See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/209729264

Textbook of Computer applications and biostatistics

Chapter · January 2011

CITATIONS READS
0 61,448

4 authors, including:

Remeth J Dias Kailas K Mali

Government Polytechnic, Jalgaon Adarsh College of Pharmacy, Vita
85 PUBLICATIONS   328 CITATIONS    73 PUBLICATIONS   307 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Liquisolid Compacts View project

Pharmaceutical Chemistry View project

All content following this page was uploaded by Kailas K Mali on 06 October 2014.

The user has requested enhancement of the downloaded file.

 TEXTBOOK OF
COMPUTER APPLICATIONS
AND BIOSTATISTICS

Dr. S. B. Bhise
M. Pharm PhD
Principal,
Singhad Institute of Pharmaceutical Sciences, Lonavala
[email protected]

Dr. R. J. Dias
M. Pharm PhD MBA
Professor,
Singhad Institute of Pharmaceutical Sciences, Lonavala
[email protected]

K. K. Mali
M. Pharm (Biopharm)
Associate Professor,
Satara College of Pharmacy, Satara
[email protected]

P. H. Ghanwat
DIE, MCA
Visiting Lecturer,
Satara College of Pharmacy, Satara
[email protected]

INNOVATE

P
TRINITY PUBLISHING HOUSE
PUBLISH
Serving Pharmacy Profession
Textbook of Computer Applications and Biostatistics

Published by

Mrs. Anita R. Dias

For Trinity Publishing House,
475/8, F-3, Suryanandan Apartments,
Near Hotel Suruban, Sadar Bazaar,
Satara - 415 001. India.
Mobile: +91 9850953955, +91 8087250235
E-mail: [email protected]

© 2011 Trinity Publishing House

All rights reserved. No part and style of this book be reproduced or transmitted, in any form, or by any
means- electronic, mechanical, photocopying, recording or otherwise, without prior permission of the
publishers and authors.

Disclaimer: As new information becomes available, changes become necessary. The

editors/authors/contributors and the publishers have, as far as it is possible, taken care to ensure that
the information given in this book is accurate and up-to-date. In view of the possibility of human error or
advances in medical science neither the editor nor the publisher nor any other party who has been
involved in the preparation or publication of this work warrants that the information contained herein is in
every respect accurate or complete. Readers are strongly advised to confirm. This book is for sale in
India only and cannot be exported without the permission of the publisher in writing. Any disputes and
legal matters to be settled under Mumbai jurisdiction only.

ISBN 978-81-920565-1-7

Rs. 350/-

Printed at
Vikram Printers Pvt. Ltd.
31 & 34, Parvati Industrial Estate,
Pune-Satara Road,
Pune- 411 009. India.
Phone: (020)24220890, 24228905. Distributed by
www.vikram-printers.com Amit Book Company Pvt. Ltd.
B-3/16, Darja Ganj,
Designed by Near The Time of India,
Srushti Computers, Satara. New Delhi - 110 002.
G-2, ‘Venna’, Adarshnagar, Khed, Satara. Phone: 011 - 43538989.
[email protected]. [email protected].
INNOVATE

PUBLISH
TRINITY PUBLISHING HOUSE
 PREFACE

We are very pleased to put forth the first edition of book, ‘Textbook of Computer
Applications and Biostatistics’. This book is intended to be an introduction to pharmacy
students regarding applications of computers and biostatistics to pharmacy. The basic
knowledge of computers and their applications is covered in details as it is essential to
students in every walk of their lives. The procedures for operating MS-Office 2003 is
discussed here as many colleges still use this version. However our second edition will have
the procedures for operating latest versions of Windows. We regret for inconvenience caused
to few readers, due to this.

The concepts of biostatistics are discussed here with minimum of maths, so as to drive away
the maths phobia in pharmacy students. Moreover, most of the statistics can be handled
through computers using excel and we have emphasized in every chapter on how to use
computers for statistical needs. This will help students to handle the data and infer about their
experiments easily.

This book is an sincere effort to bring statistical concepts in simple, understandable form so
that every student will enjoy to learn them with ease. The learning objectives, summary,
multiple choice questions and exercise in all twenty two chapters makes the book more
interesting.

We acknowledge the help and co-operation extended by various persons in bringing out this
book. We are highly indebted to the authors of the various books and articles mentioned in
bibliography which became a major source of information for writing this book. We also
thank the publishers and designers who graciously worked hard to publish this book in time.

Our request to all users of this book is to provide constructive criticism in improving further
editions of the book. We sincerely hope that readers will certainly welcome the book.

SB Bhise
Satara RJ Dias
KK Mali
January 15, 2011. PH Ghanwat

(iii)
 CONTENTS

Chapter Title of the Chapter Page

No. No.
1. Introduction to Computers 1
Introduction, History of computers, Evolution and generations of computers,
Characteristics of computer, Types of computers, Applications of computers.
2. Anatomy and Computer Peripherals 13
Anatomy of computer system, Parts of computer system, Hardware, Software,
Input devices, Output devices, Memory, Binary numbers in computers, Unit of
size, Computer language.
3. Operating System 34
Introduction, Functions, Types of operating systems, MS-DOS, MS-Windows,
Working in Paint, Wallpaper, Screensaver.
4. Microsoft Word 60
Introduction, Features of word processing, Opening Microsoft Word,
Components of the screen in MS-Word, Creating, editing and saving the
document, Formatting the text, Printing a document, Quitting Microsoft Word.
5. Microsoft Excel 75
Introduction, Features of Microsoft Excel, Opening of Spreadsheet,
Components of an Excel workbook, Entering data and saving a new workbook,
Mathematical calculations, Moving and copying data, Deleting and adding rows
and columns, Aligning data, Changing the size of row and column,
Creating a graph, Adding, renaming or deleting a sheet from the workbook,
Closing the workbook, Quitting Microsoft Excel.
6. Microsoft PowerPoint 95
Introduction, Features of PowerPoint presentation, Starting PowerPoint
presentation, Components of MS-PowerPoint, Exploring PowerPoint views,
Creating and saving a PowerPoint presentation, Adding slides, chart, picture,
text box to a presentation, Duplicating and deleting slides, Adding animation
to a presentation, Making slide show, Closing a PowerPoint presentation,
Quitting a PowerPoint presentation.
7. Computer Networking and Internet Applications 110
Introduction, Types of networking, Topology, TCP/IP protocol, Advantages
of Networking, Internet, Internet connection, Requirements for connecting to
the Internet, Internet services, WWW, DNS, URL’s, E-mail, Intranet, Net
surfing, Chatting, Computer virus.
8. Applications of Computers to Pharmacy 141
Use of computers in Manufacturing of drugs, Quality control, Quality
assurance, Pharmaceutical analysis, Inventory control, Clinical research,
Retail pharmacy, Drug information services, Marketing and sales, Hospital
pharmacy, Clinical services, Bioequivalence testing, Basic research, etc.
9. Introduction to Biostatistics 148
Definition, Types of statistics, Applications and uses of Biostatistics, Types
of variables, Identification of the type of variable.

(v)
 CONTENTS

Chapter Title of the Chapter Page

No. No.
10. Presentation of Data 157
Tabulation of data, Graphical presentation of categorical and metric data,
Charting of data using MS-Excel.
11. Shape of Distribution of Data 187
Shapes of data, Bell shaped distribution, Skewness, Kurtosis.
12. Measures of Central Tendency 193
Measures of location: Mode, Median, Mean, Measures of spread: Percentile,
Range, Interquartile range, Standard deviation, Use of Excel in measures of
central tendency.
13. Probability and Probability Distribution 231
Classic probability, Probability of simple and composite event, Probability
involving two variables and conditional probability, Probability distribution.
14. Sampling Techniques 244
Methods of Sampling, Precision, Accuracy and Bias.
15. Estimation of Confidence Interval 251
Concept of confidence interval, Standard error of the mean, Estimation of
confidence interval.
16. Hypothesis Testing 259
Hypothesis testing, Decision rule, Types of errors, One-tailed and two-tailed
test, Defining the critical region for statistical test.
17. Choice of Statistical Tests 266
Parametric and Non-parametric tests, Choice of statistical tests, Commonly
used hypothesis tests.
18. Hypothesis Testing for One Sample 271
One sample z-test, One sample t-test.
19. Hypothesis Testing for Two Samples 282
Two samples z-test, Two samples t-test, Paired t-test, Use of MS-Excel in
hypothesis testing for two samples
20. One Way Analysis of Variance (ANOVA) 305
Definition, Calculations of ANOVA by using definitional and computational
formula, One way ANOVA using MS-Excel.
21. Correlation 327
Definition, Types of correlation, Calculation of correlation coefficient by
definitional and computational formula, Correlation using MS-Excel.
22. Linear Regression 340
Definition, Meaning of regression, Regression coefficient, Linear regression
using MS-Excel.
Important Points & Formulaes At A Glance 351
Appendices 354
Bibliography 358

(vi)
 Chapter 8
APPLICATIONS OF COMPUTERS TO PHARMACY

Learning objectives
When we have finished this chapter, we should be able to:
1. Understand applications of computers to pharmacy.
2. Know various computer programmes used in different areas of pharmacy field.

Introduction
The utility of computer in collection, evaluation, organisation and dissemination of
information has made their presence virtually in every walk of life. Their potential in every field of
pharmacy has led to its extensive use encompassing research of drug, its manufacturing and till its
proper usage. The following properties of computers have made them to bring biggest revolution of
the twenty first century known as information technology. The properties are:
1. Large storage capacity
2. Speed and accuracy
3. Flexibility
4. Ease of dissemination and transmission
5. Multiple user capacity
6. Can do repetitive tasks.

Let us see the applications of computers to various fields of pharmacy as given below:
1. Use of computers for manufacturing of drugs
The manufacturing of drugs in various dosage forms requires sophisticated instruments and
machinery which are now-a-day controlled by computers. The touch screen panels provided to these
machines can be utilised for controlling various manufacturing variables thereby producing quality
medicines. Automation brought in manufacturing area has increased the efficiency, quality and safety
manifold.
Computer- aided manufacturing (CAM) is the use of computers to plan and control
manufacturing process. A well designed CAM system allows manufacturers to become much more
productive. Not only a greater number of products are produced, but also speed and quality is
increased.
Softwares available: Marg pharmaceutical software for manufacturers, DMC Medical
Manufacturing, Taylor Pharmaceutical Manufacturing, TGI Process manufacturing, MISys
Manufacturing software, etc.

141
 142 Textbook of Computer Applications and Biostatistics

2. Use of computers in quality control

The quality of medicine is utmost important for any manufacturing plant and every batch has
to pass stringent quality control norms. Computers play very important role in quality control
department by analyzing and interpreting whether raw materials and finished products match the
expected quality norms, as per the specifications given in official book.
Softwares available: Darwin LIMS software for QA/QC, DMC quality control software,
Monark software for Pharma QC system, MasterControl QC software etc.
3. Use of computers in quality assurance
The computers are used in documentation of every process for assuring quality medicine.
The preparation of standard operating procedures, validating their use, and ensuring their adherence
according to guidelines are few areas where computers are necessary.
Softwares available: Qtor Pharmaceutical QA software, EtQ software, Marg QA, DMC QA
software, etc
4. Use of computers in pharmaceutical analysis
Various analytical methods like high performance liquid chromatography (HPLC), Gas
chromatography (GC), Gas chromatography with mass spectrometry (GC-MS), Infrared
spectroscopy (IR), UV visible spectroscopy (UV) etc are used for analysing various drugs.
Computers are useful in all these techniques for identifying and interpreting the compounds. For
example, Gas liquid chromatography separates the individual components and MS identifies it. This
operation generates hundreds of spectra in few minutes containing number of peaks. Computer when
used for interpretation of this data, stores these spectra for some time and then represents it in the
graphical form. For identification of mass spectrometry computer compares the spectrum of the given
sample with the spectrum of pure compound.
Softwares available: DeWinter, Drug Testing Software Management Suit, DrugPak IPA
CoreAnalysis,Assistant Pro 4.1, etc.
5. Use of computers in inventory control
The inventory control includes purchasing of raw materials as per the demand, supply of
finished goods, distribution of raw materials to various departments, and maintaining the stocks
effectively. Softwares can be utilized to provide reports on stock status, opening and closing balance
and to prepare purchase orders.
Softwares available: mSupply, IMS Leon, Meditab IMS, CASI, DMC, etc.
6. Use of computers in clinical research
Computer is utilized for coding of data and statistical analysis of results of clinical trials.
Computer process ensures uniformity, completeness and accuracy of data collected from various
centres and helps in evaluating efficacy of design protocols on investigational drugs. Computers has
been extensively used in multicentric clinical trials on investigational drugs.
Softwares available: OpenClinica, TEMPO, Cytel, InferMed, Clinplus, TrackWise,
Metadata ClinAxys II, etc.
 Applications of Computers to Pharmacy 143

7. Use of computers in managing drug store (Retail pharmacy)

In retail pharmacy store, inventory control is done using computers. Computers also helps in
sales analysis and purchasing of the medicines. Computers can be used for storage of patient records
and the drug history profiles. The software are also available which shows drug interactions in
prescribed medication and patient instructions to be given. This allows pharmacists to do better
patient counseling job.
Software available: Dava plus, PharmaSoft, ApotheSoft Rx, PEPID, Essel, Abascus
Pharmacy, MediVision, etc.
8. Use of computers in drug information services
The vast data is available on drugs and disease state and computerization of this data is
needed to retrieve the information needed. The user oriented drug information systems are available
whereby drug information guide for patients and full text database of drug is given as ready reckoner.
Drug information centres using computers can answer drug related queries like how, why, what and
when of medication immediately.
Softwares available: MicroMedeX
Websites: RxFacts .org, medind.nic.in, www.uiowa.edu, www.health.umt.edu, Lexicomp
online, Drugs.com, Davi’s Drug Guide online, etc.
9. Use of computers in marketing and sales
The use of computers in every arena of marketing and sales department have made the field
more promising and lucrative. In this field computers are used in market research, consumer survey,
advertizing, sales promotion campaigns, product management, developing dealer distribution
network, sales analysis etc. E marketing is newer concept in which medicines are purchased or sold
online eg. Dava Bazaar.
Softwares available: Medisno Pharma CRM, Metastorm, Marg online MR software, Marg
ethical marketing software, etc.
10. Use of computers in hospital pharmacy
Use of computer in managing bigger hospitals have saved a lot of money and man hours. A
complete hospital information system is a multiterminal database management system that provides
facilities for integrated handling of information regarding registration of patients; investigation,
treatment, follow up etc.
Some of the other areas covered under this system are inventory control and purchase of
medicine, drug information database, adverse effects and drug interaction management, drug
distribution within hospital, prescription monitoring and preparation of hospital formulary.
Softwares available: Meditab IMS, WinPharm, WorkPath, MediNovs, HMS-Leon, etc.
11. Use of computers in clinical services
Computer programmes have been designed to assist in the monitoring of patients with
chronic diseases like diabetes, hypertension and asthma. Therapeutic drug monitoring based upon
plasma concentration of a drug is possible using computer programmes.
Softwares available: NAMAH, Pharmacy Plus, VirtualCare, MEDIPHOR, etc.
 144 Textbook of Computer Applications and Biostatistics

12. Use of computers in bioequivalence testing centres

Bioequivalence of generic drugs with reference innovator’s drug is very important
regulatory need for marketing off patented drugs. Various bioequivalence centres use computer
programes to accumulate, sort and use the data for bioequivalence testing.
Softwares available: WinNonlin, DDSolver, EquivTest, MONOLIX, Study Size 2.0, etc.
13. Use of computers in basic research
The softwares available today are playing important role in drug design. Computer aided
drug design (CADD), Qualitative Structure Activity Relationship (QSAR), Molecular modeling are
promising areas that help the researchers in doing their research with ease.
Available softwares: AMBER, CHARMM and GROMACS are widely used to carry out
molecular modeling. FTDock and DARWIN has tremendous application in the rational drug design
process.
Other softwares available: Prochemist, Tripos, Oxford Molecular QSAR, HYPERCHEM,
MATLAB, DRAGON, RECKON, Spartan, etc.
14. Use of computers in Medicine
Many of the modern-day medical equipment have small, programmed computers. Many of
the medical appliances of today work on pre-programmed instructions. The circuitry and logic in
most of the medical equipment is basically a computer.
Computer software is used for diagnosis of diseases. It can be used for the examination of
internal organs of the body.Advanced computer-based systems are used to examine delicate organs of
the body. Some of the complex surgeries can be performed with the aid of computers. The different
types of monitoring equipment in hospitals are often based on computer programming.
Medical imaging is a vast field that deals with the techniques to create images of the human
body for medical purposes. Many of the modern methods of scanning and imaging are largely based
on the computer technology. It has been possible to implement many of the advanced medical
imaging techniques, due to the developments in computer science. Magnetic resonance imaging
employs computer software. Computed tomography makes use of digital geometry processing
techniques to obtain 3-D images. Sophisticated computers and infrared cameras are used for
obtaining high-resolution images. Computers are widely used for the generation of 3-D images in
medicine.
Softwares available: CliniScript, Clinichem, LIFEDATA EMR, LDRA, EasyDiagnosis,
Diagnosis Pro, COMPUTER CLINIC, DICOM, 3D-DOCTOR, MIM, etc.
15. Use of computers in Biostatistics
Every data generated during the research is to be handled or manipulated carefully for
drawing inference from it. Both the types, descriptive and inferential statistics can be applied by
researcher using MS Excel or relevant softwares.
Softwares available: SigmaStat, GraphPad Prism, Minitab, SPSS, SAS, DesignExpert,
SigmaXL, etc.
 Applications of Computers to Pharmacy 145

16. Use of computers in Patent searching

Now a days computers are widely used for searching patents available online on various
countries official websites. The important sites available for the same are given in table below:
Important websites for patent search
Sr. No. Title Official websites
1. US Patent and Trademark Office www.uspto.gov
2. UK Patent Office (IPO) www.ipo.gov.uk
3. European Patent Office www.epo.org
4. Japanese Patent Office www.jpo.go.jp
5. Word Intellectual Property Organisation www.wipo.int
6. Indian Patent Office www.ipindia.nic.in
7. Australian Patent Office www.ipaustralia.gov
8. Singaporean Patent Office www.ipos.gov.sg
9. Chinese Patent Office www.chinatrademarkoffice.com
10. Google Patent Search www.google.com/patents
11. free online Patent search www.freepatentsonline.com
12. Espace Patent database http//wordwide.espacenet.com

17. Use of computers in pharmacology simulations

Now a days computers are widely used as alternative for animal experimentation. Various
simulated pharmacology experiments are generated with the help of computers and are widely used
for the learning of undergraduate students.
Softwares available: Biosoft, Neurosim, LabTutor, X-Cology, Cardiolab, MacDogLab,
Ex-pharm, PCCAL, etc.
18. Use of computers in pharmacokinetics simulations
It is a simulation method used in determining the safety levels of a drug during its
development. It gives an insight to drug efficacy and safety before exposure of individuals to the new
drug that might help to improve the design of a clinical trial.
Simcyp Simulator and GastroPlus (from Simulations Plus) are simulators that take account
for individual variabilities. GastroPlus is an advanced software program that simulates the
absorption, pharmacokinetics, and pharmacodynamics for drugs in human and preclinical species.
Softwares available: NONLIN, KINPAK, ESTRIP, STRIPACT, PK solutions, KINETICS,
SimBiology, etc.
19. High performance computing bioinformatics
The advent of high throughput technologies like genome sequencing, microarrays and
proteomics has transformed biology into a data rich information sciences. The huge data generated
needs to be organized in a structured manner to facilitate the use of data mining tools for extracting
knowledge. The ultimate objective of these efforts is to improve our understanding of human health
and thereby provide rationale solutions to overcome disease. The use of computers in this area has
made performance computing possible.
Softwares available: Amber, Alchemy, Sybyl, MOE, Cerius, Bioconductor, ISYS v.1.35,
 146 Textbook of Computer Applications and Biostatistics

MetaCore, etc.
20. Literature storage and retrieval system
Computers have been utilized to offer bibliographic, indexing and abstracting services. the
articles can be referred through keywords, titles, authors or journals. Automated on-line literature
retrieval systems like Medline and Chemline are offered by National Library of Medicine, USA.
International Pharmaceutical Abstracts (IPA) and Martindale’s Extra Pharmacopoeia are also
available on compact disks.
Databases available: Excerpta Medica, LIMS,AMA/NET Information base, etc.

Summary
Use of computer in various fields of Pharmacy is given in following table:
Sr. No. Use of Computers in Pharmacy SoftwaresAvailable
1. Manufacturing of drugs Marg, DMC, Taylor, TGI, MIsys
2. Quality control Darwin LIMS, DMC-QC, Maonark, MasterControl
3. Quality assurance Qtor, EtQ, Darwin LIMS, Marg, DMC-QA
4. Pharmaceutical analysis DeWinter, Drugpak, IPACore,Assistant Pro, DTSMS
5. Inventory control mSupply, IMS Leon, Meditab IMS, CASI, DMC
6. Clinical research OpenClinica, Clinplus, Cytel, Metadata, TrackWise
7. Retail drug store and wholesalers PharmaSoft,Apothesoft, PEPID, Medivision,Abacus
8. Drug information services MicroMedex, Lexicomp Platinum, Davi’s Drug Guide
Tarascon’s Pharmacopoeia, DIT Drug Risk Navigator
9. Marketing and sales Marg Ethical Marketing, Pharma CRM, Metastorm
10. Hospital management and pharmacy WinPharm, WorkPath, MediNous, HMS-Leon
11. Clinical services PharmacyPlus, VirtualCare, TEICTDM, NAMAH
MEDIPHOR, MW/Pharm
12. Bioequivalence testing centres WinNonlin, DDSolver, EquivTest, MONOLIX
13. Research and development Prochemist, Tripos,AMBER, DRAGON, RECKON
Spartan, CHARMM, GROMACS, MATLAB
14. Biostatistics SAS, SPSS, Minitab, SigmaStat
15. Medical Diagnosis and Imaging DICOM,3D-DOCTOR, Easy Diagnosis, MIM, LDRA
16. Patent search www.uspto.gov, www.wipo.int, www.epo.org,
www.ipo.gov.uk, www.ipindia.nic.in, etc
17. Pharmacology simulations LabTutor, X-Cology, Ex-Pharm, Biosoft, Neurosim
18. Pharmacokinetic simulation NONLIN, KINETICS, KINPAK, ESTRIP
19. Bioinformatics MetaCore, BioSpice, ISYS, 3Dslicer, Bioconductor
20. Literature survey Pubmed, Medline, Google, IPA, LIMS
 Applications of Computers to Pharmacy 147

Multiple Choice Questions:

1. _________ is used to plan and control manufacturing process.
a. CAM b. CADD c. QSAR d. NONLIN
2. ________ includes purchasing, supply, distribution and maintenance of stocks using computers.
a. Clinical research b. Inventory control
c. Quality control d. Pharmaceutical analysis
3. The following software is used in the field of Clinical Research.
a. PharmaSoft b. MicroMedex c. Prochemist d. OpenClinica
4. Automated on-line literature retrival system offered by National Library of Medicine, USA
is_____.
a. MEDIPHOR b. Medline c. IPA d. Minitab
5. Bioequivalence testing centre uses following software.
a. VirtualCare b. X-Cology c. KINETICS d. Equivtest
6. The following software is used for statistical analysis.
a. LabTutor b. MONOLIX c. SAS d. DRAGON
7. Which of the following website helps online patent search.
a. www.medind.nic.in b. www.health.umt.edu
c. www.wipo.int d. www.Rxfacts.org
8. _________ is programme that simulates pharmacokinetics of drug.
a. Expharm b. GastroPlus c. DavaPlus d. ClinPlus
9. The example of e-marketing, in which medicines are purchased or sold online is _______.
a. DavaPlus b. GastroPlus c. ClinPlus d. Dava Bazaar
10. The laboratory pharmacology simulations are given by following software.
a. LabTutor b. PharmacyPlus c. Workpath d. DD Solver

Exercise:
1. Give an account on application of computers to pharmacy.
2. Discuss the use of computers in basic research.
3. Give various softwares available for pharmaceutical manufacturing.
4. How literature storage and retrieval is possible with computers?
5. Give the use of computers in pharmacokinetics.

Answers:
Multiple Choice Questions
1. a 2. b 3. d 4. b 5. d 6. c 7. c 8. b 9. d 10. a
 Chapter 9
INTRODUCTION TO BIOSTATISTICS

Learning objectives
When we have finished this chapter, we should be able to:
1. Explain types of statistics with their components.
2. Explain the difference between nominal, ordinal, metric discrete and metric continuous variables.
3. Identify the type of a variable.

What is Biostatistics?
As defined by Daniel in 1978, “Biostatistics is a field of study concerned with the
organisation and summarisation of data from health sciences and drawing of inferences about a body
of data when only a part of the data are observed”.
In simpler terms, biostatistics can be defined as the branch of statistics applied to biological
sciences whereby collection, classification, summarizing, analysis and interpretation of data is done.

Types of Statistics
All statistical procedures can be divided into general categories- descriptive or inferential.
Descriptive statistics
As the name implies, descriptive statistics describes data that we collect or observe
(empirical data). They represent all of the procedures that can be used to organise, summarise,
display, and categorise data collected for a certain experiment or event. It includes tabulation,
graphical presentation, measures of central tendency, etc.
Inferential statistics
Inferential statistics represents a wide range of procedures (tests) that are used to infer or
make predictions about a large body of information based on sample observations. The inferential
statistics include z test, t test, analysis of variance, etc.
Statistical samples and population
Statistical data usually involve a relatively small portion of an entire population, and
decisions and interpretations (inferences) are made about that population through numerical
manipulation. The population may be defined as the entire number of observations that constitute a
particular group. Samples are generally a relatively small group of observations that have been taken
from a defined population. Parameters are characteristics of populations while statistic are
characteristics of samples representing summary measures computed on observed sample values.
Parameters or population values are usually represented by Greek symbols (e.g. m, s, y) and sample
statistic are denoted by letters (e.g. `X, S , r).
2

148
 Introduction to Biostatistics 149

Table 9.1 Examples of populations and samples

Examples (task) Population parameter Sample statistic

Characterisation of the weights of All tablets that constitute 100 tablets removed for
tablets in a particular batch the batch weighing

Measurement of the incidence of heart All inhabitants of 300 patients attending

disease in Maharashtra in patients Maharashtra over the GP clinics at specified
over 45 years of age age of 45 geographical locations
throughout Maharashtra

Evaluation of the incidence of asthma All employees of the 50 named workers at

in a certain chemical company company the company
employing 500 workers

Samples, as given in above examples, are only a small subset of a much larger population and
are used for nearly all statistical tests. By using various formulas, these descriptive sample results are
manipulated to make predictions (inferences) about the population from which they are sampled.
Population Sampling Descriptive
Parameter Sample Statistic
(Unknown) Statistic (Known)

Best Mathematical
Estimate Manipulation

Inferential
Statistic
Figure 9.1 Descriptive statistics and inferential statistics

Applications and Uses of Biostatistics

1. Problem Solving
Often the research is conducted on a limited scale due to scarcity of resources. Biostatistics
helps us in interpreting the data of whole population by taking a sample from that population.
2. Use in Clinical Trials
Biostatistics can be used in designing and analysing study whereby the relative potency of a
new drug with respect to a standard drug can be found out.
3. Use in Manufacturing of Pharmaceuticals
The changes in manufacturing variables, machines and manpower to improve the efficiency
of manufacturing of pharmaceuticals can be tested and confirmed by using statistical methods.
4. Use in Quality Control of Pharmaceuticals
The quality of pharmaceuticals can be controlled by performing various quality control tests
on limited samples and interpreting the results for whole batches manufactured.
 150 Textbook of Computer Applications and Biostatistics

5. Use in Bioequivalence study

The bioavailability of test drug can be compared to innovators drug and the decision of its
bioequivalence is based on passing the appropriate statistical test for the purpose.
6. Use in Research and Development of Drugs and Technology
Biostatistics can be applied to test the significance of any experiment in research and
development of drugs and technology by comparing the obtained results with the result given by
standard drug and technology.
7. Use inAnatomy and Physiology
Biostatistics is used in anatomy and physiology
i. to define what is normal or healthy in a population and to find limits of normality in variables. e.g.
weight and pulse rate.
ii. to find the difference between means and proportions of normal at two places or in different
periods. The mean height of boy in Maharashtra is less than the mean height in Punjab. Whether this
difference is due to chance or a natural variation or because of some other factors such as better
nutrition playing a part, has to be decided.
8. Use in Pharmacology
Biostatistics is used in pharmacology
i. to find the action of drug by giving it to animals or humans to see whether the changes produced are
due to the drug or by chance.
ii. to compare the action of two different drugs or two successive dosages of the same drug.
9. Use in Medicine
Biostatistics is used in medicine
i. to compare the efficiency of a particular drug, operation or line of treatment by comparing it with
control groups.
ii. to find an association between two attributes such as cancer and smoking.
iii. to identify signs and symptoms of a disease or syndrome. Cough in typhoid is found by chance and
fever is found in almost every case. The proportional incidence of one symptom or another indicates
whether it is a characteristic feature of the disease or not.
10. Use in Community Medicine and Public Health
Biostatistics is used in medicine and public health
i. to test usefulness of sera and vaccines in the field: Here the percentage of attacks or deaths among
the vaccinated subjects is compared with that among the unvaccinated ones to find whether the
difference observed is statistically significant.
ii. In epidemiological studies: The role of causative factors can be statistically tested. Deficiency of
iodine as an important cause of goitre in a community is confirmed only after comparing the
incidence of goitre cases before and after giving iodised salt.
11. Use in Health and Vital statistics
Health and Vital statistics are essential tools in demography, public health, medical practice
 Introduction to Biostatistics 151

and community services. Biostatistics as a science of figures can tell

a. the leading causes of death,
b. the important causes of sickness, severity of disease, its prevalence, etc,
12. Use in Biotechnology, Bioinformatics and Computational Biotechnology
It can be used in analysis of genomics data, for example from microarray or proteomics
experiment, offten concerning diseases or disease stages. Statistical methods are beginning to be
integrated into Medical informatics, public health informatics, bioinformatics and computational
biology.

Types of Variables
A variable is something whose value can vary. For example age, sex, and blood type are
variables. Data are the values we get when we measure or observe a variable. There are two major
types of variables- categorical variables and metric variables. Each of these can be further divided
into two sub-types as shown in table 9.2.
Table 9.2 Types of Variables

Type of variables Sub type Characteristics Unit

Nominal Values in arbitrary categories No units
Categorical variables
Ordinal Values in ordered categories No units
Discrete Integer values on numeric scale Counted units
Metric variables
Continuous Continuous values on numeric scale Measured units

Categorical variables
1. Nominal categorical variables
Consider the variable blood type, O, A, B and A/B. The variable ‘blood type ‘ is a nominal
categorical variable. A typical characteristics of this variable are that they do not have any units of
measurement, and the ordering of the categories is completely arbitrary. In other words, the
categories cannot be ordered in any meaningful way. Therefore, we can easily write the blood type
categories asA/B, B, O,Aor B, O,A,A/B or B,A,A/B, O, or whatever.
2. Ordinal categorical variables
The Child Pugh Score is an ordinal categorical variable. This data too do not have any units
of measurement as like that of nominal variables but the ordering of the categories is not arbitrary as
it was with nominal variables. It is now possible to order the categories in a meaningful way.
Ordinal data are not real numbers. They cannot be placed on the number line. The reason is
that the Child Pugh Score data, and the data of most other clinical scales, are not properly measured
but assessed in some way, by the clinician working with the patient. This is a characteristic of all
ordinal data.
As ordinal data are not real numbers, it is not appropriate to apply any of the rules of basic
arithmetic to sort this data. We can not add, subtract, multiply or divide ordinal values. This limitation
has marked implications for the analyses of such data.
 152 Textbook of Computer Applications and Biostatistics

Metric Variables
1. Continuous metric variables
The variable ‘weight’ is a metric continuous variable. With metric variables, proper
measurement is possible and therefore these variables produce data that are real numbers, and can be
placed on the number line. Some common examples of metric continuous variables include: Birth
weight (g), blood pressure (mmHg), blood cholesterol (mg/ml), waiting time (minutes), body mass
index (kg/m2), peak expiry flow (1 per min), and so on. These variables have units of measurement
attached to them.
In contrast to ordinal values, the difference between any pair of adjacent values of
continuous metric variables is exactly the same. The difference between birth weights of 3000 g and
3001 g is the same as the difference between 3001 g and 3002 g, and so on.
Metric continuous variables can be properly measured and have units of measurement.
2. Discrete metric variables
Continuous metric data usually comes from measuring while discrete metric data, usually
comes from counting. For example, number of deaths, number of pressure sores, number of angina
attacks, and so on, are all discrete metric variables. The data produced are real numbers, and are
invariably integer (i.e. whole number). They can be placed on the number line, and have the same
interval and ratio properties as continuous metric data. Metric discrete variables can be properly
counted and have units of measurement- ‘numbers of things’.

An aid to identify type of variable

The easiest way to identity whether data is metric or categorical, is to check whether it has
units attached to it, such as: g, mm, oC, μg/cm3, number of pressure sores, number of deaths, and so on.
If not, it may be ordinal or nominal and if the values can be put in any meaningful order then it is
ordinal. Figure 9.2 is an aid to variable type recognition.

Importance of Identifying variable

In order to select the correct inferential test procedure, it is essential that as researchers, we
should understand the variables involved with our data. The type and subtype of variable is very
much important for selecting appropriate statistical test. The details of selection of statistical tests are
given in chapter 17.
 Introduction to Biostatistics 153

Has the variable got units?

No Yes

Can the data be put in Do the data come from

meaningful order? measuring or counting?
Arbitrary Ordered Counting Measuring

Categorical Categorical Metric Metric

nominal ordinal discrete continuous
Figure 9.2 An algorithm to help identify variables type

Problem: Identify the type of the following variables:

1. Dosage form - tablet/ capsule / ointment
2. Bioavailability measurements (Cmax, Tmax,AUC)
3.Age ( in years).
4. Hypertension -Mild, Moderate and Severe
5. Smoking history (cigarattes per day)
6. Hardness
7. Dissolution test- pass or fail criteria
8. Tablet weight
9. Male vs female subjects
10. Scores for patient responses to treatment
Solution:
Sr. No. Variable Type of Variable Characteristics
1. Dosage form - tablet/ capsule / ointment Categorical, Nominal No unit, No order
2. Bioavailability measurements Metric, continuous Unit present,
(Cmax, Tmax, AUC) can be measured
3. Age ( in years) Metric, continuous Unit present,
can be measured
4. Hypertension -Mild, Moderate and Severe Categorical, Ordinal No unit, ordered form
5. Smoking history (cigarattes per day) Metric, discrete Unit- numbers of things
counting can be done
6. Hardness Metric, continuous Unit present,
can be measured
7. Dissolution test- pass or fail criteria Categorical, Nominal No unit, arbitrary
8. Tablet weight Metric, continuous Unit present,
can be measured
9. Male vs female subjects Categorical, Nominal No unit, arbitrary
10. Scores for patient responses to treatment Categorical, Ordinal No unit, ordered form
 154 Textbook of Computer Applications and Biostatistics

Summary
Biostatistics
It can be defined as the branch of statistics applied to biological sciences whereby collection,
classification, summarising, analysis and interpretation of data is done.
Types of Statistics
1. Descriptive Statistics: It describes collected data.
2. Inferential Statistics: It infers about a large body of information based on sample.
Types of Variables
1. Categorical Nominal : No any units of measurement and values in arbitrary categories.
2. Categorical Ordinal: No any units of measurement and values in ordered categories.
3. Metric Continuous: It can be properly measured and have units of measurement.
4. Metric Discrete: It can be properly counted and have units of measurement.

Multiple choice questions

1.Biostatistics is branch of statistics applied to_______ science whereby collection, classification,
summarising, analysis and interpretation of data is done.
a. pharmaceutical b. medicinal
c. biological d. chemical
2. Descriptive statistics represents all of the procedures that can be used to_______.
a. organise, summarise b. display and categorise
c. a &b d. none of above
3. The population is_________.
a. the entire number of observations that constitutes a particular group.
b. the entire number of samples that constitutes a particular group.
c. a & b d. None of above
4. Categorical variable can be divided into__________.
a. nominal & continuous b. nominal & discrete
c. discrete & continuous d. nominal & ordinal
5. Metric data can be divided into
a. nominal & continuous b. nominal & discrete
c. discrete & continuous d. nominal & ordinal
6.The goal of ___________ is to focus on summarizing and explaining a specific set of data.
a. inferential statistics b. descriptive statistics
c. none of the above d. all of the above
7. Metric variable can be properly_________.
a. measured b. counted
c. a & b d. none of the above
 Introduction to Biostatistics 155

8. In categorical variables, values are in ________ categories.

a. arbitrary & counted b. arbitrary & measured
c. arbitrary & ordered d. counted & measured
9. Bioavailability measurement is a ___________ variable.
a. metric continuous b. metric discrete
c. categorical continuous d. categorical ordinal
10. Scores for patient responses to treatment is a ____________ variable.
a. metric continuous b. metric discrete
c. categorical continuous d. categorical ordinal

Exercise
1. Define biostatistics and enumerate applications of it.
2. Give various types of variables with their characteristics.
3. Give the importance of identifying type of variable in biostatistics.
4. Identify the type of variables associated with clinical trials of a drug given below,
a. Sex
b.Age
c. Height
d. Weight
e. Blood type (A, B,AB, O)
f. Blood pressure (Mild, Moderate, Severe)
g. Blood glucose level
h. Fed vs fasted state
i. Manufacturer (generic vs brand)
j. Smoking history (no of cigarettes per day)
5. Identify the types of variables given below, associated with manufacturing a batch of tablets of
Ciprofloxacin.
a. Impurities- present or absent
b.Amount of active ingredient (content uniformity)
c. Disintegration time
d. Dissolution test- pass or fail criteria
e. Friability- pass or fail criteria
f. Hardness
g.Appearance (good, better, best)
h. Machine efficiency score (-5 to +5)
i. Weight variation test (pass of fail)
j. Human resources employed (No of persons)
 156 Textbook of Computer Applications and Biostatistics

6. Give various types of biostatistics.

7. Distinguish between parameter and statistic.
8. Distinguish between categorical and metric variables.
9. How will you identify the type of given variable?
10. What do you mean by sample and population? Explain.

Answers:
Multiple Choice Questions
1. c 2. c 3. a 4. d 5. c 6. b 7. c 8. c 9. a 10. d
Exercise
4. a. Sex- categorical nominal
b.Age- metric continuous
c. Height- metric continuous
d. Weight- metric continuous
e. Blood type (A, B,AB, O)- categorical nominal
f. Blood pressure (Mild, Moderate, Severe)- categorical ordinal
g. Blood glucose level- metric continuous
h. Fed vs fasted state- categorical nominal
i. Manufacturer (generic vs brand)- categorical nominal
j. Smoking history (no of cigarettes per day)- metric discrete

5. a. Impurities- present or absent- categorical nominal

b.Amount of active ingredient (content uniformity)- metric continuous
c. Disintegration time- metric continuous
d. Dissolution test- pass or fail criteria- categorical nominal
e. Friability- pass or fail criteria- categorical nominal
f. Hardness- metric continuous
g.Appearance (good, better, best)- categorical ordinal
h. Machine efficiency score (-5 to +5)- categorical ordinal
i. Weight variation test (pass of fail)- categorical nominal
j. Human resources employed (No of persons)- metric discrete
 Chapter 10
PRESENTATION OF DATA

Learning objectives
When we have finished this chapter, we should be able to:
1.Construct the tables of frequency, relative frequency,cumulative frequency and relative cumulative
frequency.
2. Construct grouped frequency table and a cross-tabulation table.
3. Choose the most appropriate graph for the given data type.
4. Draw pie charts, bar charts, histograms, frequency polygons and ogives.
5. Interpret and explain what a table or graph reveals.

Introduction
Whenever the data is collected for some project, it is usually in the ‘raw’ form and not in a
organised way. Descriptive statistics deals with sorting this raw data by putting it into a table or by
presenting it in an appropriate chart or summarising it numerically.
An important consideration in sorting the raw data is the type of variable concerned. The data
from some variables are best described with a table, some with a chart, and some with both. However,
a numeric summary is more appropriate for some types of variable.

Tabulation of Data
Tabulation is the first step before the data is used for analysis or interpretation. Frequency
distribution tables presents data in a relatively compact form, ready to use but certain information
may be lost. The data can be reduced to manageable form using frequency tables.
The frequency table
The frequency table can have one or all the following parameters, depending on the type of
data.
1. Frequency:
Frequency is the repetition of observations or actual number of subjects in each category.
2. Relative frequency:
Relative frequency is the frequency converted into percentage of the total number of
observations.
Number of observations in category
Relative frequency = ´100
Total number of observations ...1
3. Cumulative frequency:
It is the cumulative total of frequencies and is obtained by adding the frequency of

157
 158 Textbook of Computer Applications and Biostatistics

observations at each level point to those frequencies of the preceding level (s).
4. Cumulative relative frequency:
It is cumulative frequency converted into the percentage of the total number of observations.
Let us take the examples of various types of data and construct the frequency table.

1. Frequency table for nominal data

Example 10.1
In blood group detection camp, 95 pharmacy students were sampled to have following blood groups.
Blood groups of 95 pharmacy students were as follows:
B, AB, A, A, B, A, AB, A, B, A, A, AB, B, A, A, B, A, AB, A, B, A, A, B, A, AB, A,
B, A, B, B, A,AB, A, B, B, A, O,AB, B, A, A, AB, O, B, A, A, B, O, B, A, B, A, A, B, A, A, B,
A, B, A, A, AB, A, A, AB, B, A, A, AB, B, A, A, B, A, A, B, A, B, A , B, A, A, AB, A, AB, A,
A, O, AB, A, AB, A, A, B, A.

Solution
As we know, the ordering of nominal categories is arbitrary, and in this example they are
shown by the number of students in each – largest first. The total frequency (n = 95), is shown at the
top of the frequency column. This is helpful for the reader.
1. Frequency
Table 10.1 Frequency table showing the distribution of blood group of 95 pharmacy students

Category of Frequency
Blood group Tally marks (number of students) n=95

A |||| |||| |||| |||| |||| |||| |||| |||| |||| |||| 49
B |||| |||| |||| |||| |||| || 27
AB |||| |||| |||| 15
O |||| 04

2. Relative frequency
Table 10.2 Relative frequency table showing the percentage of students in each blood group
Category of Frequency Relative Frequency
Blood group (number of students) n=95 (% of students in each category)
A 49 (49/95)*100 = 51.6
B 27 (27/95)*100 = 28.4
AB 15 (15/95)*100 = 15.8
O 04 (04/95)*100 = 04.2
 Presentation of Data 159

3. Cumulative frequency
It makes no sense to calculate cumulative frequency for nominal data, because of the
arbitrary category order. Hence, cumulative frequency is not calculated.
2. Frequency table for Ordinal Data
When the variable in question is ordinal, we can allocate the data into ordered categories.
Example 10.2
Let us take an example of ‘level of satisfaction’ of 60 final year students regarding
infrastructure available in the college. The following data is given in numbered form for easy
understanding: (4- very satisfied, 3- satisfied, 2- neutral, 1-dissatisfied, 0- very dissatisfied).
Data: 3, 0, 2, 1, 3, 4, 0, 3, 4, 0, 2, 3, 4, 1, 3, 2, 3, 4, 0, 1, 3, 4, 3, 4, 0, 3, 2, 1, 3, 4, 2, 1, 3, 3, 1, 4,
3, 1, 3, 4, 1, 3, 4, 3, 4, 0, 3, 2, 3, 4, 1, 3, 1, 0, 3, 4, 3, 2, 1, 0
Solution:
Level of satisfaction is clearly an ordinal variable. ‘Satisfaction’ cannot be properly
measured, and has no units. But the categories can be meaningfully ordered, as they have been given
here.
The frequency values indicate that more than half of the patients were happy with their infra
structural facilities, 34 students (13+21), out of 60. Much smaller numbers expressed dissatisfaction.
Table 10.3 The frequency distributions for the ordinal variable ‘level of satisfaction’ with
infrastructure available in college
Satisfaction with Tally marks Frequency
infrastructure (number of students) n=60
Very satisfied (4) |||| |||| ||| 13
Satisfied (3) |||| |||| |||| |||| | 21
Neutral (2) |||| || 07
Dissatisfied (1) |||| |||| | 11
Very dissatisfied (0) |||| ||| 08

Table 10.4 The relative frequency distributions for data

Frequency Relative Frequency

Satisfaction with (% of students in each
(number of students)
infrastructure level of satisfaction)
n=60
Very satisfied (4) 13 (13/60)*100 = 21.7
Satisfied (3) 21 (21/60)*100 = 35
Neutral (2) 07 (07/60)*100 = 11.7
Dissatisfied (1) 11 (11/60)*100 = 18.3
Very dissatisfied (0) 08 (08/60)*100 = 13.3
 160 Textbook of Computer Applications and Biostatistics

Table 10.5 The Cumulative and relative cumulative frequency distributions for data

Frequency Cumulative Frequency Relative Cumulative Frequency

Satisfaction with (Cumulative number
(number of (Cumulative % of students)
infrastructure of students)
students) n=60
Very satisfied (4) 13 13 (13/60)*100 = 21.7
Satisfied (3) 21 13+21= 34 (34/60)*100 = 57.7
Neutral (2) 07 34+07= 41 (41/60)*100 = 68.3
Dissatisfied (1) 11 41+11= 52 (52/60)*100 = 86.7
Very dissatisfied (0) 08 52+08= 60 (60/60)*100 = 100

3. Frequency Table for Metric Continuous data

Organising raw metric continuous data into a frequency table is usually impractical, because
there are such a large number of possible values. The most useful approach with metric continuous
data is to group them first, and then construct a frequency distribution of the grouped data.
The construction of frequency distributions in case of metric continuous data requires
following three steps:
1. Choosing of class intervals,
2. Tallying the data into these classes, and
3. Counting the tallies in each class (frequency)
While choosing the class intervals, the number of observations to be grouped are very
important. We seldom use fewer than 6 or more than 15 classes. The general guide to decide number
of intervals for various sample sizes (observations) are given in table below as per Sturges’rule.
Table 10.6 Number of intervals for various sample sizes using Sturges’Rule

Sample Size K Intervals

23-45 6
46-90 7
91-181 8
182-363 9
364-726 10
727-1454 11
1455-2909 12

In a grouped frequency distribution.

1) all class intervals must be of same width, or size;
2) the intervals should be mutually exclusive and exhaustive;
3) the interval widths should be assigned so the lowest interval includes the smallest
observed outcome and the top interval includes the largest specified outcome.
 Presentation of Data 161

Let’s see one example of grouping metric continuous data

Example 10.3
The following are the weights in kg of 60 final year pharmacy students.
50, 61, 70, 61, 78, 56, 71, 63, 66, 75, 77, 52, 80, 45, 56, 57, 58, 60, 62, 72, 78, 48, 50, 63, 51,
64, 67, 52, 53, 54, 55, 56, 57, 70, 71, 62, 72, 57, 73, 64, 65, 66, 67, 62, 63, 64, 65, 52, 60, 54, 56, 63, 58,
57, 61, 76, 58, 84, 46, 50,
Solution:
Here, the smallest value in the observations is 45 while the largest one is 84, the difference of
39. So, 8 groups of width 5 can be taken, to cover all observations.

Table 10.7 The frequency distribution table for metric continuous data
Weight Tally marks Frequency
( kg) n=60
45-49 ||| 03
50-54 |||| |||| 10
55-59 |||| |||| || 12
60-64 |||| |||| |||| 15
65-69 |||| | 06
70-74 |||| || 07
75-79 |||| 05
80-84 || 02

1. Class:
The group of observations is called as class. In this example there are 8 classes ( 45-49, 50-
54, 55-59, 60-64, 65-69, 70-74, 75-79, 80-84).
2. Class limits:
The minimum value that can be included in the class is lower class limit while the maximum
value that can be included in the class is upper class limit. In this example, for class 45-49, the lower
class limit is 45 while upperclass limit is 49.
3. Class boundaries:
Consider the classes 45-49, 50-54, 55-59, 60-64, 65-69, 70-74, 75-79, 80-84 in this example.
49 is the upper class limit for 45-49 class while 50 is the lower class limit for 50-54 class. Here, class
limits are not continuos and therefore we have to subtract 0.5 from lower limit and add 0.5 from upper
limit. Thus class become continuous as shown in table below and they are called as class boundaries.
In case of classes 45-50, 50-55, 55-60, 60-65, 65-70, 70-75, 75-80, 80-85 the class limits are called
continuous and hence class limits are called as class boundaries.
 162 Textbook of Computer Applications and Biostatistics

Table 10.8 Class boundaries for non continuous class limits

Class limits Class Boundaries
45-49 44.5-49.5
50-54 49.5-54.5
55-59 54.5-59.5
60-64 59.5-64.5
65-69 64.5-69.5
70-74 69.5-74.5
75-79 74.5-79.5
80-84 79.5-84.5
Class interval:
The difference between class boundaries is called class interval or class width.
Table 10.9 The relative, cumulative and relative cumulative frequency distribution table

Weight Frequency Relative Cumulative Relative Cumulative

( kg) n=60 frequency frequency frequency
45-49 03 05 03 05
50-54 10 16.7 13 21.7
55-59 12 20 25 41.7
60-64 15 25 40 66.7
65-69 06 10 46 76.7
70-74 07 11.7 53 88.3
75-79 05 8.3 58 96.7
80-84 02 3.3 60 100

Class mark:
Class marks are simply the midpoints of the classes and they are found by adding lower and
upper limits of a class (or its lowest and upper boundaries) and dividing by two.

Lower class boundary + upper class boundary

Class mark (midpoint) = ...2
2
Open Ended Groups
One problem arises when one or two values are a long way from the general mass of the data,
either much lower or much higher. These values are called outliers. Their presence can mean having
lots of empty or near-empty rows at one or both ends of the frequency table.
For example, if one student is having weight of 24 kg in above example, then the groups
(class intervals) 40-44, 35-39, 30-34 and 25-29 will have empty cells. In this case open-ended group
can be used, here <45 group can be incorporated and frequency of one can be recorded thus avoiding
empty cells.
 Presentation of Data 163

4. Frequency Table for Metric Discrete data

Constructing frequency tables for metric discrete data is generally easy as compared to
continuous metric data, because the number of possible values which the variable can take is often
limited
Example 10.4
The data given below are the number of times inhaler used in past 24 h by 53 children with
asthma. Construct frequency, relative frequency, cumulative frequency, relative cumulative
frequency table.
Data: 0, 2, 1, 3, 5, 2, 4, 6, 1, 2, 3, 1, 0, 2, 3, 1, 2, 1, 4, 5, 2, 0, 2, 3, 4, 0, 2, 1, 4, 2, 1, 2, 1,
2, 4, 1, 0, 1, 3, 1, 5, 3, 1, 2, 1, 7, 1, 3, 1, 0, 1, 3, 6.
Table 10.10 The frequency distribution table for metric discrete data

Number of times inhaler Frequency

Tally marks
used in past 24 h n=53
0 |||| | 6
1 |||| |||| |||| | 16
2 |||| |||| || 12
3 |||| ||| 8
4 |||| 5
5 ||| 3
6 || 2
7 | 1

Table 10.11 The relative, cumulative and relative cumulative frequency distribution table

Number of times inhaler Frequency Relative Cumulative Relative cumulative

used in past 24 h n=53 frequency frequency frequency
0 06 11.32 06 11.32
1 16 30.18 22 41.50
2 12 22.64 34 64.14
3 08 15.09 42 79.23
4 05 09.43 47 88.66
5 03 05.66 50 94.32
6 02 03.77 52 98.09
7 01 01.88 53 100

Cross- tabulation
Each of the frequency tables above provides us with a description of the frequency
distribution of a single variable. However, sometimes, we need to examine the association between
two variables, within a single group of individuals. We can do this by putting the data into a table of
 164 Textbook of Computer Applications and Biostatistics

cross-tabulations, where the rows represent the categories of one variable, and the columns represent
the categories of a second variable.
Example 10.5
A study was carried out on the degree of job satisfaction among doctors and nurses in rural
and urban areas. To describe the sample a cross-tabulation was constructed which included the sex
and the residence (rural or urban) of the doctors and nurses interviewed. This was useful because in
the analysis the opinions of male and female staff had to be compared separately for rural and urban
areas.
Table 10.12 Type of health worker by residence
Type of Health Worker
Residence Doctors Nurses Total

Rural 10 (16%) 69 (38%) 79 (33%)

Urban 51 (54%) 113 (62%) 164 (67%)
Total 61 (100%) 182 (100%) 243 (100%)

Table 10.12 a shows that a higher percentage of nurses than of doctors work in rural areas, but
that, overall, a greater proportion of staff works in urban areas (67%).

Graphical Presentation of Data

1. The Pie Chart
Graphical presentation of data with appropriate chart is a good idea for describing data
effectively. Appropriate chart depends primarily on the type of data, as well as on what particular
features of it we are looking for.
Graphical presentation of Nominal and Ordinal data
The pie chart is a diagram in which the frequencies of the groups are shown in a circle. Each
segment (slice) of a pie chart should be proportional to the frequency of the category it represents.

O 4%
16%
AB

A
52%
B

28%

Figure 10.2 Pie chart of percentage blood group of pharmacy students

For example, Figure 10.2 is a pie chart of blood group of each of 95 pharmacy students
 Presentation of Data 165

shown in Table 10.2. A disadvantage of a pie chart is that it can only represent one variable (in Figure
10.2, blood group). We will therefore need a separate pie chart for each variable we want to chart. A
disadvantage of pie chart can lose clarity if it is used to represent more than four or five categories.

2. The Simple Bar Chart

An alternative to the pie chart for nominal data is the bar chart. This is a chart with frequency
on the vertical axis and category on the horizontal axis. The simple bar chart is appropriate if only one
variable is to be shown. Figure 10.3 is a simple bar chart of blood group of pharmacy students. Note
that all the bars should be of the same width, and there should be equal spaces between bars. These
spaces emphasise the categorical nature of the data.
60

50

40
Frequency

30

20

10

0
A B AB O
Blood Group
Figure 10.3 Simple bar chart of blood group of pharmacy students

3. The Clustered Bar Chart

If there are more than one group, we can use the clustered bar chart. Suppose we are
knowing the sex of the students in the above example, then it will give us two sub-groups, boys and
girls, with the data shown in Table 10.13. There are two ways of presenting a clustered bar chart.
Figure 10.3 shows one, with blood group categories on the horizontal axis. This arrangement is
helpful if we want to compare the relative sizes of the groups within each category.
Table 10.13 The frequency distribution table of blood group of 95 pharmacy students by sex

Category of Frequency Frequency

Blood group (number of Boys) n=48 (number of Girls) n=47

A 04 11
B 29 20
A 01 03
O 14 13

Alternatively, the chart can also be drawn with the categories boys and girls, on the
 166 Textbook of Computer Applications and Biostatistics

horizontal axis. This format is more useful if we want to compare category sizes within each group.

Example 10.6
Girls with blood group A, B, AB and O can be compared. Which chart is more appropriate
depends on what aspect of the data we want to examine.
35 35
A
30 30
Boys
B
25 25
Frequency

Girls
AB
20 20

Frequency
O
15 15
10 10
5 5
0 0
A B AB O Boys Sex Girls
Blood Group
Figure 10.4 Clustered bar chart of blood group of 95 pharmacy students by sex

4. The Stacked Bar Chart

Figure 10.5 shows a stacked bar chart for the blood group and sex data shown in Table 10.13.
Instead of appearing side by side, as in the clustered bar chart of Figure 10.4, the bars are now stacked
on top of each other. Stacked bar charts are appropriate if we want to compare the total number of
subjects in each group (total number of boys and girls for example), but not so good if we want to
compare category sizes between groups.
Example 10.7 Blood groups of girls with blood groups of boys can be compared.

60

50
O
40 AB
Frequency

30 B
20
A

10

0
Boys Sex Girls

Figure 10.5 Astacked bar chart of blood group of 95 pharmacy students by sex

5. Pictograms
Pictograms are similar to bar charts. They present the same type of information, but the bars
 Presentation of Data 167

are replaced with a proportional number of icons. This type of presentation for descriptive statistics
dates back to the beginning of civilization when pictorial images were used to record numbers of
people, animals or objects.
Example 10.8
Population of different districts of Western Maharashtra. Each diagram indicates one lakh
population.
Pune Satara Kolhapur Sangli

Figure 10.6 Population of different districts of Western Maharashtra

Graphical Presentation of Metric Discrete Data

1. Bar Chart
We can use bar charts to graph discrete metric data in the same way as with ordinal data.
Example 10.9:
The data given in table 10.10 are the number of times inhaler used in past 24 h by 53
children with asthma. Construct bar chart.
18
16
14
12
Frequency

10
8
6
4
2
0
0 1 2 3 4 5 6 7
No of times inhaler used in past 24h

Figure 10.7 The frequency distribution table for metric discrete data

2. Line Plot
A line chart is similar to a bar chart except that thin lines, instead of thicker bars, are used to
represent the frequency associated with each level of the discrete variable.
Example 10.10
Assay result of 30 amoxicillin capsules are as follows
 168 Textbook of Computer Applications and Biostatistics

Data: 251, 250, 253, 249, 250, 252, 247, 248, 254, 245, 250, 253, 251, 250, 249, 252, 249,
251, 246, 250, 250, 254, 248, 252, 251, 248, 250, 247, 251, 249.

Table 10. 14 Frequency distribution table for assay result of 30 amoxicillin tablets
Assay Result Frequency
Tally marks
(mg) n=30
245 | 1
246 | 1
247 || 2
248 ||| 3
249 |||| 4
250 |||| || 7
251 |||| 5
252 ||| 3
253 || 2
254 || 2

Using the data presented in Table 10.14, a corresponding line chart is presented in Figure
10.8. 8
7
6
5
Frequency

4
3
2
1
0
245 246 247 248 249 250 251 252 253 254
Assay Result

Figure 10.8 Line Plot of assay result of 30 amoxicillin tablets

3 . Point Plot
Point plots are identical to line charts, however, instead of a line, a number of points or dots
equivalent to the frequency are stacked vertically for each value of the horizontal axis. Also referred
to as dot diagram, point plots are useful for small data sets.
Using the data presented in Table 10.14, a corresponding point diagram is presented in
Figure 10.9.
 Presentation of Data 169

8
7
6
5

Frequency
4
3
2
1
0
245 246 247 248 249 250 251 252 253 254
Assay Result

Figure 10.9 Point Plot of assay result of 30 amoxicillin tablets

Graphical Presentation of Metric Continuous Data

In all work with graphs, we can use two axes. The vertical axis is always labeled as Y axis. It
is also called as “Ordinate”. The values taken along this axis are called ordinate values. The horizontal
axis is always labeled as ‘X’ axis. It is also called “Abscissa”. The X axis and the Y axis meet at right
angles at a point of origin (O).
In most of the graphs X axis is longer than Y axis. Usually a ratio of 3:2 or 4:3will result in a
good graph. If X axis is taken 18 cm long then Y axis should be 12 cm long. On X axis we will mark
different groups, scores, class intervals, and on Y axis we will usually take frequencies.

1. The Histogram
A continuous metric variable can take a very large number of values, so it is usually
impractical to plot them without first grouping the values. The grouped data is plotted using a
frequency histogram, which has frequency plotted on the vertical axis and group size on the
horizontal axis.
A histogram looks like a bar chart but without any gaps between adjacent bars. This
emphasises the continuous nature of the underlying variable. If the groups in the frequency table are
all of the same width, then the bars in the histogram will also be of the same width. One limitation of
the histogram is that it can represent only one variable at a time (like the pie chart), and this can make
comparisons between two histograms difficult.
Figure 10.10 shows a histogram of the grouped weight in kg of 60 final year pharmacy
students as per the data in Table 10.8.
 170 Textbook of Computer Applications and Biostatistics

16
14
12
10

Frequency
8
6
4
2
0
45-49 50-54 55-59 60-64 65-69 70-74 75-79 80-84
Weight in kg

Figure 10.10 A histogram of the grouped weight in kg of 60 final year pharmacy students

2. Frequency Polygon
A frequency polygon can be constructed by placing a dot at the midpoint (class mark) for
each class interval in the histogram and then these dots are connected by straight lines. This frequency
polygon gives a better conception of the shape of the distribution. The class interval midpoint for a
section in a histogram is
calculated as follows:

Midpoint = (highest + lowest point)/2 ...3

The frequency polygon is then created by listing the midpoints (class marks) on the x axis,
frequencies on the y-axis, and drawing lines to connect the midpoints for each interval.

Construction of frequency polygon

We will take the values in table 10.7 for drawing frequency polygon. Let us mark
frequencies on y-axis and groups or class intervals (weights) on x-axis.
Steps in drawing frequency polygon are given below:
1. Let us take the first group with class interval 45-49. The class intervals in this example are not
continuous and hence the class boundaries are defined first. Then the midpoint (class mark) is
calculated (see table 15) using the formula given above. So, place a point corresponding to 47 on X
axis and 3 on Y axis.
2. The same way, mid point of next group (50-54) is 52. This group is having frequency of 10. So, we
have to place a point corresponding to 52.2 on X-axis and 10 on Y-axis.
3. The same way all frequencies are marked on the corresponding mid points of the groups.
4. Then with a ruler we should connect these points with straight line.
5. Rather than leaving the graph suspended in space, we assume that there is another interval above
and below which is having frequency of zero. So, the mid points of group 40-44 and 85-89 are
assumed having frequency of zero. The group is now allowed to meet X-axis on both ends.
 Presentation of Data 171

Table 10.15
Class limits Class Boundaries Midpoint(Class Mark) Frequency
45-49 44.5-49.5 47 03
50-54 49.5-54.5 52 10
55-59 54.5-59.5 57 12
60-64 59.5-64.5 62 15
65-69 64.5-69.5 67 06
70-74 69.5-74.5 72 07
75-79 74.5-79.5 77 05
80-84 79.5-84.5 82 02

16
14
12
Frequency

10
8
6
4
2
0
40 50 60 70 80 90
Class Marks

Figure 10.11 A Frequency Polygon of the grouped weight in kg of 60 final year pharmacy students

3. Relative Cumulative Frequency Curve (Ogive)

With continuous metric data, there is assumed to be a smooth continuum of values, so we can
chart relative cumulative frequency with a correspondingly smooth curve, known as a relative
cumulative frequency curve, or ogive.
Construction of Ogive
Ogive is a graph of the cumulative relative frequency distribution. So, to draw Ogive we
should convert ordinary frequency distribution into relative cumulative frequency.
Example: Construct relative cumulative frequency diagram for the data given in Example 10.3
Solution
1. Construct the table of the cumulative frequency distribution as shown below. Here cumulative
frequency means “Total number of students in each particular weight range from lowest value to that
particular value”. It is obtained by cumulating the frequency of previous classes including the class in
question.
 172 Textbook of Computer Applications and Biostatistics

Table 10.16

Class limits Class Boundaries Midpoint Frequency Cumulative Relative cumulative

frequency frequency
45-49 44.5-49.5 47 03 03 05.0
50-54 49.5-54.5 52 10 13 21.7
55-59 54.5-59.5 57 12 25 41.7
60-64 59.5-64.5 62 15 40 66.7
65-69 64.5-69.5 67 06 46 76.7
70-74 69.5-74.5 72 07 53 88.3
75-79 74.5-79.5 77 05 58 96.7
80-84 79.5-84.5 82 02 60 100
2. Now plot the relative cumulative frequencies on y-axis while midpoints of class intervals on x-axis.
So, place a point corresponding to 47.5 on x-axis and 5 on y-axis. Likewise plot all relative
cumulative frequencies on corresponding midpoints of the group.
3. Now, join the points with a free hand to give a smooth curve. This is the Ogive as shown below.

100
% Relative cumulative frequency

75th percentile
75

50th percentile
50

25th percentile Q3
25 Q2
Q1

0
40 50 60 70 80 90
Class Marks

Figure 10.12 Ogive of the grouped weight of 60 final year students

Applications of Ogive
1. By using Ogive we can locate any percentile that will divide the series into two parts.
2. Quartiles: There are three different points located on the entire range of variable (here it is weight in
kg). These are Q1, Q2, Q3.
Q1 or lower quartile will have 25% observations falling in its left and 75% observations on
its right side.
Q2 is the median, i.e., 50% values lies on either side.
Q3 is the upper quartile, will have 75% observations falling on its left side and 25%
 Presentation of Data 173

observations on its right side.

By using these quartiles we can calculate semi inter quartile range and inter quartile range
(Q1-Q3).
3. Quintiles: This divides the distribution into 5 equal parts. So, 20th percentile or 1st quintile will
have 20% observations falling to its left and 80% to its right.
4. Deciles: This divides the distribution into 10 equal parts. First decile (10th percentile) will have
10% values to its left and 90% values to its right. 5th decile is the median and contains 50% values on
either side.

4. Stem and Leaf plot

The stem-and-leaf plot is a visual representation for continuous data and contains features
common to both the frequency distribution and dot diagrams. Digits, instead of bars are used to
illustrate the spread and shape of the distribution. Each piece of data is divided into "leading" and
"trailing" digits. For example, based on the range of data points, the observation 116 can be divided
into either 11 and 6, or 1 and 16, as the leading and trailing digits. All the leading digits are sorted from
lowest to highest and listed to the left of a vertical line. These digits become the stem. The trailing
digits are then written in the appropriate location to the right of the vertical line. These become the
leaves.

Example 10. 11
Plot the Stem and Leaf graph for following data of age in years of 27 patients.
8, 13, 16, 25, 26, 29, 30, 32, 37, 38, 40, 41, 44, 47,
49, 51, 54, 55, 58, 61, 63, 67, 75, 78, 82, 86, 95
Solution
1. We can group 27 patients (8-95 yrs) according to age groups. Here we will take groups of 10 yrs
each.
0-9 10-19 20-29 30-39 40-49 50-59 60-69 70-79 80-89 90-99
2. We can now write these groups on left side of the vertical line and place the last digit of the age on
the right of the vertical line to the corresponding group (see figure 10.13).
3. Now, groups can be considered as stem and corresponding digits on the right as leaves. This can be
rewritten in figure 10.14.
4. From stem and leaves, we can read the exact ages of the patients and the data is not lost.

5. Box and Whiskers plot

One simple plot that displays a great deal of information about a continuous variable is the
box-and-whisker plot. The box plot illustrates the bulk of the data as a rectangular box in which the
upper and lower lines represent the third quartile (75% of observations below Q3) and first quartile
(25% of observations below Ql), respectively. The second quartile (50% of the observations below
this point) is depicted as a horizontal line through the box. Vertical lines (whiskers) extend from the
 174 Textbook of Computer Applications and Biostatistics

top and bottom lines of the box to an upper and lower adjacent value. The details of drawing box and
whiskers plot is discussed in next chapter (Example 12.5).

Stem Leaves
0-9 8
10-19 36
20-29 569
30-39 0278
40-49 01479
50-59 1458
60-69 137
70-79 58
80-89 26
90-99 5
Figure 10.13 Stem and Leaf plot of grouped data

Stem Leaves
0 8
1 3 6
2 569
3 0278
4 01479
5 1458
6 137
7 5 8
8 2 6
9 5
Figure 10.14 Stem and Leaf plot of ungrouped data

6. Scatter diagram or Scatter plot

A scatter diagram is an extremely useful presentation for showing the relationship between
two continuous variables. The two dimensional plot has both horizontal and vertical axes which cover
the ranges of the two variables. Plotted data points represent paired observations for both the x and y
variable (Figure 10.14). These types of plots are valuable for correlation and regression inferential
tests.
 Presentation of Data 175

Example 10.12:
The table shows the age and the weight of a child. Plot the scatter graph.
Table 10.17 Age and weight of five children

Age in years Weight in pounds

2 20
3 25
4 28
5 30
6 35

Solution
1. Take the weight of children on y-axis while age in years on x-axis.
2. Take appropriate scale on y-axis covering the weight from 20-35.
3. Take appropriate scale on x-axis covering the age from 1-6.
4. Now put the marks of weight to the corresponding ages of children.
5. Draw the line connecting the marks (points) to get the scatter plot.

40

35
Weight (Pound)

30

25

20

15
0 2 4 6 8
Age (years)

Figure 10.15 Scatter plot

7. Time Series chart

If the data collected are from measurements made at regular intervals of time (minutes,
weeks, years, etc.), we can present the data with a time series chart. Usually these charts are used with
metric data, but may also be appropriate for ordinal data. Time is always plotted on the horizontal
axis, and data values on the vertical axis.
 176 Textbook of Computer Applications and Biostatistics

Charting of Data Using MS - Excel

1. Construction of Pie Chart
Example 10.13
Blood group of 95 students are as follows
Category of blood group A B AB O
Number of students 49 27 15 04
Excel Solution
Step I 1. Open New MS Excel File.
2. Put labels in column A1 as “Category of blood group” and B1 as “Number of students”.
3. Put values of blood group from cell A2 toA5 and number of students from Cell B2 to B5.
4. Sheet will appear as shown in figure 10.16.

Figure 10.16 Entering data into worksheet

Step II 1. Select cells fromA1 to B5.
2. Click on Insert from menu bar. Instantly pull down menu will appear.
3. Select ‘Chart’from pull down menu. Instantly “Chart Wizard- Step 1 of 4 - Chart Type”
display box will appear as shown in figure 10.17.

Figure 10.17 Window of Chart type

4. Select ‘Pie’ from Chart type as shown in figure 10.7 and click on Next button
 Presentation of Data 177

Figure 10.18 Window of selecting Pie Chart

5. Instantly Chart Wizard- Step 2 of 4- Chart Source Data will appear as shown in figure
10.19.
6. If the appeared graph is correct then click on next button.
7. If Graph is not correct click on series and make the corrections by changing Values and
Category Labels, as shown in figure 10.20. Then click on Next button.

Figure 10.19 Window of Chart Source Data Figure 10.20 Window of selecting Series option

8. Instantly Chart Wizard - Step 3 of 4- Chart Options will appear.

9. Click to Value from ‘Label Contains’ as shown in figure 10.21and then click to Next
178 Textbook of Computer Applications and Biostatistics

button.

Figure 10.21 Window of Chart Options

10. Instantly Chart Wizard - Step 4 of 4 - Chart Location will appear as shown in figure 10.22

Figure 10.22 Window of Chart Location

11. Ignore the things and click on ‘Finish’button. Following graph will be displayed on same
sheet.
Number of Students
4

15

49

27

Figure 10.22 Pie Chart

13. We can make changes in the graph properties according to our needs.
 Presentation of Data 179

2. Construction of Line or Scatter Graph in Excel

1. First put the values of X and Y in the excel sheet and label the colums. Make sure that your X values
are in one column and that Y values are in a second column with each row containing the matching X,
Y values.
2. Now click on the picture of a graph in the toolbar at the top. Alternatively, we can go to “Insert
Chart” from the drop-down menus.
3. Step 1 of the Chart Wizard will appear as shown in previous example. Pick the picture and
description that best matches the type of graph we are trying to draw. Choose the “XY (Scatter)”
category. We can additionally pick “Scatter with data points connect by smoothed lines” or “Scatter
with data points connected by lines” to connect the points with a smoothed line or with straight lines.
Click “Next”.
4. Step 2 of the Chart Wizard, Chart Source Data, will appear. Go to the tab ‘Series’. It contains the
following fields. Fill in the fields as listed below:
a. Name: Fill in the words that will appear in the legend.
b. X Values: Click on the picture of the spreadsheet at the right of the blank. This takes us to the
spreadsheet where we can highlight data for X (just the numbers, not the column heading).
c. Y Values: Click on the spreadsheet, and highlight data for Y (just the numbers, not the column
heading).
Click Next.
5. Step 3 of the Chart Wizard has several different tabs, as shown below:
a. Titles tab –
Chart Title: This is the graph title that will appear above the graph by default.
Value X axis: Enter the X axis label here. Don't forget to include appropriate units in
parentheses.
Value Y axis: Enter the Y axis label here. Don't forget to include appropriate units in
parentheses.
b.Axes tab – We can alter the appearance of the X or Y-axis.
c. Gridlines tab – We can check the boxes on this tab to make gridlines appear or disappear.
d. Legend tab – We can check the boxes on this tab to make series legend appear or disappear
and even we can indicate where on the page legend should appear.
e. Data Labels tab – We can check boxes to show specific data labels and values. click “Next”
after making the choices on step 3.
6. Step 4 of the chart wizard simply asks where we want graph to appear. We can check for the graph
to appear by itself on a separate sheet in the workbook, or to appear as an object embedded in the
current sheet of the notebook, next to the data.

3. Construction of a Bar Graph in Excel

1. Put the X axis and Y axis label in the Excel sheet. Then type the category data (word or numbers) in
column of X axis label and type corresponding frequency values in column of Y axis label.
 180 Textbook of Computer Applications and Biostatistics

2. Now click on the picture of a graph in the toolbar at the top. Alternatively, we can go to “Insert
Chart” from the drop-down menus.
3. Step 1 of the Chart Wizard will appear. A bar graph is called a “Column Chart” in Excel; so select
“Clustered Column Chart”. Select that and click “Next”
4. Step 2 of the Chart Wizard has 2 tabs, the “Data Range” & “Series” tabs.
a. The “Data Range” tab asks us to pick data range. If data is not highlighted previously we
can use the selection tool to do so from this screen.
b. Click on the “Series” tab.
In the space labeled “Name” we can type the title of the graph.
In the space labeled “Category (x) axis labels”, we can highlight a data range that contains
the words or numbers that we would like to appear under each bar. Click on the picture of the
spreadsheet beside the blank to allow us to highlight the range on the spreadsheet. Hit “enter”, then
click “Next”.
5. Step 3 of the Chart Wizard has several different tabs.
a. Titles tab –
Chart Title: This is the graph title that will appear above the graph by default. We have
already entered title under step 2.
Category X axis : Enter the X axis label here.
Value Y axis: Enter the Y axis label here.
b. Axes tab – We can change the type of scaling on X axis & Y axis.
c. Gridlines tab – We can check the boxes on this tab to make gridlines appear or disappear.
d. Legend tab – We can check the boxes to make series legend appear or disappear.
e. Data Labels tab – We can check boxes to show specific data labels and values.
f. Data Table tab – We can use this tab showing the data table on the graph.
g. Hit “Next” after making choices on step 3.
6. Step 4 of the chart wizard simply asks where we want the graph to appear. We can check for the
graph to appear by itself on a separate sheet in the workbook, or to appear as an object embedded in the
current sheet of the notebook, next to the data.

Summary
1.Tabulation of data
Frequency distribution table
1. Frequency: Repetition of observations
2. Relative frequency: Frequency converted into Percentage
3.Cumulative frequency: Cumulative total of frequencies
4. Cumulative relative frequency: Cumulative frequency converted into percentage
2 . Graphical presentation of nominal and ordinal data
1. Pie Chart 2. Simple Bar Chart 3. Clustered Bar Chart
4. Stacked Bar Chart 5. Pictogram
 Presentation of Data 181

3. Graphical presentation of metric discrete data

1. Bar Chart 2 . Line Chart 3. Point Chart
4. Graphical presentation of metric continuous data
1. Histogram 2. Frequency polygon
3. Ogive 4. Stem and Leaf plot
5. Box and Whisker plot 6. Scatter plot
7. Time series plot.
Choosing an appropriate chart
Type of data
Chart Type Nominal Ordinal Metric discrete Metric Continuous
Pie chart yes no no no
Bar chart yes yes yes no
Histogram no no yes yes
Frequency polygon no no yes yes
Ogive no no yes (cumulative) yes (cumulative)
Box & Whiskers no no no yes
Stem & Leaf no no no yes

Multiple choice questions

1. Which of the following would be most suitable for displaying the proportions of a budget spent on
different items by pharmaceutical company?
a. Pie chart b. Bar chart
c. Line graph d. Histogram
2. Bar charts may be distinguished from histograms at a glance because:
a. bar charts are not used for time series data
b. histograms are used to display discrete data
c. bar charts are based on area under the curve
d. histograms do not have spaces between consecutive columns
3.Agraph that uses vertical bars to represent data is called a ____.
a. Line graph b. Bar graph
c. Scatter plot d. Vertical graph
4. Ogive is ______.
a. Subset of population b. Data collected over a period of time
c.Avariable with qualitative data d.Agraph of cumulative distribution
5. Which of the following is used for graphical presentation of metric discrete data.
a. Bar Chart b. Line Chart
c. Point Chart d.All of above
 182 Textbook of Computer Applications and Biostatistics

6. The difference between class boundaries is called _______.

a. class limit b. class interval
c. class frequency d. class mark
7. When the value is away from general mass of data, it is called as _________.
a. mean b. open ended
c. outlier d. close ended
8. _______ divides the distribution into 5 equal parts.
a. Quartile b. Quintiles
c. Deciles d. Median
9. _______ plot is useful for showing the relationship between two continuous variables.
a. Scatter b. Point
c. Stem and leaf d. Box and Whiskers
10. Cumulative frequency converted into percentage is called________.
a. relative frequency b. cumulative relative frequency
c. frequency d. none of above

Exercise
1. Based on 1210 patients involved in clinical trials of an anticancer drug, it was observed that 841
experienced no adverse effects, while 256, 91 and 22 subjects suffered mild, moderate and severe
adverse effects respectively. Prepare tabular and graphical presentation for this data with proper
reasoning for choosing particular graphs.
2. The following assay results (percentage of label claim) were observed in 50 random samples during
a production run.
102, 100, 96, 99, 101, 102, 100, 105, 97, 100, 92, 103, 101, 100, 99, 102, 96, 100,
101, 98, 107, 95, 98, 100, 100, 99, 97, 104, 101, 103, 98, 101, 100, 105, 99, 101, 102, 100,
87, 98, 101, 103, 93, 99, 101, 97, 100, 102, 99, 104.
Tabulate the data and report results as a stemplot.
3. Construct table of frequency, relative frequency, cumulative frequency and relative cumulative
frequency for following data of weights in kg of 50 pharmacy students and present data graphically.
51, 53, 52, 39, 58, 48, 45, 56, 62, 64, 66, 67, 42, 48, 43, 44, 45, 47, 52, 54, 50, 49, 38, 39, 38,
32, 34, 36, 33, 34, 36, 38, 42, 43, 48, 49, 50, 51, 52, 30, 31, 30, 32, 45, 45, 47, 46, 48, 49, 58.
4. Prepare Ogive curve for the following data.
Height of groups (cm) 72-76 76-80 80-84 84-88 88-92 92-96 96-100
Frequency 5 7 10 15 11 9 7
5. Construct scatter plot of the responses given by a drug in 15 patients.
Concentration of drug (mg/l) 4 6 8 10 12 14 16 18 20 22 24 26
Response 1 2 3 4 5 6 7 8 9 10 11 12
6. Following data provides the number of deaths for several leading causes for the year 2010. Display
 Presentation of Data 183

the results in a pie chart.

Cause of death Number of deaths
Heart disease 10382
Cancer 8299
Cerebrovascular disease 2830
Accidents 1381
Other causes 11476
7. Draw a frequency polygon for the data given in Exercise 3.
8. The following is the distribution of number of painkiller tablet taken by 60 patients in last one week.
Number of tablets 0-1 2-3 4-5 6-7 8-9
Frequency 16 25 13 4 2
Find class marks, class boundaries and class intervals of the distribution.
9. List the data which correspond to the following stem and leaf plots:
a) 12 | 4 0 3 8 7 6 6 5
b) 34 | 42 05 19 61
10.Among histograms, bar charts, and pie charts, which ones can be used to represent
a) nominal data b) ordinal data c) metric data

Answers:
Multiple Choice Questions
1. a 2. d 3. b 4. d 5. d 6. b 7. c 8. b 9. a 10. b

Exercise
1. Tabular presentation of data
Grade of Frequency Relative Cumulative Relative Cumulative
adverse effect n=1210 frequency frequency Frequency
0 841 69.50 841 69.50
1 256 21.15 1097 90.66
2 91 7.52 1188 98.18
3 22 1.81 1210 100
No adverse effect (0); Mild adverse effect (1); Moderate adverse effect (2); Severe adverse effect (3)
Graphical presentation of data: The given data is ordinal categorical type and hence can be
represented as Bar chart.
 184 Textbook of Computer Applications and Biostatistics

900

800

700

600

No of patients
500

400

300

200

100

0
No adverse Mild adverse Moderate Severe
effects effects adverse adverse
effects effects
2. Stem plot: The given data varies from 87 to 107 and therefore the stems should be 8, 9 and 10

8 7
9 235667778888999999
10 0000000000111111112222233344557
Stems Leaves
Histogram is to be given.

3. Tabular presentation of data

Frequency Relative Cumulative Relative Cumulative

Class interval frequency frequency Frequency
n=50
30-35 8 16 8 16
35-40 7 14 15 30
40-45 5 10 20 40
45-50 14 28 34 68
50-55 9 18 43 86
55-60 3 6 46 92
60-65 2 4 48 96
65-70 2 4 50 100

As the data is metric continuous, histogram is the best choice

 Presentation of Data 185

16

14

12

10

Frequency
8

0
30-35 35-40 40-45 45-50 50-55 55-60 60-65 65-70
Weight in kg

4. Ogive 120
% Cumulative frequency

100

80

60

40

20

0
70 80 90 100
Mind point of height of groups (cm)

5. Scatter plot
14

12

10

8
Response

0
0 10 20 30
Concentration of drug (mg/l)
 186 Textbook of Computer Applications and Biostatistics

6. Pie Chart
Other causes 34%
Accidents 4%

Cerebrovascular disease 8%

Heart disease 30%

Cancer 24%

Leading cause of Death

7. Frequency polygon
16

14

12

10
Frequency

0
25 35 45 55 65 75
Weight in kg

8. a) Class marks: 0.5 2.5 4.5 6.5 8.5

b) Class boundaries: -0.5 1.5 3.5 5.5 7.5 9.5
c) 2
9. a) 124 120 123 128 127 126 126 125
b) 3442 3405 3419 3461
10. a) nominal data: Histogram
b) ordinal data: Bar chart
c) metric data: Nominal data
 Chapter 11
SHAPE OF DISTRIBUTION OF DATA

Learning objectives
When we have finished this chapter, we should be able to:
1. Explain what is meant by the ‘shape’of a frequency distribution.
2. Sketch and explain: negatively skewed, symmetric and positively skewed distributions.
3. Sketch and explain a bimodal distribution.
4. Describe the approximate shape of a frequency distribution from a frequency table or chart.
5. Sketch and describe a Normal distribution.

Shape of distribution of data

The choice of the most appropriate procedures for summarising and analysing data will not
only depend on the type of variable but also on the shape of the distribution.
The shape of distribution of data may be of following types:
1. Uniform distribution: The values are fairly evenly spread throughout their possible range. This is
a uniform distribution.
5

0
0 2 4 6 8 10
Figure 11.1 Chart showing uniform distribution of data

2. Positively skewed distribution: The values are concentrated towards the bottom of the range, with
progressively fewer values towards the top of the range. This is a right or positively skewed
distribution. See figure 11.2 for positively skewed distribution.
3. Negatively skewed distribution: The values are concentrated towards the top of the range, with
progressively fewer values towards the bottom of the range. This is a left or negatively skewed
distribution. See figure 11.3 for negatively skewed distribution.

187
 188 Textbook of Computer Applications and Biostatistics

0
0 2 4 6 8 10

Figure 11.2 Chart showing positively skewed distribution of data

0
0 2 4 6 8 10
Figure 11.3 Chart showing negatively skewed distribution of data

4. Symmetric or Mound-shaped distribution: The values are clumped together around one
particular value, with progressively fewer values both below and above this value. This is a
symmetric or mound-shaped distribution
5

0
0 2 4 6 8 10

Figure 11.4 Chart showing symmetric or mound shaped distribution of data

 Shape of Distribution of Data 189

5. Bimodal or Multimodal distribution: The values are clumped around two or more particular
values. This is a bimodal or multimodal distribution.
5

0
0 2 4 6 8 10
Figure 11.5 Chart showing bimodal distribution of data
One simple way to assess the shape of a frequency distribution is to plot a bar chart, or a
histogram.
5

0
0 2 4 6 8 10
Figure 11.6 Frequency curve superimposed on a histogram

Bell Shaped distribution (Normal distribution)

There is one particular symmetric bell-shaped distribution, known as the Normal
distribution. Many human clinical features are distributed normally, and the normal distribution has a
very important role to play.
Characteristics of normal distribution
1. It is bell shaped curve
2. It is symmetrical in distribution; variables on either side of mean are equal in number.
3. Its maximum height is at the mean.
4. Mean= mode= median , in case of normal distribution coincide.
 190 Textbook of Computer Applications and Biostatistics

5. Skewness of the curve is zero.

6. It is asymptotic, in that tails never touch baseline.
7. Total area of curve is one and standard deviation is also one.
8. It has two curves. Central part is convex and when it comes down, it becomes concave on both
sides.

35

30

25

20

15

10

0
0 1 2 3 4 5 6 7 8 9 10
Figure 11.7 Frequency curve superimposed on a histogram

Skewness
Skewness measures asymmetry around the mean. The parameter is best interpreted as
relative to the normal distribution (whose skewness equals to zero). The interpretation of the
skewness is
- Skewness > 0 asymmetric tail with more values above the mean
- Skewness < 0 asymmetric tail with more values below the mean
Skewned data is required to be treated using non parametric tests while normal curve data is
treated using parametric tests.

Kurtosis
Kurtosis is a property associated with a frequency distribution and refers to the shape of the
distribution of values regarding its relative flatness and peakedness. Compared with normal
distribution, the interpretation of the kurtosis is:
Kurtosis > 0 peaked relative to Normal distribution
Kurtosis < 0 flat relative to Normal distribution
Here are some examples of the shapes described above.
 Shape of Distribution of Data 191

6 Peaked (Leptokurtic)
5

4 Normal (Mesokurtic)
3
Flatter ( Platykurtic)
2

0
0 2 4 6 8 10
Figure 11.8 Chart showing Kurtosis of distribution of data

Summary
Shapes of data:
1. Uniform distribution
2. Positively skewed distribution
3. Negatively skewed distribution
4. Symmetric or Mound-shaped distribution
5. Bimodal or Multimodal distribution
If the data is skewed it is required to be treated using non parametric tests while normal data
is treated using parametric tests.
Normal distribution:
It is a bell shaped curve, symmetric and usually has Mean = Mode = Medium
Skewness
Skewness measures asymmetry around the mean.
Kurtosis
Kurtosis measures relative flatness and peakedness.

Multiple Choice Questions

1. If a distribution is skewed to the left, then it is __________.
a. Negatively skewed b. Positively skewed
c. Symmetrically skewed d. Symmetrical
2. If a test was generally very easy, except for a few students who had very low scores, then the
distribution of scores would be _____.
a. positively skewed b. negatively skewed
c. not skewed at all d. normal
 192 Textbook of Computer Applications and Biostatistics

3. A measure of asymmetry is given by_________.

a. kurtosis b. normal distribution
c. skewness d. uniform distribution
4. A bell shaped distribution of data is said to be________.
a. normal distribution b. positively skewed distribution
c. negatively skewed distribution d. bimodel distribution
5. If the shape of distribution is flatter as compared to normal distribution, then the curve is _______.
a. Leptokurtic b. Platykurtic
c. Mesokurtic d. Skewed

Exercise:
1. Describe shapes of data.
2. Write characteristics of normal distribution.
3. Why knowing the shape of the data is important?
4. Explain skewness with the help of diagram.
5. Explain kurtosis with the help of diagram.

Answers:
Multiple Choice Questions
1. a 2. b 3. c 4. a 5. b
 Chapter 12
MEASURES OF CENTRAL TENDENCY

Learning objectives
When we have finished this chapter, we should be able to:
1. Explain what a summary measure of location is, and calculate the mode, median and mean for a set
of values.
2. Demonstrate the role of data type and distributional shape in choosing the most appropriate
measure of central tendency.
3. Explain what a percentile is, and calculate any given percentile value.
4. Explain what a summary measure of spread is, and calculate, the range, the interquartile range and
the standard deviation.
5. Estimate percentile values from an ogive.
8. Demonstrate the role of data type and distributional shape in choosing the most appropriate
measure of spread.
9. Draw a boxplot and explain how it works.
10. Explain the use of Excel in estimating all measures of central tendency.

Introduction
As we have seen in the previous two chapters, we can ‘describe’ raw data by charting it, or
arranging it in table form or we can examine its shape. These procedures helps us to see patterns in the
data. However, it is often more useful to summarise the data numerically. There are two principal
features of a set of data that can be summarised with a single numeric value:
1. Measure of Location: A value around which the data has a tendency to congregate or cluster, is
called as summary measure of location.
2. Measure of Dispersion: A value which measures the degree to which the data are spread out is
called a summary measure of spread or dispersion.
With these two summary values we can compare different sets of data quantitatively.

Summary measures of location

A summary measure of location is a value around which most of the data values tend to
congregate or centre. Let us discuss three measures of location: the mode; the median; and the mean.

1. The Mode
The mode is that category or value in the data that has the highest frequency (i.e. it occurs
most often).The mode is not particularly useful with metric continuous data where no two values are

193
 194 Textbook of Computer Applications and Biostatistics

same. The other shortcoming of this measure is that there may be more than one mode in a set of data.

1. Mode for Ungrouped data

Example 12.1
Calculate mode for ungrouped data given below
6, 8, 9, 7, 12, 3, 2, 4, 8, 1, 8, 5
Solution
1. First arrange the data in ascending order
1, 2, 3, 4, 5, 6, 7, 8, 8, 8, 9, 12.
2. Mode is frequently occuring value. In this example 8 is frequently occuring and hence mode is 8.

2. Mode for grouped data

The mode for grouped data can be calculated by using the formula
æ f1 - f 0 ö
Mode = l1 + çç ÷÷ ´ i
è 2f1 - f 0 - f 2 ø
...1
Where,
l1 = Lower limit of modal class
f1 = Frequency of modal class
f0 = Frequency before the modal class
f2 = Frequency after the modal class
i = Class interval of modal class

Example 12.2
Calculate mode for following grouped data
Table 12.1 Frequency distribution of Sales Per day

Sales volume (Class interval) 53-56 57-60 61-64 65-68 69-72 72 and above
Number of days (Frequency) 2 4 5 4 4 1

Solution
Since the largest frequency corresponds to the class interval 61-64, hence it is the modal
class.
l1 = Lower limit of modal class = 61; f1 = Frequency of modal class = 5;
f0 = Frequency before the modal class = 4; f2 = Frequency after the modal class = 4
i = Class interval of modal class = 3
Formula
æ f1 - f 0 ö
Mode = l1 + çç ÷÷ ´ i
è 2f1 - f 0 - f 2 ø
 Measures of Central Tendency 195

æ 5- 4 ö
Mode = 61 + ç ÷ ´ 3 = 62.5
è 10 - 4 - 4 ø
Hence, the modal sale is of 62.5 units.

Example 12.3
Calculate mode for following grouped data
Table 12.2 Frequency distribution for the heights of the Pharmacy students

Height (inches) 57-59 59-61 61-63 63-65 65-67 67-69

Frequency 47 23 65 51 20 08

Solution
Since the largest frequency corresponds to the class interval 61-63, hence it is the modal
class.
l1 = Lower limit of modal class = 61; f1 = Frequency of modal class = 65;
f0 = Frequency before the modal class = 23; f2 = Frequency after the modal class = 51
i = Class interval of modal class = 2
Formula
æ f1 - f 0 ö
Mode = l1 + çç ÷÷ ´ i
è 2f1 - f 0 - f 2 ø

æ 65 - 51 ö
Mode = 61 + ç ÷ ´ 2 = 61.5
è 130 - 23 - 51 ø
Hence, the modal height is of 61.5 inches.

2. The Median
If we arrange the data in ascending order of size, the median is the middle value. Thus, half of
the values will be equal to or less than the median value, and half will be equal to or above it. The
median is thus a measure of central-ness. If we have an even number of values, the median is the
average of the two values either side of the ‘middle’.
An advantage of the median is that it is not much affected by skewness in the distribution, or
by the presence of outliers. However, it discards a lot of information, because it ignores most of the
values, apart from those in the centre of the distribution.

1. Median for ungrouped data

In this case the data is arranged in either ascending or descending order of magnitude. If the
number of observations (n) is an odd number, then the median is represented by the numerical value
corresponding to (n+1)/2th ordered observation.
 196 Textbook of Computer Applications and Biostatistics

æ n +1ö
Median = Size or value of ç ÷ th observation
è 2 ø ...2
If the number of observations (n) is an even number, then the median is defined as the
arithmetic mean of the numerical values of n/2th and (n/2 + 1)th observations in the data array.
n æn ö
th + ç + 1÷ th observation
Median =
2 è 2 ø ...3
2
Example 12.4
Calculate the median of the following data that relates to the number of patients per day in the
outpatient ward in a Civil Hospital.
100, 200, 120, 170,130, 150, 180.
Solution:
1. First arrange the data in an ascending order
100, 120, 130, 150, 170, 180, 200.
2. Since the number of observations in the data array are odd (n=7), the median for this data is
æ n +1ö
Median = Size or value of ç ÷ th observation
è 2 ø
æ 7 +1ö
Median = ç ÷ = 4th observation
è 2 ø
4th observation in the data array = 150.
Thus the median number of patients examined per day in OPD in a Civil Hospital are 150.

Example 12.5
Calculate the median of the following data that relates to the sale in lakh per month of a
company in last one year. 12, 18, 15, 14, 13, 12, 20, 10, 11, 18, 19, 16
Solution:
1. First arrange the data in ascending order
10, 11, 12, 12, 13, 14, 15, 16, 18, 18, 19, 20.
2. Since the number of observations in the data array are even (n=12), the median for this data is
n æn ö
th + ç + 1÷ th
Median =
2 è2 ø
2

12 æ 12 ö
th + ç + 1÷th
Median =
2 è2 ø = (6th value + 7th value) = 14 + 15 = 14.5
2 2 2

Thus the median sale of company per month is 14.5 lakhs.

 Measures of Central Tendency 197

2. Median for grouped data (Metric continuous)

To find the median for grouped data, first we need to identify the class interval which
contains the median value or (n/2)th observation of the data set. To identify such class interval, we
should find the cumulative frequency of each class until the class for which the cumulative frequency
is equal to or greater than the value of (n/2)th observation. The value of the median within that class is
found by using interpolation. It is assumed that the observation values are evenly spaced over the
entire class interval. The following formula is used to determine the median of grouped data:
(n/2) - c.f.
Median = l1 + ´ i ...4
f
Where
l1 = Lower limit of median class.
c.f. = Cumulative frequency of the class prior to the median class.
f = Frequency of median class.
i = Class interval of median class.
th
Median class is the class in which n/2 observation lies. To use above formula, data should be
continuous.

Example 12.6
A survey was conducted to determine the age (in years) of 130 pharmacists. The result of
such a survey is as follows:
Table 12.3 Frequency distribution of age of Pharmacist

Age of pharmacist 20-25 25-30 30-35 35-40 40-45 45-50 50-55

Number of Pharmacist 6 13 29 48 22 8 4

Solution:
First, we should find the cumulative frequencies to locate the median class ( see table 12.4).

Table 12.4 Calculations for median age of pharmacist

Age of pharmacist No of pharmacist Cumulative frequency

(in years) (f) (cf)
20-25 06 06
25-30 13 19
30-35 29 48
35-40 48 96 Median class
40-45 22 118
45-50 08 126
50-55 04 130
n=130
 198 Textbook of Computer Applications and Biostatistics

Here the total number of observations are n = 130. Median is the size of (n/2)th = 130/2 =
65th observation in the data set. This observation lies in the class interval 35-40.
l1 = Lower limit of median class = 35
c.f.= Cumulative frequency of the class prior to the median class interval = 48
f = Frequency of median class = 48
i= Class interval of median class = 5
Formula
(n/2) - c.f.
Median = l1 + ´ i
f
(130/2) - 48 17
Median = 35 + ´ 5 = 35 + ´ 5 = 35 + 1.77 = 36.77
48 48
Hence the median age of pharmacist is 36.77 years.

Example 12.7
A survey was conducted to determine the height (in inches) of 50 pharmacists. The result of
such a survey is as follows:
Table 12.5 Frequency distribution of age of Pharmacist

Height of pharmacist 44-48 48-52 52-58 58-62 62-66 66-70 70-74

Number of Pharmacist 1 2 7 8 18 12 2

Solution:
First, we should find the cumulative frequencies to locate the median class ( see table 12.6).

Table 12.6 Calculations for median age of pharmacist

Height of pharmacist No of pharmacist Cumulative frequency

(in inches) (f) (cf)
44-48 01 01
48-52 02 03
52-58 07 10
58-62 08 18
62-66 18 36 Median class
66-70 12 48
70-74 02 50
n=50
Here the total number of observations are n = 50. Median is the size of (n/2)th = 50/2 = 25th
observation in the data set. This observation lies in the class interval 62-66.
l1 = Lower limit of median class = 62
 Measures of Central Tendency 199

c.f.= Cumulative frequency of the class prior to the median class interval = 18
f= Frequency of median class = 36
i= Class interval of median class = 4
(n/2) - c.f.
Median = l1 + ´i
f
(50/2) - 18
Median = 62 + ´ 4 = 62.78
36
Hence the median age of pharmacist is 62.78 inches.

3. Median for Metric Discrete data

Median for metric discrete data is calculated as follows
Median = value of (n/2)th observation
Example 12.8
The information on the number of defective components in 1000 boxes, is given below
Table 12. 7 Data of number of defective components in 1000 boxes

No. of defective components 0 1 2 3 4 5 6

Number of boxes 25 306 402 200 51 10 6
Calculate the median of defective components for the whole of the production line.
Solution
For the calculation of median defective components for the whole production line the
frequency table as shown in table 12.8 is required.
Table 12.8 Frequency table for calculation of median for Metric discrete data

No. of defective No. of Boxes Cumulative frequency

components (f) (cf)
0 25 25
1 306 331
2 402 733
3 200 933
4 51 984
5 10 994
6 06 1000
n = 1000

Median = value of (n/2)th observation

= value of (1000/2) th observation
= value of 500th observation
 200 Textbook of Computer Applications and Biostatistics

In cumulative frequency column 500th observation comes after 331 and before 733.
Hence median is 2 defective components.

3. The Mean
The mean, or the arithmetic mean, is more commonly known as the average. One advantage
of the mean over the median is that it uses all of the information in the data set. However, it is affected
by skewness in the distribution, and by the presence of outliers in the data. This may, sometimes,
produce a mean that is not very representative of the general mass of the data. Moreover, it cannot be
used with ordinal data (as ordinal data are not real numbers, so they cannot be added or divided).

1. Mean for Ungrouped data

The mean of ungrouped data is calculated by adding the values of all observations and
dividing the total by the number of observations.
Sum of all observations ( å X)
Mean ( X ) = ...5
Total number of observations (N)

Example 12.9
The following data gives weight of 20 paracetamol tablets in mg. Calculate average weight
of a paracetamol tablet.
625, 617, 633, 630, 620, 631, 618, 620, 619, 632,
625, 628, 626, 624, 622, 625, 627, 631, 619, 624.
Solution:
Sum of all observations (å X) 12496
Mean (X) = = = 624.8
Total number of observations (N) 20

Hence mean weight of paracetamol tablet is 624.8.

Example 12.10
The blood serum cholesterol levels of 10 subjects are given as:
245 262 292 247 253 286 274 265 279 252
Calculate mean.
Solution:
Sum of all observations (å X) 2655
Mean (X ) = = = 265.5
Total number of observations (N) 10

Hence mean cholesterol levels of 10 subject is 265.5.

 Measures of Central Tendency 201

2. Mean for metric continuous data (grouped data)

Mean for metric continuous grouped data can be calculated by using following formula

Mean ( X ) = A +
å fd ´ i
N ...6
Where
A=Assumed mean
f = frequency of ith class interval
N = Summation of all frequencies
d = (mi -A)/ i, deviation from assumed mean
m = mid value of ith class interval
i = width of the class interval

Example 12.11
A company is planning to improve plant safety. For this, accident data for the last 50 weeks
was compiled. These data are grouped into the frequency distribution as shown below. Calculate
mean of the number of accidents per week.
Table 12.9 Number of accidents in last 50 weeks
No. of accidents 0-4 5-9 10-14 15-19 20-24
Number of weeks 5 22 13 8 2

Solution
1. Construct frequency distribution table as per given in table 12.10.
Table 12.10 Calculations for mean accidents per week

No. of Accidents No of weeks Mid value of Deviation from

(f) class interval (m) assumed mean (d ) fd
class interval
0-4 5 02 -2 -10
5-9 22 07 -1 -22
10-14 13 12 A 0 0
15-19 8 17 1 8
20-24 2 22 2 4
N= 50 å fd = -20

2. Take the mid point of any class interval as assumed mean (A). Here we have taken the mid point of
class interval of 10-14, which is 12, as assumed mean.
3. Set third column as d. That gives deviation from assumed mean, which is calculated by using
formula:
(mid point of ith class interval - assumed mean)/ width of class interval.
Here class interval is 5.
 202 Textbook of Computer Applications and Biostatistics

4. Finally set forth column fd and calculate å fd.

5. Now put the values in formula for calculation of mean for grouped data
Data:
A=Assumed mean = 12
N = Summation of all frequencies = 50
åf d = -20
i = width of the class interval = 5
Formula:

Mean ( X ) = A +
å fd ´ i
N

- 20
Mean ( X ) = 12 + ´ 5 = 10
50

Hence mean of the number of accidents per week is 10.

Example 12.12
The following are the weights in kg of 60 final year pharmacy students. Calculate mean
weight.
Table 12.11 Weights of 60 final year students

Weight (kg) 45-49 50-54 55-59 60-64 65-69 70-74 75-79 80-54
Frequency 3 10 12 15 6 7 5 2

Solution
1. Construct frequency distribution table as per given in table 12.12.
Table 12.12 Calculations for mean weight of final year students

Weight Frequency Mid value of Deviation from

( kg) class interval (m) assumed mean (d) fd
n=60
45-49 03 47 -3 -9
50-54 10 52 -2 -20
55-59 12 57 -1 -12
60-64 15 62 A 0 0
65-69 06 67 1 6
70-74 07 72 2 14
75-79 05 77 3 15
80-84 02 82 4 8
N= 60 å fd = 2
 Measures of Central Tendency 203

2. Take the mid point of any class interval as assumed mean (A). Here we have taken the mid point of
class interval of 60-64, which is 62, as assumed mean.
3. Set third column as d. That gives deviation from assumed mean, which is calculated by using
formula:
(mid point of ith class interval - assumed mean)/ width of class interval.
Here class interval is 5.
4. Finally set forth column fd and calculate å fd.
5. Now put the values in formula for calculation of mean for grouped data
Data:
A=Assumed mean = 62
N = Summation of all frequencies = 60
åf d = 2
i = width of the class interval = 5
Formula:

Mean ( X ) = A +
åf d ´i
N
2
Mean ( X ) = 62 + ´ 5 = 62 .16
60
Hence mean weight of final year pharmacy students is 62.16 kg.

3. Mean for Metric Discrete data (frequency data)

When observations are grouped as a frequency distribution, then mean is calculated by using
following formula:

Mean (X) = åf Xi i

N ...7
Where
N = å fi
fi = frequency with which variable Xi occurs

Example 12.13
Following data gives number of times that inhaler used in past 24h by 53 children with
asthma. Find mean number of times inhaler used by children with asthma.
Table 12.13 Number of times inhaler used by children

No. of times inhaler used 0 1 2 3 4 5 6 7

Number of children 6 16 12 8 5 3 2 1
 204 Textbook of Computer Applications and Biostatistics

Solution:
1. Construct frequency distribution table as per given in table 12.14.
2. Set third column as fiXi.
3. Determine summation of fi and fiXi.
4. Now put the values in formula for calculation of mean for grouped discrete data

Table 12.14 Calculation of mean for metric discrete data

Number of times inhaler used in past 24 h Number of children fiXi

(Xi) (fi)
0 06 0
1 16 16
2 12 24
3 08 24
4 05 20
5 03 15
6 02 12
7 01 7
N= åfi = 53 å fiXi = 118
Applying formula 7, the mean is

Mean (X) =
åf X i i
=
118
= 2.23
N 53

Hence mean number of times inhaler used by children with asthma in past 24 h was 2.23.

Example 12.14
Calculate mean number of living children per woman from the following table.
Table 12.15 Number of living children per woman

No. of living children 0 1 2 3 4 5

Number of women 42 49 57 40 31 22

Solution:
1. Construct frequency distribution table as per given in table 12.16.
2. Set third column as fiXi.
3. Determine summation of fi and fiXi.
4. Now put the values in formula for calculation of mean for grouped discrete data.
 Measures of Central Tendency 205

Table 12.16 Calculation of mean for metric discrete data

Number of living children Number of Women fiXi

(Xi) (fi)
0 42 0
1 49 49
2 57 114
3 40 120
4 31 124
5 22 110
N= åfi = 241 å fiXi = 517

Applying formula 7, the mean is

Mean (X) =
åf Xi i
=
517
= 2.145
N 241

Hence mean number of living children per woman was 2.145.

Choosing the most appropriate measure

The most appropriate measure of location for given set of data is given below in the table
12.10. The main thing to remember is that the mean cannot be used with ordinal data (because they are
not real numbers), and that the median can be used for both ordinal and metric data (particularly when
the latter is skewed).
Table 12.17 Aguide to choosing an appropriate measure of location

Summary measure of location

Type of variable Mode Median Mean

Nominal yes no no
Ordinal yes yes no
Metric continuous yes yes, if skewed yes
Metric discrete yes yes, if skewed yes

4. Percentiles
The measures of central values discussed so far are averages. They locate the centre or mid
point of a distribution. It may also be of interest to locate other points in the range like percentiles.
They are values of a variable which divide the total observations by an imaginary line into two parts,
expressed in percentages such as 10% and 90% or 25% and 75%, etc. In all, there are 99 percentiles.
Percentiles are values in a series of observations arranged in ascending order of magnitude which
divide the distribution into 100 equal parts. Thus, the median is 50th percentile. The 50th percentile
 206 Textbook of Computer Applications and Biostatistics

will have 50% observations on either side. Accordingly, 10th percentile will have 10% observations
to the left and 90% to the right.
Quartiles are three different points located on the entire range of a variable- Q1, Q2 and Q3.
Q1 or lower quartile will have 25% observations falling on its left and 75% on its right: Q2 or median
will have 50% observations on either side and Q3 or upper quartile will have 75 % observations on its
left and 25% on its right.
Quintiles are four in number and divide the distribution into 5 equal parts. So 20th
percentile or first quintile will have 20% observations falling to its left and 80% to its right.
Deciles are nine in number and divide the distribution into 10 equal parts, first decile or 10th
percentile will divide the distribution into 10% and 90% while 9th decile will divide into 90% and
10% .

Calculating a percentile value

Percentile value is calculated by using formula
P
Pth Percentile = (N + 1)th value
100 ...8
Where
P = percentile
N = number of values

Example 12.15
Following table records the percentage mortality in 26 ICUs. Calculate the 25th and 75th
percentiles for the ICU percent mortality values.
Table 12.18 Percent mortality in 26 ICUs

ICU 1 2 3 4 5 6 7 8 9 10 11 12 13
% mortality 15.2 31.3 14.9 16.3 19.3 18.2 20.2 12.8 14.7 29.4 21.1 20.4 13.6
ICU 14 15 16 17 18 19 20 21 22 23 24 25 26
% mortality 22.4 14 14.3 22.8 26.7 18.9 13.7 17.7 27.2 19.3 16.1 13.5 11.2

Solution:
1. First, arrange values in ascending order for 26 ICUs as shown below

ICU 1 2 3 4 5 6 7 8 9 10 11 12 13
% mortality 11.2 12.8 13.5 13.6 13.7 14 14.3 14.7 14.9 15.2 16.3 16.1 17.7
ICU 14 15 16 17 18 19 20 21 22 23 24 25 26
% mortality 18.2 18.9 19.3 19.3 20.2 20.4 21.1 22.4 22.8 26.7 27.2 29.4 31.3
th
2. Now, the 25 percentile can be calculated by using formula
 Measures of Central Tendency 207

P
Pth Percentile = (N + 1)th value
100
P= 25th percentile = 25
N= no. of ICU = 26
25
25 th Percentile = (26 + 1) = 0.25 ´ 27th value = 6.75th value
100
th th
The 6 value is 14 % while the 7 value is 14.3%,
A difference is of 0.3,
th
The 25 percentile is 14% + 0.75 of 0.3,
14% + 0.75 x 0.3% = 14% + 0.225 =14.225
th
So, the 25 percentile for the ICU percent mortality is 14.225.

th
3. Now, the 75 percentile can also be calculated using same formula, P = 75 and N = 26.
75
75th Percentile = (26 + 1) = 0.75 ´ 27th value = 20.25th value
100
th st
The 20 value is 21.1 % while the 21 value is 22.4%, a differences of 1.3,
The 75th percentile is 21.1% + 0.25 of 1.3 = 21.1% + 0.25 x 1.3% = 21.1% + 0.325 =21.425
So, the 75th percentile for the ICU percent mortality is 21.425.

Summary measures of spread

There are three main measures of spread or dispersion in common use. Here also, the type of
data influences the choice of an appropriate measure.

1. The Range
The range is the distance from the smallest value to the largest. The range is not affected by
skewness, but is sensitive to the addition or removal of an outlier value.
Range = Lowest value to Highest value. ...9

Example 12.16
Weight of six paracetamol tablets in mg are given below. Find weight range for
paracetamol tablet.
Data: 625, 612, 615, 632, 628, 618
Solution:
1.Arrange data in ascending order
612, 615, 618, 625, 628, 632
2. Lowest value = 612 mg
3. Highest value = 632 mg
 208 Textbook of Computer Applications and Biostatistics

Range is 612 to 632 mg.

2. The interquartile range (iqr)

One solution to the problem of the sensitivity of the range to extreme value (outliers) is to
chop a quarter (25 per cent) of the values off both ends of the distribution, which removes any
troublesome outliers, and then measure the range of the remaining values. This distance is called the
interquartile range, or iqr. The interquartile range is not affected either by outliers or skewness, but it
does not use all of the information in the data set since it ignores the bottom and top quarter of values.

Calculating interquartile range

To calculate the interquartile range, first we need to determine two values:
1. The value which cuts off the bottom 25th percentile of values; this is known as the first quartile
and denoted as Q1.
2. The value which cuts off the top 75th percentile of values, known as the third quartile and denoted
as Q3.
The interquartile range is then written as (Q1 to Q3).

Example 12.17
Calculate the iqr for the ICU percentage mortality values in table 12.19 .
Table 12.19 Percent mortality in 26 ICUs

ICU 1 2 3 4 5 6 7 8 9 10 11 12 13
% mortality 15.2 31.3 14.9 16.3 19.3 18.2 20.2 12.8 14.7 29.4 21.1 20.4 13.6
ICU 14 15 16 17 18 19 20 21 22 23 24 25 26
% mortality 22.4 14 14.3 22.8 26.7 18.9 13.7 17.7 27.2 19.3 16.1 13.5 11.2

Solution
1.Arrange the data in ascending order
2. Calculate 25th percentile as a Q1
3. Calculate 75th percentile as a Q3
4. We have already calculated the 25th and 75th percentiles before in example 12.15
Q1 = 14.225%,
Q3 = 21.425%
Therefore interquartile range = (14.225 to 21.425) %.

Estimating the median and interquartile range from Ogive

We can also estimate interquartile range from ogive. If we draw horizontal lines from the
values 25 per cent, 50 per cent and 75 per cent on the y axis, to the ogive, and then down to the x axis,
the points of intersection on the x axis approximately gives values for Q1, Q2 (the median), and Q3.
 Measures of Central Tendency 209

Example 12.18
Estimate iqr for the data of weight in kg of 60 final year pharmacy students given in Table
10.6.
Table 12.20 Weights of 60 final year students

Weight (kg) 45-49 50-54 55-59 60-64 65-69 70-74 75-79 80-54
Frequency 3 10 12 15 6 7 5 2

Solution
We have already drawn ogive of this example.
Ogive is drawn by plotting % cumulative frequency on Y axis against midpoint of class
interval on X axis.
Now, draw perpendicular to ogive at 25th percentile (Q1), 50th percentile (Q2) and 75th
percentile (Q3) as shown in figure below.
% Relative cumulative frequency

100

Q3= 66
75

Q2= 58.5
50

Q1= 53
25

0
40 50 60 70 80 90
Mid point of class interval

Figure 12.1 % cumulative frequency curve of weight to estimate median and iqr

The weight values corresponding to these perpendiculars are taken as Q1, Q2 and Q3 and here
they are Q1=53, Q2 = 58.5 and Q3 =66
Interquartile range (IQR)= Q1 to Q3
Interquartile range (IQR)= 53 to 66
The interquartile range for the data of weights of final year students is 53 to 66 kg.

The Boxplot
Now that we have discussed the median and interquartile range, we can now better
understand the boxplot which we had seen in chapter 11. Boxplots provide a graphical summary of
the three quartile values, the minimum and maximum values, and any outliers. They are usually
plotted with value on the vertical axis. Like the pie chart, the boxplot can only represent one variable
at a time, but a number of boxplots can be set alongside each other.
 210 Textbook of Computer Applications and Biostatistics

Let’s understand the boxplot,

1. The bottom end of the lower ‘whisker’(the line sticking out of the bottom of the box), corresponds
to the minimum value.
2. The bottom of the box is the 1st quartile value, Q1.
3. The line across the inside of the box, is the median, Q2. The median line will always not be at the
centre. The more asymmetric (skewed) the distributional shape, the median line will be further away
from the middle of the box. If it is closer to the top of the box, it indicates negative skew while if it is
closer to the bottom of the box, it indicates positive skew.
4. The top of the box is the third quartile Q3.
5. The top end of the upper whisker is the ‘maximum’. This is the maximum value that can be
considered still to be part of the general mass of the data.
6. There is one outlier. This is, the actual maximum value in the data.

120
Percentage of label claim

110

100

90

80
n = 50
Figure 12.1 Boxplot

Example 12.21
Sketch the box plot for the percentage mortality in ICUs shown in table 12.10. We have
already calculated Q1 and Q3 for same.

Solution:
Calculation of Q2, Median
The 50th percentile can be calculated using formula

P
Pth Percentile = (N + 1)th value
100
th
P = 50 percentile = 50; N = 26
 Measures of Central Tendency 211

50
50th Percentile = (26 + 1) = 0.5 ´ 27th value = 13.5th value
100
The 13th value is 17.7 % while the 14th value is 18.2%. A differences of 0.5.
The 50th percentile is 17.7% + 0.5 of 0.5 = 17.7% + 0.5 x 0.5% = 17.7% +0.25 =17.95
So, the 25th percentile for the ICU percent mortality is 14.225.
So, the 50th percentile for the ICU percent mortality is 17.95.
So, the 75th percentile for the ICU percent mortality is 21.425.
Now, the box plot can be sketched as under

Maximum

75th percentile
50th percentile
25th percentile
Minimum

Figure 12.2. Box plot for percentage mortality in ICUs

3. Standard Deviation and Variance

The limitation of the interquartile range as a summary measure of spread is that, it is not
using all the information in the data, since it omits the top and bottom quarter of values. An alternative
approach we can use is to measure the mean (average) distance of all the data values from the overall
mean of all of the values. The smaller this mean distance is, the narrower the spread of values will be,
and vice versa. This approach is known as the standard deviation, or s.d.
Standard deviation can be calculated by using following steps:
1. Subtract the mean of the sample from each of the n sample values in the sample, to give the
difference values.
2. Square each of these differences.
3. Add these squared values together (called the sum of squares).
4. Divide the sum of squares by (N – 1); i.e. divide by 1 less than the sample size.
5. Take the square root. This is the standard deviation
Variance is the square of standard deviation and is used instead of standard deviation.
Example 12.23
A microbiologist found 8, 10, 19, 12, 6 and 5 E. coli in six culture. Calculate standard
 212 Textbook of Computer Applications and Biostatistics

2
deviation, s and variance, s .
Solution
1. Let us calculate the mean
8 + 10 + 19 + 12 + 6 + 5
Mean, X = = 10
6
2. Now, construct the table as shown below:
x (x - `x ) (x - `x )2
08 -2 4
10 0 0
19 9 81
12 2 04
06 -4 16
05 -5 25
å(x - `x )2=130
3. Let us use å(x - `x ) value in the formula
2

Formula
å(X - X) 2
Standard deviation (s) = ...10
N -1
130
Standard deviation (s) = = 26 = 5.1
6 -1
2
4. Thus, standard deviation, s was found to be 5.1 while variance, s was found to be 26.
However, the calculations using this formula can be tedious when the numbers and mean are
in decimals. So, the computing formula used is as follows
Formula
(å X)2
å X2 -
Standard deviation (s) = N ...11
N -1
Let us calculate standard deviation using this formula
2
x x
08 64
10 100
19 361
12 144
06 36
05 25
åx = 60 åx =730
2
 Measures of Central Tendency 213

Hence: N = 6, åX = 60, åX =730

2

(60) 2
730 -
Standard deviation (s) = 10 = 5.1
6 -1
2
The, standard deviation, s by this method is also 5.1 and variance, s is 26.

1. Standard deviation (s) in ungrouped data

For calculation of standard deviation in ungrouped data following formula is used, where
assumed mean method is used so that squaring of large values can be avoided.

(å X )2
å X2 - ... 12
Standard deviation (s) = N
N -1
Where,
N = number of observations
X = deviation from assumed mean = x-A.
A=Assumed mean
x = given observations

Example 12.23
Find standard deviation of incubation period of smallpox in 9 patients where it was found to
be 14, 13, 11, 15, 10, 7, 9, 12, and 10.
Solution
1. Construct calculation table as per given in table no 12.21.
2. Set any assumed mean from set of observations (x).
3. Set second column as deviation from assumed mean i.e. X=x-A.
2
4. Set third column as square of deviation i.e X
2
5. Determine summation of deviation from assumed mean (X) and square of deviation (X ).
6. Finally put the values in equation 10 so as to get standard deviation for ungrouped data.
 214 Textbook of Computer Applications and Biostatistics

Table 12.21 Calculation of standard deviation for ungrouped data

Incubation period of smallpox Deviation from assumed mean Square of deviation

(x) (X= x - 11) X2
14 3 9
13 2 4
11 A 0 0
15 4 16
10 -1 1
7 -4 16
9 -2 4
12 1 1
10 -1 1
Data åX = 2 åX = 52
2

N=9
Summation of deviation from assumed mean, X =2
2
Summation of square of deviation, X = 52
Formula
(å X)2
å X2 -
Standard deviation (s) = N
N -1

22
52 -
9 = 51.56
Standard deviation (s) = = 6 .443 = 2.54
9 -1 8

Therefore, standard deviation of incubation period of smallpox in 9 patients was 2.54.

Example 12.24
The following data gives weight of 10 paracetamol tablets in mg. Find standard deviation.
625 617 633 630 620 631 618 620 619 632
Solution
1. Construct calculation table as per given in table no 12.22.
2. Set any assumed mean from set of observations (x).
3. Set second column as deviation from assumed mean i.e. X=x-A.
2
4. Set third column as square of deviation i.e X
2
5. Determine summation of deviation from assumed mean (X) and square of deviation (X ).
6. Finally put the values in equation 10 so as to get standard deviation for ungrouped data.
 Measures of Central Tendency 215

Table 12.22 Calculation of standard deviation for ungrouped data

Weight of Paracetamol tablets Deviation from assumed mean Square of deviation
(x) (X= x - 625) X2
625 A 0 0
617 -8 64
633 8 64
630 5 25
620 -5 25
631 6 36
618 7 49
620 -5 25
619 -6 36
632 7 49
åX= -5 åX = 373
2

Data
N = 10
Summation of deviation from assumed mean, X = -5
2
Summation of square of deviation, X = 373
Formula
(å X) 2
å X2 -
Standard deviation (s) = N
N -1

- 52
373 -
Standard deviation (s) = 10 = 6.416
10 - 1

Therefore, standard deviation of weight of 10 paracetamol tablets was 6.416.

2. Standard deviation (s) in grouped data

Standard deviation in grouped data is calculated by using formula:

( å fd ) 2
å fd 2 -
Standard deviation (s) = N ´i ...13
N -1
Where,
f = frequency
N = summation of frequency, f = Sf
 216 Textbook of Computer Applications and Biostatistics

d = (m.p. - A)/ i
m.p. = mid point
A=Assumed mean
i = class interval

Example 12.25
Calculate SD of following data
Table 12.23 IQ of 50 students

IQ 0-20 20-40 40-60 60-80 80-100 100-120 120-140 140-160

Number of students 3 4 3 4 13 12 8 3

Answer:
1. Construct calculation table as per given in table no 12.24.
2. Construct third column as mid value of class interval.
3. Set any assumed mean (A) from set of mid points (m.p.).
4. Set forth column as deviation from assumed mean i.e.d= (m.p. -A)/ i.
2
5. Determine fd and fd .
2
6. Determine summation of f, fd and fd .
7. Finally put the values in equation 11 so as to get standard deviation for grouped data.

Table 12.24 Calculation of SD in grouped data

No of Students Mid value of d
IQ d
2
fd f d2
(f) class interval (m.p.) (mp - A)/i

0-20 3 10 -4 16 -12 48
20-40 4 30 -3 9 -12 36
40-60 3 50 -2 4 -6 12
60-80 4 70 -1 1 -4 4
80-100 13 90 A 0 0 0 0
100-120 12 110 1 1 12 12
120-140 8 130 2 4 16 32
140-160 3 150 3 9 9 9
N=Sf = 50 Sfd=3 Sfd2= 171
Data:
Class interval, i = 20;
Assumed mean, A= 90;
N=Sf = 50;
Sfd=3;
 Measures of Central Tendency 217

Sfd = 171
2

Formula:
( å fd ) 2
å fd 2 -
Standard deviation (s) = N ´i
N -1

Solution: Now, putting the values in above formula, we get

( 3) 2
171 -
Standard deviation (s) = 50 ´ 20 = 3.48 ´ 20 = 37 .2
50 - 1
Therefore standard deviation of given IQ data is 37.2.

Example 12.26
Calculate SD of following data of intelligence quotient (IQ) of 50 students.
Table 12.25 Given data
Classes 0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90
Frequency 5 10 15 20 25 12 8 4 1

Answer:
1. Construct calculation table as per given in table no 12.26.
2. Construct third column as mid value of class interval.
3. Set any assumed mean (A) from set of mid points (m.p.).
4. Set forth column as deviation from assumed mean i.e. d= (m.p. -A)/ i.
2
5. Determine fd and fd .
6. Determine summation of f, fd and fd2.
7. Finally put the values in equation 11 so as to get standard deviation for grouped data.
Table 12.26 Calculation of SD in grouped data
Classes Frequency Mid value of d 2
d fd f d2
(f) class interval (m.p.) (mp - A)/i
0-10 5 5 -4 16 -20 80
10-20 10 15 -3 9 -30 90
20-30 15 25 -2 4 -30 60
30-40 20 35 -1 1 -20 20
40-50 25 45 A 0 0 0 0
50-60 12 55 1 1 12 12
60-70 8 65 2 4 16 32
70-80 4 75 3 9 12 36
80-90 1 85 4 16 4 16
N=Sf = 100 Sfd=-56 Sfd = 346
2
 218 Textbook of Computer Applications and Biostatistics

Data:
Class interval, i = 10;
Assumed mean, A= 45;
N=Sf = 100;
Sfd=-56;
Sfd = 346
2

Formula:
( å fd ) 2
å fd 2 -
Standard deviation (s) = N ´i
N -1
Solution:
Now, putting the values in above formula, we get

( - 56 ) 2
346 -
Standard deviation (s) = 100 ´ 10 = 17.8
100 - 1

Therefore standard deviation of given data is 17.8.

3. Standard deviation in metric discrete data

Standard deviation for metric discrete data is given by the formula:

(å f i X i ) 2
å fiXi -
2

Standard deviation (s) = N ...14

N -1

Where,
fi = frequency with which variable Xi occurs
N = åfi

Example 12.27
Calculate standard deviation of Example 12.13

No. of times inhaler used 0 1 2 3 4 5 6 7

Number of children 6 16 12 8 5 3 2 1
 Measures of Central Tendency 219

Solution
1. Construct the frequency table as shown below:
Number of times inhaler used in past 24 h Number of children fiXi Xi2 fiXi2
(Xi) (fi)
0 06 0 0 0
1 16 16 1 16
2 12 24 4 48
3 08 24 9 72
4 05 20 16 80
5 03 15 25 75
6 02 12 36 72
7 01 7 49 49
N= åfi = 53 å fiXi = 118 åfiXi = 412
2

Data
N= åfi = 53
å fiXi = 118
åfiXi2 = 412
Formula
(å fiX i )2
å fiX i -
2

Standard deviation (s) = N

N -1

2. Now put the values in the formula

(118 ) 2
412 -
53 = 412 - 262.71
Standard deviation (s) = = 1.72
53 - 1 53 - 1

So, the standard deviation was found to be 1.72.

Measurement related to sample standard deviation

The variability of data may often be better described as a relative variation rather than as an
absolute variation (i.e., the standard deviation). This can be accomplished by calculating the
coefficient of variation (CV) that is the ratio of the standard deviation to the mean.
SD ...15
Coefficient of variation (CV) =
mean
The CV is usually expressed as a percentage (relative standard deviation or RSD) and can be
useful in many instances because it places variability in perspective to the distribution center.
Relative Standard Deviation = CV x 100 ...16
 220 Textbook of Computer Applications and Biostatistics

Example 12.28
In two series of adults aged 21 years and children 3 months old, following values were
obtained for the height. Find which series shows greater variation?
Data Persons Mean height SD
Adults 160 cm 10 cm
Children 60 cm 5 cm
Formula:
CV = SD/mean
RSD = CV x 100
Solution:
Put the values in equation
Adults
CV of adults = 10/160 = 0.0625
RSD = 0.0625 x 100 = 6.25 %
Children
CV of children = 5/60 = 0.0833
RSD = 0.0833 x 100 = 8.33 %
Thus, heights in children show greater variation than adult.

Example 12.29
Chest circumference in cm of 10 malnourished children and normal children aged one year,
are given below. Find which series shows greater variation?
Data Children Mean height SD
Malnourished 48.5 4.53
Normal 34.8 4.57
Formula:
CV = SD/mean
RSD = CV x 100
Solution:
Put the values in equation
Malnourished
CV = 4.53/48.5 = 0.09
RSD = 0.093 x 100 = 9.3 %
Normal
CV = 4.57/34.8 = 0.131
RSD = 0.131 x 100 = 13.1 %
Thus, Normal children show greater variation than malnourished children.
 Measures of Central Tendency 221

Choosing an appropriate measure of spread

Table 12.27 Choosing an appropriate measure of spread
Summary measure of spread
Type of variable
Range Interquartile range Standard deviation
Nominal no no no
Ordinal yes yes no
Metric yes yes, if skewed yes

Use of Excel in Measures of Central Tendency

We can easily calculate both measures of central tendency i.e measures of location and
measures of dispersion in excel where we need not have to worry about any formulas. Excel will
readily calculate these values for us. So, lets see how we can use excel for calculating measures of
central tendency.

Installing theAnalysis ToolPak in Excel

Excel includes a number of add-in tools to assist with a number of data handling, reporting
and analysis functions. So, first we will install this analysis toolpak as given below:
1. Open New workbook from MS-Excel and click to Tools menu from Menu bar.
2. Instantly, it will display pull-down menus.
3. Select onAdd-Ins option.
4. In the Manage box, selectAdd-ins-Analysis ToolPak as shown in figure.
6. Then, click on OK button.
This unpacks the chosen Excel Add-in tools and makes them available for use. If the Add-in
list is empty, there may have been a limited installation of MS Office. In this case, the MS Office CDs
will be needed to install theseAdd-in tools.

Figure 12.3 Window of Add-Ins

 222 Textbook of Computer Applications and Biostatistics

Use of Excel in Descriptive Statistics

The Data Analysis ToolPak has a Descriptive Statistics tool that provides us with an easy
way to calculate summary statistics for a set of sample data. Summary statistics includes Mean,
Standard Error, Median, Mode, Standard Deviation, Variance, Kurtosis, Skewness, Range,
Minimum, Maximum, Sum, and Count. This tool eliminates the need to type individual functions to
find each of these results. Excel includes elaborate and customizable toolbars. We can follow the
below steps:
Step 1. Open New Excel file. In sheet 1 put labels in first row (i.e A1, B1) and enter the data below
these labels.
Step 2. Select the Tools menu from menu bar. Then, it will display pull-down menus.
Step 3. Click on the data analysis option. Instantly, dialog box will appear.
Step 4. Choose Descriptive Statistics from Analysis Tools list of Data Analysis dialog box. Lastly,
click on OK button. This will display the Descriptive Statistics dialog box on the screen.

Figure 12.4 Window of Data Analysis

Step 4. When the dialog box appears; Enter range of data or select range (along with labels) in the
Input Range box. Tick mark on Labels in first row and summary statistics of Output options, as
shown below:
Note: Ensure that labels are given in first row.
Step 5. Ignore remaining settings and click on Ok button.

Figure 12.5 Window of Descriptive statistics

 Measures of Central Tendency 223

Step 6. We will get result of descriptive statistics as shown below.

Column1
Mean
Standard Error
Median
Mode
Standard Deviation
Sample Variance
Kurtosis
Skewness
Range
Minimum
Maximum
Sum
Count

Summary
The mean, median and mode, are common measures of location which represent either a
typical or representative score and/or a value about which the data tend to center.
Mode
The mode (denoted by M) represents the most frequently occurring score. When more than
two scores occur with the greatest frequency, the data set is said to be multimodal.
Mode for grouped data
æ f1 - f 0 ö
Mode = l1 + çç ÷÷ ´ i
è 2f1 - f 0 - f 2 ø
Median
th
The median of a set of scores represents the middle value (50 percentile) when the scores are
arranged as an array in order of increasing (or decreasing) magnitude.
Median for ungrouped data
Odd number
æ n +1ö
Median = Size or value of ç ÷ th observation
è 2 ø
n æn ö
Even number th + ç + 1 ÷ th
Median = value of
2 è2 ø observation
2

Median for grouped data

(n/2) - c.f.
Median = l1 + ´ i
f

Median for Metric Discrete data

Median = value of (n/2)th observation
 224 Textbook of Computer Applications and Biostatistics

Mean
Mean represents the most appropriate measure of central tendency for continuous-type data.
It is obtained by adding all of the scores and dividing this sum by the number of scores.
Mean for ungrouped data
Sum of all observations (å X)
Mean (X) =
Total number of observations (N)

Mean for grouped data

Mean ( X ) = A +
å fd ´ i
N

Mean for Metric Discrete data

Mean (X) = åf Xi i

N
Percentile
Percentiles are values in a series of observations arranged in ascending order of magnitude
which divide the distribution into 100 equal parts.
P
Pth Percentile = (N + 1)th value
100
The range, interquartile range, standard deviation are common measures of spread and
represents the extent of spread of data
Range
The range is the distance from the smallest value to the largest.
Range = Lowest value to Highest value.
The interquartile range
The interquartile range is Q1 to Q3
Q1 is the value which cuts off the bottom 25th percent of values and is known as the first
quartile (25th percentile),while Q3 cut off top 25th percent of values and is known as third quartile (75th
percentile).
Standard Deviation (SD)
The standard deviation measures the variation of scores about the mean (average) score, and
can be defined as the “root mean squared deviation.” Variance is square of standard deviation.
Standard deviation of ungrouped data
(å X )2
å X2 -
Standard deviation (s) = N
N -1
Standard deviation of grouped data
( å fd ) 2
å fd 2 -
Standard deviation (s) = N ´i
N -1
 Measures of Central Tendency 225

Standard deviation for metric discrete data

(å f i X i ) 2
å f i X i2 -
Standard deviation (s) = N
N -1

Measurement related to sample standard deviation

SD
Coefficient of variation (CV) =
mean
Relative Standard Deviation = CV x 100

Choosing an appropriate measure of central tendency

Type of Summary measures of location Summary measures of spread

variable Mode Median Mean Range Iqr SD

Nominal yes no no no no no
Ordinal yes yes no yes yes no
Metric yes yes, if skewed yes yes yes, if skewed yes

Multiple choice questions

1. The range of a sample gives an indication of the________.
a. way in which the values cluster about a particular point
b. number of observations bearing the same value
c. maximum variation in the sample
d. degree to which the mean value differs from its expected value.
2. The observation which occurs most frequently in a sample is the__________.
a. median b. mean deviation c. standard deviation d. mode
3. What is the median of the sample 5, 5, 11, 9, 8, 5, 8 ?
a. 5 b. 6 c. 8 d. 9
4. What is the median of the following set of scores? 18, 6, 12, 10, 14 ?
a. 10 b. 14 c. 18 d. 12
5. All of the following are measures of central location except________.
a. range b. arithmetic mean c. median d. mode
6. The measure of central location that has half of the observations below it and half of the
observations above it is the__________.
a. arithmetic mean b. geometric mean c. median d. mode
7. The most commonly used measure of central location is the________.
a. mean b. range c. median d. mode
 226 Textbook of Computer Applications and Biostatistics

8.All of the following are measures of dispersion except______.

a. interquartile range b. percentile c. variance d. standard deviation
9. The ______ is the value we calculate when we want the arithmetic average.
a. Mean b. Median c. Mode d. All of the above
10. The _______ is often the preferred measure of central tendency if the data are severely skewed.
a. Mean b. Median c. Mode d. Range
11. Which of the following is the formula for range?
a. H + L b. L x H c. L - H d. H – L
12. Why are variance and standard deviation the most popular measures of variability?
a. They are the most stable and are foundations for more advanced statistical analysis
b. They are the most simple to calculate with large data sets
c. They provide nominally scaled data
d. None of the above
13. Which of the following is NOT a measure of variability?
a. Median b. Variance c. Standard deviation d. Range
14. Which range characterizes the interquartile range?
a. From 5th percentile to 95th percentile
b. From 10th percentile to 90th percentile
c. From 25th percentile to 75th percentile
d. From 1 standard deviation below the mean to 1 standard deviation above the mean
15. The measure of dispersion most commonly used in conjunction with the arithmetic mean is the:
a. interquartile range b. range c. standard deviation d. variance

Exercise
1. Amount of 10 tablets of Atenolol were determined for its quality control. Calculate the mode of the
observed amounts.
Data: 20, 22, 18, 24, 24, 24, 18, 19, 18, 18.

2. Calculate mode for following data

Data: 8, 5, 10, 10, 9, 6, 8, 8, 10, 8, 6, 8, 10, 10, 7.

3.Anovel antidiabetic drug is developed by researcher. The reduction in glucose in 10 patients after 2
hours administration of drug is recorded. Calculate the mean reduction in blood glucose levels in the
10 patients.
Data: 20, 15, 22, 18, 25, 17, 23, 27, 19, 21.

4. The weights of 20 tablets, removed from a batch for quality control purpose are given below.
Calculate the mean and median values of the tablet weights.
 Measures of Central Tendency 227

Data: 250, 252, 255, 245, 251, 260, 258, 256, 248, 275,
268, 240, 257, 262, 270, 280, 266, 279, 258, 265.

5. Serum bilirubin levels are measured for 10 cirrhosis patients as given below. Determine the range.
Data: 1, 8, 5, 7, 10, 2, 15, 12, 3, 9.

6. Drug content of injection were determined using ultraviolet spectroscopy. Results are reported in
mg/ml. Calculate mean and standard deviation of the drug content of injection.
Data: 50.3, 48.1, 48.9, 51.5, 52.5, 50.6, 49.5, 53, 47.5, 50.

7. Time taken for dissolution of 50% of the original mass of drug from 10 tablets are summarized
below. Determine the mean and standard deviation of the time required for the release of 50% of the
original mass of drug.
t50% (h): 24.7, 20.1, 25.3, 22, 22.5, 28.6, 21.6, 20.4, 23.5, 25.1.

8. Calculate mean and standard deviation of 15 readings in the preliminary study of urinary lead
concentrations, micro mol/24h.
Data: 0.1, 0.4, 0.6, 0.8, 1.1, 1.2, 1.3, 1.5, 1.7, 1.9, 1.9, 2.0, 2.2, 2.6, 3.2

9. Results of new therapeutic agent with 50 mg/tablet administered to six healthy volunteers are listed
below. Report median and mean.
Volunteer number 1 2 3 4 5 6
Cmax mg/l 60 71 111 46 81 96

10. Calculate median and mean of following assay data of 30 tetracycline capsules.
Assay result 245 246 247 248 249 250 251 252 253 254
Frequency 1 1 2 3 4 7 5 3 2 2

11. Calculate mode, median, mean and standard deviation for the following data.
Age in years 0-10 10-20 20-30 30-40 40-50 50-60 60-70
No of persons 4 6 10 20 10 6 4

12. In addition to studying the lead concentration in the urine of 140 children, the paediatrician asked
how often each of them had been examined by a doctor during the year 2010. Calculate mean, median
and standard deviation.
No of Visits to doctor 0 1 2 3 4 5 6 7
No of Children 2 8 27 45 38 15 4 1
 228 Textbook of Computer Applications and Biostatistics

13. Age at Death (in days) of 78 cases of sudden infant death syndrome (SIDS) are given in following
table.
Age in days 1-30 31-60 61-90 91-120 120-150 151- 180 181-210 211-240 241-270
No. of deaths 6 13 23 18 7 5 3 2 1

Determine Mean, median and standard deviation.

14. Ages of Patients Diagnosed with Multiple Sclerosis are given in following table.
Age in days 20-29 30-39 40-49 50-59 60-69 70-79 80-89
No. of deaths 4 44 124 124 48 25 4

Determine Mean and standard deviation.

15. The elimination half-lives of two synthetic steroids have been determined using two groups, each
containing 15 volunteers. The results are shown below, with the values ranked from lowest to highest
for each steroid. Find the median, range and quartiles.
Steroid 1
Rank 1 2 3 4 5 6 7 8 9 10 11 12 13 14
Half life (h) 3.9 4.0 5.4 6.4 6.5 7.2 7.8 8.6 9.2 9.3 10 10.6 11.1 15.8
Steroid 2
Rank 1 2 3 4 5 6 7 8 9 10 11 12 13 14
Half life (h) 4.4 4.5 5.5 5.8 5.9 6.1 6.6 7.2 7.2 7.3 7.8 8.5 8.6 9.1

16. From the 140 children whose urinary concentration of lead were investigated, 40 were chosen
who were aged at least 1 year but under 5 years. The following concentrations of copper were found.
Find the median, range and quartiles.
Data 0.70, 0.45, 0.72, 0.30, 1.16, 0.69, 0.83, 0.74, 1.24, 0.77,
0.65, 0.76, 0.42, 0.94, 0.36, 0.98, 0.64, 0.90, 0.63, 0.55,
0.78, 0.10, 0.52, 0.42, 0.58, 0.62, 1.12, 0.86, 0.74, 1.04,
0.65, 0.66, 0.81, 0.48, 0.85, 0.75, 0.73, 0.50, 0.34, 0.88

17. Erythromycin contents of 10 tablets from an Alpha and a Bravo tabletting machine was
determined and given below. Determine coefficient of variation and relative standard deviation. Give
your comment on it.
Mean SD
Alpha machine 248.7 8.72
Bravo machine 251.1 3.78
 Measures of Central Tendency 229

18. Fifteen patients were provided with their drugs in a child-proof container of a design that they had
not previously experienced. A note was taken of the time it took each patient to get the container open
for the first time. Determine median and quartile. The results are shown below
Times taken to open a child-proof container (s)
2.2 3 3.1 3.2 3.4 3.9 4 4.1 4.2 4.5 5.1 10.7 12.2 17.9 24.8

19. The percentage of ideal body weight was determined for 18 randomly selected insulin-dependent
diabetics. The outcomes (%) are
107 119 99 114 120 104 124 88 114
116 101 121 152 125 100 114 95 117
Determine mean, CV and RSD.

20. Calculate all the following measures of central tendency for data given below:
9.4 7 7.6 6.3 6.7 8.6 6.8 10.6 8.9 9.4
a. Mean b. Median c. 25th percentile
d. 75th percentile e. 10th percentile f. Range
g. Iqr h. Standard deviation i. Variance
j. Coefficient of variance k. Relative standard deviation

Answers:
Multiple Choice Questions
1. c 2. d 3. c 4. d 5. a 6. c 7. a 8. b 9. a 10. b
11. c 12. a 13. a 14. c 15. c
Exercise
1. Mode 18.
2. Mode 10.
3. Mean 20.7.
4. Median 258, mean 259.75.
5. Range 1 to 15.
6. Mean 50.2, standard deviation 1.8.
7. Mean 23.4, standard deviation 2.6.
8. Mean 1.5, standard deviation 0.84.
9. Mean 3.1, median 3.
10. Mean 250, median 250.
11. Median 35, Mean 35 , Mode 35, standard deviation 15.40 .
12. Mean 3.25, median 3, standard deviation 1.25.
13. Median 86, Mean 56.68 , standard deviation 51.85.
230 Textbook of Computer Applications and Biostatistics

14. Mean 51.94 , standard deviation 11.42.

15. Steroid 1. Median 8.2, Range: 3.9 to 15.8, Quartile Q1 6.15, Q3 10.15.
Steroid 2. Median 6.9, Range: 4.4 to 9.1, Quartile Q1 5.73 , Q3 7.97.
16. Median 0.71, Range (0.1 to 1.24), Quartile Q1 0.54, Q3 0.83.
17.Alpha: CV 0.035, RSD 3.5; Bravo: CV 0.015, RSD 1.5.
The tablets prepared by Alpha machine show more variability than those prepared by Bravo
machine.
18. Median 4.1, Quartile Q1 3.3, Q3 7.9.
19. Mean 112.78, SD 14.42, CV 0.128, RSD 12.8.
20.
a. Mean 8.13 b. Median 8.1
c. 25th percentile 6.925 d. 75th percentile 9.4
e. 10th percentile 6.34 f. Range 6.3 to 10.6
g. Iqr 6.925 to 9.4 h. Standard deviation 1.44
i. Variance 2.09 j. Coefficient of variance 0.177
k. Relative standard deviation 17.7%.
 Chapter 13
PROBABILITY AND PROBABILITY DISTRIBUTION

Learning objectives
When we have finished this chapter, we should be able to:
1. Define probability and calculate simple probabilities.
2. Explain the proportional frequency approach to calculate probability.
3. Explain probabilities of simple and composite outcomes, probability involving two variables and
conditional probability.
4. Explain how probability can be used with the area properties of the normal distribution.
5. Explain probability distribution, binomial distribution and poisson distribution.

Classic Probability
Statistical concepts are essentially derived from probability theory. Thus, it would be only
logical to review some of the fundamentals of probability.
Probability is a measure of the chance of getting some outcome of interest from some event.
The event might be rolling a dice and the outcome of interest might be getting a six. Some basic ideas
about probability are given below:
1. The probability of a particular outcome from an event will lie between zero and one.
2. The probability of an event that is certain to happen is equal to one. For example, the probability
that everybody dies eventually.
3. The probability of an event that is impossible is zero. For example, throwing a seven with a normal
dice.
4. If an event has as much chance of happening as of not happening (like tossing a coin and getting a
head), then it has a probability of 1/2 or 0.5.
5. If the probability of an event happening is p, then the probability of the event not happening is 1 – p.

Calculating Probability
We can calculate the probability of a particular outcome from an event by using the
following formula:
Number of outcomes that favour the event (m) ...1
Probability [P E ] =
Total number of outcomes (N)
The probability of a particular outcome from an event is equal to the number of outcomes that favour
that event, divided by the total number of possible outcomes.

231
 232 Textbook of Computer Applications and Biostatistics

Example 13.1
What is the probability of getting an odd number when we roll a dice?
Solution
Total number of possible outcomes = 6 (1 or 2 or 3 or 4 or 5 or 6)
Total number of outcomes favouring the event ‘an odd number’= 3 (i.e. 1 or 3 or 5)
So probability of getting an odd number = 3/6 = ½= 0.5
The above method for determining probability works well with experiments where all of the
outcomes have the same probability, e.g. rolling dice, tossing a coin, etc. In the real world, however,
we will require to use the proportional frequency approach, which uses existing frequency data as the
basis for probability calculations.

Example 13.2
Each box contains 10 strips of paracetamol. As a check on quality such 50 boxes were tested
and distribution regarding number of defectives and the proportional frequency is calculated as
category frequency divided by total frequency.
Table 13.1 Frequency table showing number of defective strips in 50 boxes
No. of defectives Frequency Proportional
(n=50) frequency
0 30 30/50 = 0.6
1 10 10/50 = 0.2
2 5 5/50 = 0.1
3 3 3/50 = 0.06
4 1 1/50 = 0.02
5 1 1/50 = 0.02
Notice that the proportional frequencies sum to one, similar to probability.
Now if we ask the question, ‘What is the probability that if we chose one of these 50 boxes at
random and we will get one defective strip in it? The answer is the proportional frequency for the
‘one’ defective strip i.e. 0.2. In other words, we can interpret proportions as equivalent to
probabilities.

Probability of simple outcomes

The outcomes that are mutually exclusive and exhaustive are called as simple outcomes. Let
us see one example of determining probability of simple outcome.
Example 13.3
There are 52 cards in a deck, of which 26 are red; therefore, the probability of drawing a red
card is
P(Red) = 26/52 = 0.50
Note that cards must be red or black, and cannot be both; thus representing mutually
 Probability and Probability Distribution 233

exclusive and exhaustive simple outcomes.

Probability of composite outcomes

When the probability of group of simple outcomes is collectively taken it is referred to as
probability of composite outcomes.
Addition theorem
According to addition theorem, the likelihood of two or more mutually exclusive outcomes
equals the sum of their individual probabilities.
P(Ei or Ej)= P(Ei) + P(Ej) ...2
Example 13.4
The probability of a composite outcome of drawing a face card (jack, queen or king) would
equal the sum of their probabilities
P (Face card) = P (King) + P (Queen) + P (Jack) = 1/13 +1/13 +1/13 = 3/13 = 0.231.
Probability of complementary event
For any outcome E, there is a complementary event (`E ) which can be considered "not E,"
Since either E or `E must occur, but cannot occur at the same time then P(E) + P(`E ) = 1 or written for
the complement.
p(`E ) = 1 - p(E) ...3
Probability involving two variables
In the case of two different variables, it is necessary to consider the likelihood of both
variables occurring, p(A) and p(B), which are not mutually exclusive. A conjoint or union (AÈB) is
used when calculating the probability of either A or B occurring. An intersect (AÇB) or joint
probability is employed when calculating the probability of bothAand B occurring at the same time.
1. Probability of Intersect (AÇB)
The probability of an intersect is easily determined using the multiplication theorem, in
which p(AÇB) = p(A) x p(B) ifAand B are independent of each other.
p(Aand B) = p(A) x p(B)
Example 13.5
What is the probability of drawing a card which is both a queen and a heart (figure 13.1)
Intersect Universe
`Q `H

Q H

Figure 13.1 Probability of drawing a card which is both a queen and a heart
 234 Textbook of Computer Applications and Biostatistics

p(queen and heart) = p(QÇH ) = 1/ 52

p(queen and heart) = p(queen) x p(heart) =1 /13 x 1/ 4 =1/ 52
In this case there is obviously only one queen of hearts in a deck of cards.
2. Probability of conjoint
The probability of either A or B occurring i.e conjoint can be determined by using addition
theorem and subtracting intersect. The p(A or B) equals the sum of the two probabilities minus the
probability associated with the intersect.
P(AÈB) =p (A) + p (B) - p(AÇB)
Example 13.6
Take example of probability of drawing either queen or heart.
Conjoint Universe
`Q `H

Q H

Figure 13.2 Schematics of Conjoint Probability Distribution

Solution:
We can compute the conjoint as given below
p(queen or heart) = P(QÈH) = p(Q) + p(H) - p(QÇH)
p(queen or heart) = P(QÈH) = 4/ 52 + 13/52 - 1/52 =16/52
Here there are 13 heart cards and four queens for a total of 17, but one of the queens is also a
heart, thus the 16 possible outcomes.
Let us take example illustrating conjoint and intersect.

Example 13. 7
Consider that in a survey of 407 villages in Satara district, 219 villages had medical shops
while 305 had doctors in their villages; and 192 villages had both doctor and medical shops.
Assuming that this sample is representative of villages nationally,
1. What is the probability of selecting a village at random and finding that this village has a medical
shop (MS)?
2. What is the probability of selecting a village at random and finding that this village has doctors?
3. What is the probability of selecting a village at random and finding that this village does not have
doctor?
 Probability and Probability Distribution 235

4. What is the probability of selecting a village at random that has both medical shop and doctor
(intersect)?
5. What is the probability of selecting a village at random who has either medical shop or doctor but
both (conjoint)?
Solution:
1. Probability of medical shop, p(MS)
m(MS) = No of outcomes favoring event (here medical shop) = 219
N= No of possible outcomes (here villages) = 407
p(MS)= m(MS)/N= 219/407=0.538
The probability of selecting a village having medical shops at random is 0.538.
2. Probability of Doctor, p(D)
m(D) = 305; N= 407
p(D) = m(D) /N= 305/407= 0.749
The probability of selecting a village having doctor at random is 0.749.
3. Probability of no Doctor, p(nD)
m(nD) = (407-305); N= 407; p(D) =0.749
p(n D) = m(n D)/N = (407 -305)/407 = 0.251
or
p(n D) = 1- p(D) = 1- 0.749 = 0.251
The probability of selecting a village having no doctor at random is 0.251.
4. Probability of medical shop and doctor
m (MSÇD) = 192; N = 407
p(medical shop and doctor)= p(MSÇ D) = m (MSÇD)/ N= 192/407 =0.472
The probability of selecting a village having both medical shops and doctor at random is
0.472.
5. Probability of medical shop or doctor p(MS or D)
p(MS)= 0.538, p(D) = 0.749, p (MSÇD) = 0.472
p (medical shop or doctor) = p(MSÈD)= p(MS) + p(D) - p (MSÇD)
= 0.538 + 0.749 -0.472 = 0.915
The probability of selecting a village having medical shop or doctor at random is 0.915.

Conditional Probability
Many times it is necessary to calculate the probability of an outcome, given that a certain
value is already known for a second variable. For example, what is the probability of event A
occurring given the fact that only a certain level (or outcome) of a second variable (B) is considered.
p(A) given B = p(A|B) = p(AÇB) /p(B) ... 4
 236 Textbook of Computer Applications and Biostatistics

Example 13.8
What is the probability of drawing a queen from a stack of cards containing all the hearts
from a single deck?
Solution
p(QÇH) = 1/52, p(H)=13/52
Probability of (Queen| Heart), p(Q|H)= p(QÇH)/p(H)= (1/52)/(13/52)=1/13
In this example, if all the hearts are removed from a deck of cards, 1/13 is the probability of
selecting a queen from the extracted hearts.

Example 13.9
1. From the previous example 13.7, if a selected village has a medical shop, what is the probability
that this same village also has doctor?
2. If the selected village has doctor, what is the probability that this same village also has access to a
medical shop?
Solution:
1. If selected village is having medical shop, probability that this same village will have doctor is:
p(D|MS) = p(MSÇD) / p(MS) = 0.472/0.538 =0.877
2. If selected village is having doctor, probability that this same village will have medical shop is:
p(MS|D) = p(MSÇD)/ p(D) = 0.472/0.749 =0.630

Probability Distribution
Inferential statistics employs probability theory to make assumptions about the properties of
populations on the basis of data recorded from smaller samples taken from population. An
instrumental component of such estimations is the use of probability distributions, i.e. the
relationships between particular variable and their probability of occurrence.
1. Discrete probability distribution
Discrete probability distributions are those in which the probability of the occurrence of
discrete events is calculated and graphically portrayed. To illustrate these, consider firstly the
numerical outcome following the rolling of one die and then two dice (Table 13.2)
The plot of theoretical probabilities against defined variables is referred to as a probability
distribution. Frequency distributions represents the distribution of data derived from analysis of a
sample taken from a population, whereas probability distributions reflect the distributions of a
variable in a population.
In a probability distribution, the sum of all the individual probabilities is always 1 (the area
under the plotted distribution is 1).
 Probability and Probability Distribution 237

Table 13.2 Probabilities associated with defined numerical values obtained following the
rolling of one or two dice
One dice Two dice
Variable Probability Variable Probability
(Numerical outcome) (Numerical outcome)
1 0.167 2 0.038
2 0.167 3 0.056
3 0.167 4 0.083
4 0.167 5 0.111
5 0.167 6 0.139
6 0.167 7 0.167
8 0.139
9 0.111
10 0.083
11 0.056
12 0.038
0.2 0.2

0.16 0.16
Probability
Probability

0.12 0.12

0.08 0.08

0.04 0.04

0 0
1 2 3 4 5 6 2 3 4 5 6 7 8 9 10 11 12
Numerical outcome
Numerical outcome
a b
Figure 13.2 Probability distributions for the numerical values shown in table 13.2 after one dice (a)
and two dice (b)
The distribution shown in figure 13.2 is a discrete probability distribution because the
variable can adopt a countable number of values.

2. Binomial distribution
One of the distributions most commonly employed in the pharmaceutical and life science is
the binomial distribution. This distribution is used whenever the outcome of an event consists of only
two categories. An example of a binomial event has been described previously, namely tossing of a
coin. The other examples of binomial data include:
a. the outcome of a quality control assessment, which is either a pass or a fail.
b. a new formulation may produce side effects: the outcomes is either positive or negative (no effects)
c. the gender distribution in a population: the disease is either present or absent
d. a new pharmaceutical agent is either clinically efficacious or non-efficacious.
 238 Textbook of Computer Applications and Biostatistics

In addition to the requirement for only two possible outcomes, each binomial trial must be
independent, i.e. the occurrence of one events must not influence subsequent events. In the generation
of the binomial distribution, it is assumed that the proportion of observations (or individuals) in one
category is p and, consequently, the proportion of observations in the other category is 1-p, i.e. q.
The probability of events using binomial data can be calculated by expansion of the binomial
n
term, (p+q) , in which n denotes the sample size, p is probability of the occurrence of the first event, q
is the probability of the occurrence of the second event.
The probability of X observations in a sample of size n that has been removed from a
binomial distribution may be mathematically described as follows:

æn ö
P(X) = çç ÷÷ p X q n - X ...6
èXø
X
Where p denotes the probability of a sample composed of X observations possessing a
n-x
possibility p and q denotes the probability of a sample composed of n-X observations possessing a
probability q.
The above equation can be rewritten by substituting
æn ö n!
çç ÷÷ =
è ø
X X ! (n - X )! ...7
n!
P(X) = pXq n -X ...8
X!(n - X)!

Example 13.3
In pharmaceutical manufacturing, tablets from a batch may be categorized into those that
pass or those that fails quality control. As part of QC process for a batch of tablets, the probability
associated with the pass category was 0.95 whereas the probability associated with fail category was
0.05. Using the binomial expansion calculate the probability of selecting three defective tablets in a
sample of three.
Solution:
The probabilities were p (defective)= 0.05,
q (non defective) = 0.95,
The number of observations (X) = 3,
Sample size (n) = 3
æn ö æ3ö
P(X ) = çç ÷÷ p X q n - X = çç ÷÷ ( 0 . 05 ) 3 ( 0 . 95 ) 3 - 3
èXø è3ø
3!
= ´ 0.000125 = 0.000125
3!(0!)
Thus, the probability of selecting three defective tablets in a sample of three is 0.000125.
 Probability and Probability Distribution 239

3. Poisson distribution
The Poisson distribution is another discrete data distribution that is commonly employed to
describe random occurrences when the probability of observing an event is small. the Poisson
distribution approximates to the binomial distribution when the sample size is large and the
probability of a specified event is small. Mathematically, the Poisson distribution is described by
e- m m X mX
P(X) = = m
X! e X! ...9
Where
p(X)= the probability of an event occurring in a single observation
m = the mean number of occurrences (number of observations, x and probability, Np).
The above equation may be expanded to enable calculation of the probabilities of occurrence
of an event or events (Table 13.3).
Table 13.3 Expansion of the equation that defines the Poisson distribution

Variable Probability
(Numerical outcome)

0 P(0) = e-m
1 P(1) = e-mm
2 P(2) = e-mm2/2!
3 P(3) = e-mm3/3!
4 P(4) = e-mm4/4!

Probability and Normal Distribution

If data is Normally distributed then about 95 per cent of the values will lie no further than two
standard deviations from the mean (see Figure 13.3 ). In probability terms, we can say that there is a
probability of 0.95 that a single value chosen at random will lie no further than two standard
deviations from the mean.
1.About 68 % of the observations lie within one standard deviation either side of mean.
2.About 95 % of the observations lie within two standard deviation either side of mean.
3.About 99 % of the observations lie within three standard deviation either side of mean.

-3SD -2SD -1SD `X +1SD +2SD +3SD `X

Figure 13.3 The area of properties of the normal distribution
 240 Textbook of Computer Applications and Biostatistics

Summary
Probability is a measure of the chance of getting some outcome of interest from some
event.
Number of outcomes that favour the event (m)
Probability [P E ] =
Total number of outcomes (N)

Probability of composite outcomes

P(Ei or Ej)= P(Ei) + P(Ej)
Probability of complementary event
p(`E ) = 1 - p(E)
Probability of intersect
p(Aand B) = p(AÇB)=p(A) x p(B)
Probability of conjoint
p(Aor B)= p(AÈB) =p (A) + p (B) - p(AÇB)
Conditional probability
p(A) given B = p(A|B) = p(AÇB) /p(B)
Probability of binomial distribution

æn ö æn ö n!
P(X) = çç ÷÷ p X q n - X çç ÷÷ =
èXø è X ø X!(n - X)!
n!
P(X) = pXq n -X
X!(n - X)!
Poisson distribution
e-m m X mX
P(X) = = m
X! e X!

Multiple Choice Questions

1. Approximately what percentage of scores fall within one standard deviation of the mean in a
normal distribution?
a. 34% b. 95%
c. 99% d. 68%
2. The sum of all probabilities is________.
a. 1 b. 2
c. 100 d. 10
 Probability and Probability Distribution 241

3. The distribution used whenever the outcome of an event consists of only two categories
is________.
a. Normal distribution b. Poisson distribution
c. Binomial distribution d. Uniform distribution
4. The distribution used when the probability of observing an event is small is________.
a. Normal distribution b. Poisson distribution
c. Binomial distribution d. Uniform distribution
5. The likelihood of two or more mutually exclusive outcome equals the sum of their individual
probabilities is given by
a. Conjoint b. Intersect
c. Multiplication theorem d.Addition theorem
6. The probability of an intersect p(AÇB) is easily determined by using ______.
a.Addition theorem b. Multiplication theorem
c. Baye theorem d.Addition and subtracting theorem
7. The probability of an conjoint p(AÈB) is easily determined by using ______.
a.Addition theorem b. Multiplication theorem
c. Bayes theorem d.Addition and subtracting theorem
8. The outcomes that are mutually exclusive and exhaustive are _________ outcomes.
a. simple b. composite
c. complementary d. conjoint
9. The probability of an event that is certain to happen is equal to _______ .
a. 100 b. 1
c. 10 d. 0
10.About 99% of the observations lie within _______ standard deviation either side of mean.
a. 1 b. 2
c. 3 d. 4

Exercise
1. A newly designed shipping containers for ampules was compared to the existing one to determine if
the number of broken units could be reduced. One hundred shipping containers of each design (old
and new) were subjected to identical rigorous abuse. The containers were evaluated and failures were
defined as containers with more than 1% of the ampules broken. A total of 15 failures were observed
and 12 of those failures were with the old container. If one container was selected at random:
a. What is the probability that the container will be of the new design?
b. What is the probability that the container will be a "failure"?
c. What is the probability that the container will be a "success"?
d. What is the probability that the container will be both an old container design and a
“failure"?
 242 Textbook of Computer Applications and Biostatistics

e. What is the probability that the container will be either of the old design or a "failure"?
f. If one container is selected at random from only the new containers, what is the probability
that the container will be a "failure"?
g. If one container is selected at random from only the old container design, what is the
probability that the container will be a "success"?

2. Total of 150 healthy females volunteered to take part in a multi-center study of a new urine testing
kit to determine pregnancy. One-half of the volunteers were pregnant, in their first trimester. Urinary
pHs were recorded and 62 of the volunteers were found to have a urine pH less than 7.0 (acidic) at the
time of the study. Thirty-six of these women with acidic urine were also pregnant.
If one volunteer is selected at random:
a. What is the probability that the volunteer is pregnant?
b. What is the probability that the volunteer has urine that is acidic, or less than a pH 7?
c. What is the probability that the volunteer has a urine 'which is basic or a pH equal to or
greater than 7?
d. What is the probability that the volunteer is both pregnant and has urine which is acidic
or less than pH 7?
e. What is the probability that the volunteer is either pregnant or has urine which is acidic
or less than pH 7?
f. If one volunteer is selected at random from only those women with acidic urinary pHs,
what is the probability that the volunteer is also pregnant?
g. If one volunteer is selected at random from only the pregnant women, what is the
probability that the volunteer has a urine pH of 7.0 or greater?

3. If a medicine cures 80% of the people who take it, what is the probability that out of the eight people
who take the medicine, 5 will be cured?

4. If a microchip manufacturer claims that only 4% of his chips are defective, what is the probability
that among the 60 chips chosen, exactly three are defective?

5. If a telemarketing executive has determined that 15% of the people contacted will purchase the
product, what is the probability that among the 12 people who are contacted, 2 will buy the product?

6. A department store buys 50% of its appliances from Manufacturer A, 30% from Manufacturer B,
and 20% from Manufacturer C. It is estimated that 6% of Manufacturer A's appliances, 5% of
Manufacturer B's appliances, and 4% of Manufacturer C's appliances need repair before the warranty
expires. An appliance is chosen at random. If the appliance chosen needed repair before the warranty
expired, what is the probability that the appliance was manufactured by Manufacturer A?
Manufacturer B? Manufacturer C?
 Probability and Probability Distribution 243

Answers:
Multiple Choice Questions
1. d 2. a 3. c 4. b 5. d 6. b 7. d 8. a 9. b 10. c
Exercise
1.
a. 0.5, b. 0.075,
c. 0.925, d. 0.06,
e. 0.515, f. 0.03,
g. 0.88
2.
a. 0.5, b. 0.413,
c. 0.587, d. 0.24,
e. 0.673, f. 0.581,
g. 0.52
3. 0.1464
4. 0.2137
5. 0.292.
6.A- 0.03, B-0.015, C-0.008
 Chapter 14
SAMPLING TECHNIQUES

Learning objectives
When we have finished this chapter, we should be able to:
1. Explain various methods of sampling techniques
2. Explain characteristics of good sample

Samples are considered to be the true representatives of that population. It is not possible to
include all the members of the population in an experimental study because of constraints of cost,
time and labour involved. Hence, appropriate sampling techniques are utilised to ensure that the
samples are randomly selected and that every observation is independently measured.
Various sampling techniques used are given below:
1. Random sampling
2. Systematic sampling
3. Stratified sampling
4. Cluster sampling
5. Multistage sampling
6. Multiphase sampling
7. Sequential sampling

1. Random Sampling
In this method, every unit of the population has an equal chance of being selected. This
method is applicable when population is small, homogeneous and readily available. This method is
also called as ‘Unrestricted Random Sampling’.
Randomization may be accomplished for a smaller number of units by using a random
numbers table or by numbers generated at random using a calculator or computer.
Lottery Method
This is the simplest and most popular method of obtaining a random sample. Under this
method, various units of population we are going to study are numbered on small and identical slips of
paper, which are folded and put them into a box or a bag. Then they are thoroughly mixed and required
number of slips for the sample are picked one after the other without replacement. While doing this, it
has to be ensured that in successive drawings each of the remaining slips of population has equal
chance of being chosen.
Use of Random Number
Common method of drawing sample is by making use of published tables of random

244
 Sampling Techniques 245

numbers. First give serial numbers to all the people of the study population. Then select at random,
any page of the random number table and pick up the number in any row and column at random. The
population units corresponding to the numbers are selected.
Advantages of random sampling
1. Scientific method
2. More representative
3. More economical
Disadvantages of random sampling
1. It needs complete list of study population which is often difficult to get.
2. If the sample size is small, this sample will not be a true representative of the universe.
3. Cases selected by random sampling tends to be widely dispersed geographically and cost
of collecting data becomes too large.

2. Systematic Sampling
This method is popularly used in those cases when a complete list of population from which
sample is to be drawn, is available. It is more often applied to field studies when the population is
large, scattered and homogeneous.
The most practical way of sampling is to select every ith item on a list. Sampling of this type
is known as systematic sampling. An element of randomness is introduced into this kind of sampling
by using random numbers to pick up the unit with which to start. Sample interval for systematic
sampling is calculated using following formula:
Sample interval = Total population / Sample size desired
For instance, if a 4 per cent sample is desired, the first item would be selected randomly from
the first twenty-five and thereafter every 25th item would automatically be included in the sample.
Thus, in systematic sampling only the first unit is selected randomly and the remaining units of the
sample are selected at fixed intervals. Although a systematic sample is not a random sample in the
strict sense of the term, but it is often considered reasonable to treat systematic sample as if it were a
random sample.
Advantages
It can be taken as an improvement over a simple random sample in as much as the systematic
sample is spread more evenly over the entire population. It is an easier, accurate and less costlier
method of sampling and can be conveniently used even in case of large populations.
Disadvantages
If there is a hidden periodicity in the population, systematic sampling will prove to be an
inefficient method of sampling. In practice, systematic sampling is used when lists of population are
available and they are of considerable length.

3. Stratified Sampling
If a population from which a sample is to be drawn does not constitute a homogeneous group,
 246 Textbook of Computer Applications and Biostatistics

stratified sampling technique is generally applied in order to obtain a representative sample. Under
stratified sampling the population is divided into several sub-populations that are individually more
homogeneous than the total population (the different sub-populations are called ‘strata’) and then we
select items from each stratum to constitute a sample. Since each stratum is more homogeneous than
the total population, we are able to get more precise estimates for each stratum and by estimating more
accurately each of the component parts, we get a better estimate of the whole. In brief, stratified
sampling results in more reliable and detailed information.
Advantages
1. More representative
2. Greater accurate
3.Administrative convenience
4. More advantageous when distribution of population is skewed.
Disadvantages
1. It is very difficult task to divide the population into homogenous strata. This may require
considerable time and money and statistical expertise.
2. The supplementary information to set up strata is not available some times.
3. Sometimes the different strata may overlap and the sampling would not be representative.

4. Multistage Sampling
As the name implies this method refers to the sampling procedures carried out in several
stages using random sampling techniques. Ordinarily multi-stage sampling is applied in big inquiry
extending to a considerable large geographical area, say, the entire country. In the first stage, random
numbers of districts are chosen in all the states, followed by random numbers of talukas, villages and
units, respectively.
Advantages
1. It is easier to administer than most single stage designs mainly because of the fact that
sampling frame under multi-stage sampling is developed in partial units.
2. Alarge number of units can be sampled for a given cost under multistage sampling.
3. It introduces flexibility in sampling.
4. It enables the use of existing divisions and subdivisions which saves extra labour.

5. Cluster Sampling
If the total area of interest happens to be a big one, a convenient way in which a sample can be
taken is to divide the area into a number of smaller non-overlapping areas and then to randomly select
a number of these smaller areas called clusters, with the ultimate sample consisting of all units in these
small areas or clusters.
Thus in cluster sampling the total population is divided into a number of relatively small
subdivisions which are themselves clusters of still smaller units and then some of these clusters are
randomly selected for inclusion in the overall sample.
 Sampling Techniques 247

Advantages
1. Data collection method is simple and economic.
2. Cluster sampling reduces cost by concentrating surveys in selected clusters.
3. Involves less time and cost.
4. Estimates based on cluster samples are usually more reliable per unit cost.
Disadvantages
1. Gives higher standard error.
2. It is less precise than random sampling.
3. There is not as much information in observations within a cluster.

6. Multiphase Sampling
In this method part of information is collected from the whole sample and part of information
is from sub sample.
Advantages
1. Less cost
2. Less laborious
3. More purposeful.

7. Sequential Sampling
This sampling design is some what complex sample design. The ultimate size of the sample
under this technique is not fixed in advance, but is determined according to mathematical decision
rules on the basis of information yielded as survey progresses. This is usually adopted in case of
acceptance sampling plan in context of statistical quality control. When a particular lot is to be
accepted or rejected on the basis of a single sample, it is known as single sampling; when the decision
is to be taken on the basis of two samples, it is known as double sampling and in case the decision rests
on the basis of more than two samples but the number of samples is certain and decided in advance,
the sampling is known as multiple sampling. But when the number of samples is more than two but it
is neither certain nor decided in advance, this type of system is often referred to as sequential
sampling. Thus, in brief, we can say that in sequential sampling, one can go on taking samples one
after another as long as one desires to do so.

Characteristics of Sample
The main characteristics of a representative sample are as follows:
1. Precision andAccuracy
2. Unbiased character
1. Precision andAccuracy
Precision refers to how closely data are grouped together or the compactness of the
sample data. Precision measures the variability of a group of measurements. A precise set of
 248 Textbook of Computer Applications and Biostatistics

measurements is compact and is reflected by a small relative standard deviation.

Assume that the smaller box within samples A, B and C represent the true value for the
population from which the samples were taken. In this example, even though samples A and C
have good precision, only sample C is accurate as a predictor of the population.
Precision is measured by the formula,
N
Precision =
s
Where,
N is number of sample.
s is standard deviation.
Accuracy is concerned with "correctness" of the results and refers to how closely the
sample data represents the true value of the population. It is desirable to have data that is both
accurate and precise.
SampleA Sample B Sample C

Figure 15.1 Sample comparing precision and accuracy

2. Bias
Bias can be thought of as systematic error which causes some type of constant error in
the measurement with the measurement system.
Selection bias occurs when certain characteristics make potential observations more (or
less) likely to be included in the study. For example, always sampling from the top of storage
drums may bias the results based on particle size, assuming smaller particles settle to the lower
regions of the drums. Bias can result from an incorrect sampling, inappropriate experimental
design, inadequate blinding, or mistakes (blunders) in observing or recording the data.

The following are the characteristics of a good sample

1. The sample is randomly selected.
2. Every sample is independent of other.
3. The sample is precise and accurate.
4. The sample selection is unbiased.
5. The sample is reliable and valid.
 Sampling Techniques 249

Summary
Sample
Samples are relatively small group of observations that are taken from a defined population
and considered to be the true representative of that population.
Sampling techniques
1. Random sampling
2. Systematic sampling
3. Stratified sampling
4. Cluster sampling
5. Multistage sampling
6. Multiphase sampling
7. Sequential sampling
Characteristics of Sample
1. Precision and accuracy
2. Unbiased character

Multiple choice questions

1. Which of the following techniques yields a simple random sample?
a. Choosing volunteers from an introductory psychology class to participate
b. Listing the individuals by ethnic group and choosing a proportion from within each
ethnic group at random.
c. Numbering all the elements of a sampling frame and then using a random number
table to pick cases from the table.
d. Randomly selecting schools, and then sampling everyone within the school.
2. Which of the following sampling techniques is an equal probability selection method in which
every individual in the population has an equal chance of being selected?
a. Simple random sampling b. Systematic sampling
c. Proportional stratified sampling d.All of the above
3. Which of the following will give a more "accurate" representation of the population from
which a sample has been taken?
a.Alarge sample based on the convenience sampling technique
b.Asmall sample based on simple random sampling
c.Alarge sample based on simple random sampling
d.Asmall cluster sample
4. Which of the following would generally require the largest sample size?
a. Cluster sampling b. Simple random sampling
 250 Textbook of Computer Applications and Biostatistics

c. Systematic sampling d. Proportional stratified sampling

5. Which of the following would usually require the smallest sample size because of its
efficiency?
a. One stage cluster sampling b. Simple random sampling
c. Two stage cluster sampling d. Quota sampling
6.Atechnique used when selecting clusters of different sizes is called _____.
a. Cluster sampling b. One-stage sampling
c. Two-stage sampling d. Probability proportional to size or PPS
7. The process of drawing a sample from a population is known as _________.
a. Sampling b. Census c. Survey research d. None of the above
8. ___________ is a set of elements taken from a larger population according to certain rules.
a. Sample b. Population c. Statistic d. Element
9. Determining the sample interval (represented by k), randomly selecting a number between 1
and k, and including each kth element in your sample are the steps for which form of sampling?
a. Simple Random Sampling b. Stratified Random Sampling
c. Systematic Sampling d. Cluster sampling
10. A______ is a subset of a _________.
a. Sample, population b. Population, sample
c. Statistic, parameter d. Parameter, statistic

Exercise
1. Explain various methods of sampling techniques.
2. What are the characteristics of good sample?
3. From the batch of 50 tablets, sampling of 10 tablets is to be done. Explain how the systematic
sampling and random sampling method can be used?
4. Write a note on stratified sampling.
5. Explain the terms: a) Precision b)Accuracy c) Bias

Answers:
Multiple Choice Questions
1. c 2. d 3. c 4. a 5. a 6. d 7. a 8. a 9. c 10. a
 Chapter 15
ESTIMATION OF CONFIDENCE INTERVAL

Learning objectives
When we have finished this chapter, we should be able to:
1. Explain what the standard error of the sample mean is and calculate its value.
2. Explain how we can use the probability properties of the Normal distribution to measure the
preciseness of the sample mean as an estimator of the population mean.
3. Estimate the confidence interval of the population mean.
4. Calculate and interpret a 95 per cent confidence interval for a population mean.

Concept of Confidence Interval

The mean and standard deviation of sample data are employed to estimate the true mean and
true standard deviation. However, it is reasonable to know how reliable is the sample data at
representing the population data. After calculating the mean and standard deviation of a sample, as is
the normal approach in the pharmaceutical sciences, we need to provide an indication of the
reliability of the data.
In light of the small sample size, it is unreasonable to predict that the population mean and
standard deviation will be identical to these observed sample values, as each sample will produce
different mean and standard deviation values. Therefore, when reporting the mean of sample data it is
good practice to present some indication of the reliability of the data, i.e. the quality of the estimation
of the true mean from the sample mean. This is performed using confidence intervals. The confidence
intervals are quoted as a mean and range , the latter representing the probability of observing the true
mean.

Standard Error of the Mean

The standard deviation (s or SD) describes the variability within a sample; whereas, the
standard error of the mean (SEM) represents the possible variability of the mean itself. The SEM is
sometimes referred to as the standard error (SE) and describes the variation of all possible sample
means and equals the standard deviation (SD) of the sample data divided by the square root of the
sample size. The distribution of sample means (the standard error of the mean) will always be smaller
than the dispersion of the sample (the standard deviation). The SEM is calculated by the following
formula:
SD
SEM = ...1
N

251
 252 Textbook of Computer Applications and Biostatistics

Where
SEM = Standard error of mean
SD = Standard deviation
N = number of observations
Standard error of the mean can be considered as a measure of precision. Obviously, the
smaller the SEM, the more confident we can be that our sample mean is closer to the true population
mean. A general rule of thumb is that with samples of 30 or more observations, it is safe to use the
sample standard deviation as an estimate of population standard deviation.
Example 15.1
Diastolic blood pressure of 322 males was taken. Mean BP was found to be 95 and SD 12
mm. Determine SEM.
Answer:
Data: N =322,
s = 12 mm
Formula:
s
SE M =
N
Solution:
12
SEM = = 0.67
322
SEM was found to be 0.67.

Example 15.2
Hardness of tablets from same batch was determined using Pfizer type hardness tester. The
data is as follows:
5.9 6.4 5.7 5.4 4.9 6.2 5.8 5.5 5.3 5.2
Calculate SEM.
Solution
1. First, calculate standard deviation of the given data
Formula:
(å X ) 2
å X2 -
Standard deviation (s) = N
N -1
Solution:
Where
åX = 56.3
åX = 318.89
2
 Estimation of Confidence Interval 253

2
( 56 . 3) 2 X X
318 . 89 - 5.9 34.81
Standard deviation (s) = 10 6.4 40.96
10 - 1 5.7 32.49
Standard deviation (s) = 0 .21 5.4 29.16
4.9 24.01
Standard deviation (s) = 0.458 6.2 38.44
5.8 33.64
5.5 30.25
5.3 28.09
5.2 27.04
2. Now determine SEM 56.3 318.89
N = 10; s = 0.458
Formula:
s
SE M =
N
Solution:
0.458
SEM = = 0 .144
10
SEM was found to be 0.144

Estimation of Confidence Interval

We have already seen in chapter 13 that we can be 95 percent confident that any sample mean
is going to be within plus or minus two standard errors of the population mean.

-2.58 SE -1.96 SE -1.65 SE m +1.65 SE +1.96 SE +2.58 SE X

90%
95%
99%

Figure 15.1 Confidence limits of sample mean (`X) from the population mean (m) are (m± 1.96 SE
and m± 2.58 SE )
 254 Textbook of Computer Applications and Biostatistics

From this we can say that:

At 95% confidence interval level,
Population mean = sample mean ± 1.96 × standard error ...2
That is:
1. We can be 95 per cent confident that the interval, from the sample mean – 1.96 × standard error, to
the sample mean + 1.96 × standard error, will include the population mean.
2. Or in probability terms, there is a probability of 0.95 that the interval from the sample mean – 1.96
× standard error, to the sample mean + 1.96 × standard error, will contain the population mean.
In other words, if we pick one out of all the possible sample means at random, there is a
probability of 0.95 that it will lie within two standard errors of the population mean. We call the
distance from the sample mean – 1.96 × SEM (`X), to the sample mean + 1.96 × SEM(`X), the
confidence interval.
Confidence interval for mean can be calculated using following formula:

z1-a/2 s
p%=X± ...3
N
Where,
p % = selected confidence interval (90%, 95% or 99%)
`X = observed mean
Z(1-a/2) = z value corresponding to the percentage confidence interval (1.65 for 90%), (1.96 for
95%) and (2.58 for 99%)
s = standard deviation of population
N = number of observations

Example 15.3
Clinical study of new anticoagulant drug has been performed in 30 patients, in which the
volume of distribution has been calculated as 10.2 ± 1.9 l. Calculate the 95% confidence interval of
the mean value.
Answer:
Data:
`X = observed mean = 10.2
Z(1-a/2) = z value corresponding to the 95% percentage confidence interval=1.96
s = standard deviation =1.91
N = number of observations = 30
Formula:
z1-a/2 s
p 95% = X ±
N
 Estimation of Confidence Interval 255

Solution
1.96 ´ 1.91
p95% = 10.2 ± = 10.2 ± 0.68
30
The 95% confidence interval was found to be 10.2±0.68 L, i.e. 9.52-10.88 L.

Example 15.4
The following are the results of assay
Tablet No 01 0 2 0 3 0 4 05 06 0 7 0 8 0 9 10 11 12 13 14 15
Assay (mg) 75 74 72 78 78 74 75 77 76 78 73 77 75 74 72
Tablet No 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
Assay (mg) 75 76 75 73 76 79 73 76 75 80 77 74 76 77 74
Assume that three assays are selected at random to give sample of tablets: 4, 18 and 26.
Estimate 90%, 95% and 99% confidence intervals assuming that population standard
deviation is 2.04.
Answer:
Taking the assay results of sampled tablets 4, 18 and 26, let us calculate mean
Data:
78 75 77
Formula
Sum of all observations ( å X)
Mean ( X ) =
Total number of observations (N)
Solution
78 + 75 + 77 230
Mean ( X ) = = = 76.66
3 3
Mean = 76.66 mg
1. 90 % confidence interval
Data:
`X = observed mean = 76.66
Z(1-a/2) = z value corresponding to the 90% percentage confidence interval=1.65
s = standard deviation =2.04
N = number of observations = 3
Formula
z 1-a /2 s
p 90% = X ±
N
Solution
1.65 ´ 2.04
p 90% = 76.66 ±
3
 256 Textbook of Computer Applications and Biostatistics

p 90% = 76.66 ± 1 .95

74.71 < m < 78.61
90% confidence interval for population mean lies between 76.66± 1.95 mg.
2. 95 % confidence interval
Data:
`X = observed mean = 76.66
Z(1-a/2) = z value corresponding to the 95% percentage confidence interval=1.96
s = standard deviation =2.04
N = number of observations = 3
Formula
z 1 -a /2 s
p 95% = X ±
N
Solution
1 .9 6 ´ 2 .0 4
p 9 5 % = 7 6 .6 6 ±
3

p 95% = 76.66 ± 2 .3

74.36 < m < 78.96

95% confidence interval for population mean lies between 76.66± 2.30 mg.
3. 99 % confidence interval
Data:
`X = observed mean = 76.66
Z(1-a/2) = z value corresponding to the 99% percentage confidence interval= 2.58
s = standard deviation =2.04
N = number of observations = 3
Formula
z s
p 9 9 % = X ± 1 -a /2
N
Solution
p 99% = 76.66 ± 3.01
2.58 ´ 2.04
p 99% = 76.66 ±
3
73.65 < m < 79.67
99% confidence interval for population mean lies between 76.66± 3.01 mg.
 Estimation of Confidence Interval 257

Summary
Standard error of Mean
SD S
SEM = =
N N
At 95% confidence, Population mean = Sample mean± 1.96 x SE
Confidence interval
z s
p % = X ± 1 - a /2
N

Multiple choice questions

1. Standard error of mean is a measure of __________.
a.Accuracy b. Precision c. Bias d.All of above
2. What does it mean when we calculate a 95% confidence interval?
a. The process we used will capture the true parameter 95% of the time in the long run
b. We can be "95% confident" that the interval will include the population parameter
c. We can be "5% confident" that the interval will not include the population parameter
d.All of the above statements are true
3. What would happen (other things equal) to a confidence interval if we calculate a 99 percent
confidence interval rather than a 95 percent confidence interval?
a. It will be narrower b. It will not change
c. The sample size will increase d. It will become wider
4. What is the standard deviation of a sampling distribution called?
a. Sampling error b. Sample error
c. Standard error d. Simple error
5.As a general rule, researchers tend to use ____ percent confidence intervals.
a. 99% b. 95%
c. 50% d. none of the above
6. z value corresponding to the 95 percent confidence interval is ________.
a. 1.96 b. 1.65
c. 2.58 d. 1.80
7. Standard error of mean (SEM) represents the variability of _________.
a. sample b. mean itself
c. SD d. sample size
8. With samples less than 30, _______ is an estimate of population SD.
a. SEM b. sample SD
c. sample mean d. population mean
 258 Textbook of Computer Applications and Biostatistics

9. With samples more than 30, _______ is an estimate of population SD.

a. SEM b. sample SD
c. sample mean d. population mean
10. The confidence intervals are quoted as a mean and _______.
a. SD b. SEM
c. range d. percent

Exercise:
o
1. 100 aspirin tablets were removed from batch and the stability of drug was determined at 40 C. The
degradation rate constant for the drug was 0.09±0.01 per hour. Calculate the 95% confidence interval.
2. Six bottles from batch of Roxithromycin suspension were assayed by spectrophotometric method.
The results are given below. Estimate confidence interval.
52 57 49 51 53 52
3. Estimate 95% confidence interval, if the mean assay of 20 Paracetamol tablets is 524.3 mg and
standard deviation is 3.5 mg.
4. Estimate 99% confidence interval for the following data:
Mean = 18.85; N = 80, standard deviation = 5.55.
5. Estimate 90% confidence interval for the hardness of tablets:
Mean = 5.62; N = 40, standard deviation = 0.68.

Answers
Multiple Choice Questions
1. b 2. d 3.d 4.c 5.b 6. a 7. b 8. a 9. b 10. c

Exercise:
1. 95 % confidence interval = 0.09± 0.00196.
2. Mean = 52.33, SD=2.683, 95 % confidence interval = 52.33± 2.147.
3. 95 % confidence interval = 524.3± 1.533.
4. 99 % confidence interval = 18.85± 1.60.
5 .90 % confidence interval = 5.62± 0.177.
 Chapter 16
HYPOTHESIS TESTING

Learning objectives
When we have finished this chapter, we should be able to:
1. Explain how a research question can be expressed in the form of a testable hypothesis.
2. Explain what a null hypothesis is.
3. Summarise the hypothesis test procedure.
4. Explain what a p-value is.
5. Use the p-value to appropriately reject or not reject a null hypothesis.
6. Describe type I and type II errors, and their probabilities.

Introduction
Hypothesis testing is the process of inferring from a sample whether or not to accept a certain
statement about a population or populations. The sample is assumed to be a small representative
proportion of the total population. Two errors can occur, rejection of a true hypothesis or failing to
reject a false hypothesis.
Using an inferential statistics there are two possible outcomes:
H0: Hypothesis Under Test (Null Hypothesis)
Ha:Alternative Hypothesis (Research Hypothesis)
By convention, the Null hypothesis is stated as no real differences in the outcomes.
For example, if we are comparing three levels of a discrete independent variable (m1, m2, m3), the null
hypothesis would be stated as m1 = m2 = m3. The evaluation then attempts to nullify the hypothesis of no
significant difference in favor of an alternative research hypothesis.
Null hypotheses reflect the conservative position of no difference, no risk, no effect, etc.,
hence the name, ‘null’ hypothesis. To test this null hypothesis, researchers will have to take samples
and measure outcomes, and decide whether the data from the sample provides strong enough
evidence to be able to refute or reject the null hypothesis or not. If evidence against the null hypothesis
is strong enough for us to be able to reject it, then we are implicitly accepting that some specified
alternative hypothesis, usually labeled Ha, is probably true.

Hypothesis Testing Procedure

The hypothesis testing process can be summarised as given below:
1. Select a suitable outcome variable.
2. Use the research question to define an appropriate and testable null hypothesis involving this

259
 260 Textbook of Computer Applications and Biostatistics

outcome variable.
3. Collect the appropriate sample data and determine the relevant sample statistic, e.g. sample mean,
sample proportion, sample median, (or their difference or ratio), etc.
4. Use a decision rule that will enable us to judge whether the sample evidence supports or does not
support the null hypothesis.
5. Thus, on the strength of this evidence, either reject or do not reject the null hypothesis.
Let’s take a simple example. Suppose we want to test whether a coin is fair, i.e. not weighted
to produce more heads or more tails than it should. The null hypothesis is that the coin is fair, i.e. it will
produce as many heads as tails, so that the population proportion π, equals 0.5. The outcome variable
is the sample proportion of heads, p. If we toss the coin 100 times, and get 41 heads, then p = 0.41. Is
this outcome compatible with our hypothesised value of 0.5? Is the difference between 0.5 and 0.41
statistically significant or could it be due to chance? We decide what proportion of heads we might
expect to get if the coin is fair, by using two approaches.
1. creation of a confidence interval or
2. a comparison with a critical value.
The estimation of confidence interval has already been studied in the previous chapter, This
can be estimated by using the formula
s
p% = X ± z % ´
N ...1
Population Mean = Estimate Sample Mean ± Reliability Coefficient x Standard Error.
In the second method we would calculate a "test statistic', a value based on the manipulation
of sample data. This value is compared to a preset "critical" value (usually given in a special table)
based on a specific acceptable error rate (i.e., 5%). If the test statistic is extremely rare it will be to the
extreme of our critical value and we will reject the null hypothesis under test in favor of the research
hypothesis.
Decision rule for hypothesis testing
1. Determine the p-value for the output we have obtained. A p-value is the probability of getting the
outcome observed (or one more extreme), assuming the null hypothesis to be true.
2. Compare it with the critical value, usually 0.05.
3. If the p-value is less than the critical value, reject the null hypothesis; otherwise do not reject it.
It’s important to stress that the p-value is not the probability that the null hypothesis is true or
not true. It’s a measure of the strength of the evidence against the null hypothesis. The smaller the p-
value, the stronger the evidence. This means it is less likely that the outcome we have got is occurred
by chance. Note that the critical value, usually 0.05 or 0.01, is called the significance level of the
hypothesis test and is denoted as α (alpha).

Types of Error
Whenever we decide either to reject or not to reject a null hypothesis, we may be making a
mistake. After all, we are basing the decision on sample evidence. Even if we have done everything
 Hypothesis Testing 261

right, the sample could still, by chance, not be very representative of the population. Moreover, the
test might not be powerful enough to detect an effect if there is one.
There are two possible errors associated with hypothesis testing. Type I error is the
probability of rejecting a true null hypothesis (H0) and Type II error is the probability of accepting a
false H0. Type I error is also called the level of significance and is denoted by the symbol a.
Alternatively, level of confidence is given as (1-a). Type II error is symbolized using the Greek letter
b. The probability of rejecting a false H0 is called power (1-b). Type I error is called as false positive
while Type II error is called as false negative.

Table 16.1 Summary of the relationships between statistical outcome and statistical errors
Statistical outcome Real results
(decision)
Null hypothesis is true Null hypothesis is false

Non rejection of the Correct decision Type II (b) error

null hypothesis
Rejection of the null Type I (a) error Correct decision
hypothesis/ acceptance of
alternative hypothesis

One-tailed and two-tailed tests

If there are two possible statistical outcomes in the study, then a two tailed test must be used.
Conversely, if there is only one outcome of interest to the investigator, the test statistic must be
interpreted using a one tailed outcome.
The decision as to whether the calculated test statistic should be evaluated as a one tailed test
or a two tailed test is crucial. An incorrect choice may result in the incorrect interpretation of the
statistical analysis because a significant difference between treatments may be declared insignificant
and vice versa.

Defining the critical region for statistical test

Assuming a = 0.05, the critical region of the standardised normal distribution may be given
in three ways. In all three cases, the chosen level of significance is 0.05, as denoted by shaded region.
Calculated values of the z statistic (the test statistic that relates to the standardised normal
distribution) that reside within these regions allow the analyst to reject the null hypothesis.
In a two tailed test, the rejection region is equally divided into two sections, each relating to a
single tail that resides at either extreme of the probability distribution. As a = 0.05, each tail occupies
2.5% of the distribution. In the case of the standardised normal distribution, the critical regions are
denoted by z values of either ³ + 1.96 or £ - 1.96. Conversely, in the one tailed test, the rejection region
is distributed at either tail of the distribution.
 262 Textbook of Computer Applications and Biostatistics

-3 -2 -1 0 +1 +2 +3
z statistic
Figure 16.1 Standardised normal distribution showing the (shaded) region of the z statistic for a two-
tailed test.
If the alternative hypothesis states, that ‘the mean of a treatment group is greater than the
mean of another group’, then the critical value of z is ³ + 1.65 and test is referred to as a positive one
tailed test.
When the alternative hypothesis states, that the mean of a treatment group is lower than the
mean of another group, the rejection region resides at the left-hand (negative) side of standardised
normal distribution and is defined by z values that are £ - 1.65.

-3 -2 -1 0 +1 +2 +3 -3 -2 -1 0 +1 +2 +3
z statistic z statistic
Positive one tailed test Negative one tailed test
Figure 16.2 Standardised normal distribution showing the (shaded) region of the z statistic for a one-
tailed test.

Key stages in hypothesis testing

1. State the null hypothesis.
e.g. m1 = m2 ; no difference in means of two groups.
2. State the alternative hypothesis.
e.g. m1 ¹ m2 ; there is difference in means of two groups.
3. Select the level of significance
e.g. a = 0.05 or 0.01 or 0.001. Usually 0.05 is taken as level of significance. We have to
set the level of significance before start of the study.
 Hypothesis Testing 263

4. Select the number of tails

e.g. one tailed m1 < m2 ; the mean of group 2 is more than mean of group 1.
m1 > m2 ; the mean of group 1 is more than mean of group 2.
two tailed m1 ¹ m2 ; the mean of group 1 and group 2 are not equal but we
don’t know which is greater.
5. Test the statistics
Determine p value by suitable statistical test.
6. Compare table and observed value
p value calculated is compared with table value (critical value).
7. Decide whether to accept or reject null hypothesis
Based on the comparison, either reject or accept the null hypothesis. If the obtained p value is
less than critical value, reject null hypothesis or otherwise accept it.

Summary
Hypothesis testing
It is the process of inferring from a sample whether or not to accept a certain statement about
a population (s).
Key stages in hypothesis testing
1. State the null hypothesis
2. State the alternative hypothesis
3. Select the level of significance
4. Select number of tails
5. Test the statistics
6. Compare table and observed value
7. Decide whether to accept or reject null hypothesis
Decision rule
1. Determine p value.
2. Compare it with the critical value, usually 0.05.
3. If the obtained p value is less than critical value, reject null hypothesis; otherwise accept it.
Types of error
Type I error is the probability of rejecting a true null hypothesis
Type II error is the probability of accepting a false H0.

Multiple choice questions

1. Hypothesis testing is the process of inferring from a ______ whether or not to accept a certain
statement about a________.
a. sample, population b. population, sample
 264 Textbook of Computer Applications and Biostatistics

c. sample, sample d. none of above

2. Which of the following statements sounds like a null hypothesis?
a. The coin is not fair b. There is a correlation in the population
c. There is no difference between male and female incomes in the population
d. The defendant is guilty
3. Which of the following is the researcher usually interested in supporting when he or she is engaged
in hypothesis testing?
a. The alternative hypothesis b. The null hypothesis
c. Both the alternative and null hypothesis d. Neither the alternative or null hypothesis
4. _________ are the values that mark the boundaries of the confidence interval.
a. Confidence intervals b. Confidence limits
c. Levels of confidence d. Margin of error
5. _____ results if we fail to reject the null hypothesis when the null hypothesis is actually false.
a. Type I error b. Type II error
c. Type III error d. Type IV error
6. Identify which of the following steps would not be included in hypothesis testing.
a. State the null and alternative hypotheses
b. Set the significance level before the research study
c. Eliminate all outliers
d. Obtain the probability value using a computer program such as Excel.
7. When the researcher rejects a true null hypothesis, a ____ error occurs.
a. Type I b. TypeA
c. Type II d. Type B
8. If there is only one outcome of interest, then the test statistic is interpreted using ________test.
a. two tailed b. null hypothesis
c. alternative hypothesis d. one tailed
9. AType I error is also known as a ______.
a. false positive b. false negative
c. double negative d. positive negative
10.AType II error is also known as a ______.
a. false positive b. false negative
c. double negative d. positive negative

Exercise
1. Write the alternative hypothesis for each of the following null hypotheses:
a. mA = mB
b. mH ³ mI.
 Hypothesis Testing 265

c. ml = m2 = m3 = m4 = m5 = m6
d. m A £ mB
e. m =115
f. populations C, D, E, F and G are the same
g. both samples come from the same population
2. What is hypothesis testing?
3. Give key stages in hypothesis testing.
4. Give various types of errors in hypothesis testing.
5. What is one tailed and two tailed test?

Answers:
Multiple Choice Questions
1. a 2. c 3. a 4. b 5. b 6. c 7. a 8. b 9. a 10. b
 Chapter 17
CHOICE OF STATISTICAL TESTS

Learning objectives
When we have finished this chapter, we should be able to:
1. Choose appropriate statistical test for given type of data.
2. Understand parametric and non parametric statistical tests.

One of the major steps in the process of statistical hypothesis testing involves the choice of
statistical test. The most appropriate statistical test is chosen according to the desired power of the
study, the nature of the population from which the observations are taken and the nature of
measurement of the variable.

Parametric and non parametric analysis

Statistical tests may be differentiated into two categories, known as parametric and non-
parametric tests. The following assumptions should hold before a parametric statistical method is
selected and used:
1. The samples should be removed from a normally distributed population.
2. The samples should be independent.
3. The variances of the populations under examination should be similar. This is termed as
homoscedasticity.
4. The variable under examination must be metric, discrete or continuous.
Therefore, parametric tests are utilised for the metric data while nominal and ordinal data is
analysed by non-parametric tests.
Let us see statistical tests used according to the type of data.
1. Nominal data
The statistical analysis of nominal data may be performed using a c (Chi square) analysis or
2

a binomial-based test. Characterization of the association within nominal data is performed using the
contingency coefficient.
2. Ordinal data
Many non-parametric tests are referred to as ranking tests and may be employed for the
analysis of ordinal data. Characterization of the association within ordinal data is commonly
performed using the Spearman correlation coefficient.
3. Metric discrete data
Metric discrete data may be analysed using parametric statistical tests. Indeed, if all the
assumptions of parametric statistics are valid, statistical comparisons of groups of interval data

266
 Choice of Statistical Tests 267

should be performed using parametric tests, e.g. t test, analysis of variance, etc.
4. Metric continuous data
Metric continuous data scales may be manipulated using conventional arithmetic and may
therefore be conveniently analysed using either parametric or non-parametic methods.
Characterisation of the association within normally distributed interval or ratio data is commonly
performed using the correlation coefficient.
More specifically, non-parametric methods are exclusively employed to analyse nominal
and ordinal data, whereas parametric and non-parametric methods may be used to examine
continuous metric data. If metric data is skewed then non parametric tests are used or otherwise
parametric tests are choice for metric data.
Chart Showing Choice of Statistical tests
No. of samples Statistical test
Type of Matched/
data to be Unmatched Parametric test Non-parametric test
compared
1 N/A One sample binomial test

2 Unmatched > 20, Chi-square test

Categorical
< 20, Fisher’s Exact test
data

2 Matched McNemar’s test

>2 Unmatched Chi-square test

1 N/A One sample t-test One sample Wilcoxon
for means; One signed rank test
sample Chi-square
test for variances

2 Unmatched Students t-test Man Whitney U test

Metric
data 2 Matched Paired t test Wilcoxon signed rank test

>2 Unmatched One-way ANOVA Kruskal Wallis

for means; Bartlett’s one way ANOVA test
test of homogenity
for variances

>2 Matched Repeated Measures Friedman rank sum test

ANOVA
 268 Textbook of Computer Applications and Biostatistics

Some Commonly Used Hypothesis tests

Parametric tests
1. Two-sample t test
Used to test whether or not the difference between two independent population means is zero
(i.e. the two means are equal). Both variables must be metric and Normally distributed. In addition
the two population standard deviations should be similar, but for larger sample sizes this becomes less
important.
2. Matched-pairs t test
Used to test whether or not the difference between two paired population means is zero.
Both variables must be metric, and the differences between the two must be Normally distributed.
3. One wayANOVA
Used to test whether or not the differences between three or more independent groups means
is zero. The variables should be metric and normally distributed.

Non parametric tests

1. Mann-Whitney U test
Used to test whether or not the difference between two independent population medians is
zero. Variables can be either metric or ordinal. No requirement as to shape of the distributions, but
they need to be similar. This is the non-parametric equivalent of the two-sample t test.
3. Kruskal-Wallis test
Used to test whether the medians of three or more independent groups are the same.
Variables can be either ordinal or metric. Distributions may be of any shape, but all need to be similar.
This non-parametric test is an extension of the Mann-Whitney test.
4. Wilcoxon signed rank test
Used to test whether or not the difference between two paired population medians is zero.
Variables can be either metric or ordinal. Distributions may be of any shape, but the differences
should be distributed symmetrically. This is the non-parametric equivalent of the matched-pairs t test.
5. Chi-squared test (χ2)
Used to test whether the proportions across a number of categories of two or more
independent groups is the same. Variables must be categorical. The chi-squared test is also a test of
the independence of the two variables.
6. Fisher’s Exact test
Used to test whether the proportions in two categories of two independent groups is the
same. Variables must be categorical. This test is an alternative to the 2 × 2 chi-squared test, when cell
sizes are too small .
7. McNemar’s test
Used to test whether the proportions in two categories of two matched groups is the same.
Variables must be categorical.
 Choice of Statistical Tests 269

Summary
Choice of statistical test
Parametric Non-Parametric Frequency
DataType
Ratio, Interval Ordinal Nominal
c Goodness of fit
2
Single sample z-test, t-test Sign test, K-S test
Two independent > 20, Chi-squared
z-test, t-test Mann-Whitney U
samples < 20, Fisher’s Exact
Two independent
Paired t-test Wilcoxon Signed ranks McNemar’s test
paired samples
More than two
One-way ANOVA Kruskall-Wallis One-way ANOVA
independent samples
c Test of
2

Two factors Two-way ANOVA

Independence
Correlation Pearson Spearman Phi

Multiple Choice Questions

1. If metric data is skewed, the choice of statistical test should be__________ .
a. parametric tests b. non parametric tests
c. one sample t test d. z test
2. If data is ordinal or nominal, which of the following test is not used.
a. Non parametric test b. Kruskal Wallis test
c. Parametric test d. Mann Whitney U test
3. If the data is paired but skewed, the following test is used
a. t test b. Paired t test
c. Mann Whitney U test d. Wilcoxon signed rank test
4. If the data is paired and metric, the following test is used
a. t test b. Paired t test
c. Mann Whitney U test d. One sample t test
5. The following test is used to test whether or not the difference between two population means is
zero.
a. t test b. Paired t test
c. Mann Whitney U test d. One sample t test
6. The test for knowing association between two metric variable is _________.
a. Pearson correlation b. Spearman correlation
c. Phi correlation d. None of above
 270 Textbook of Computer Applications and Biostatistics

7. The association between two ordinal variables is tested by ______ .

a. Pearson correlation b. Spearman correlation
c. Phi correlation d. None of above
8. The statistical test used to test the difference in medians of three independent groups is _____.
a. One wayANOVA b. Kruskal-Wallis test
c. Wilcoxon signed rank test d. Friedman rank test
9. The statistical analysis of nominal data is performed by _______.
a. t test b. z test
c. Chi-square d.ANOVA
10. When the variances of two groups of population are similar, it is called as ________.
a. homogeneous b. homoscedasticity
c. heterogeneous d. heteroscedasticity

Exercise
1. Give the choice of statistical tests for following
a. If data is metric, continuous and three groups are to be compared.
b. If data is nominal and paired and two groups are to compared.
c. If data is ordinal and more than three groups are to be compared.
d. If data is ordinal, non parametric and paired.
e. If data is metric, non-parametric and two groups are to be compared.
2. How will you choose statistical test according to type of data.
3. Give the assumptions of parametric tests.
4. Give the non parametric equivalents of following parametric tests.
a. Paired t-test b. One wayANOVA
c. t-test d. Pearson correlation
5. If more than 2 groups are to be compared which tests can be used? Explain.

Answers:
Multiple Choice Questions
1. b 2. c 3. d 4. a 5. a 6. a 7. b 8. b 9. c 10. b
 Chapter 18
HYPOTHESIS TESTING FOR ONE SAMPLE MEAN

Learning objectives
When we have finished this chapter, we should be able to:
1. Understand and estimate z statistic for one sample z test.
2. Understand and estimate t statistic for one sample t test.

One sample testing involves the estimation of whether sample data, generated from an
experimental procedure, are derived from a defined population or not. More specifically, one sample
tests evaluate whether the mean of sample and the mean of the population are different.
Two parametric one sample tests are described here, one sample z test and one sample t test.
The non parametric tests, the chi-square one sample test and the Kolmogorov-Smirnov one sample
test are used but are out of scope of this book.

1. One Sample z test

One sample z test are employed for the analysis of one sample hypothesis when the sample
size is large and, in addition, there is sufficient knowledge about the properties of the population with
which the experimentally derived sample is being compared. Typically, z tests (and indeed t tests)
compare whether the mean of the sample differs from that of a defined population.
The z statistics is calculated using the following formula

X - m0
z= ...1
s/ N
Where
`X = observed sample mean
m0 = hypothesised mean
s = standard deviation of the population
N = sample size

Example 18.1
A pharmaceutical company has purchased a new liquid filling machine for use in its liquid
oral division. The liquid filling properties of existing machine linked to the current manufacturing
process are 5.04 ± 0.27 per bottle. A pilot study has been engaged to evaluate whether (or not) the
filling performance of the new machine is similar to that of previous machine using identical process

271
 272 Textbook of Computer Applications and Biostatistics

conditions. Therefore, 100 vials of the product were removed from a pilot and the mean fill volume
was determined gravimetrically to be 5.13 ml. Is there a difference between the performance of new
and existing filling machines?
Solution: As the mean filling volume of one machine is given and the number of samples are more
than 30, one sample z test is employed here. The following steps gives the systematic approach of
hypothesis testing.
1. State the null hypothesis
The null hypothesis (H0) states that there is no significant difference between the sample
mean and the population mean.
H0: m = 5.04 ml
2. State the alternative hypothesis
The alternative hypothesis states that there is a difference between the sample mean and the
population mean.
Ha: m ¹ 5.04 ml
3. State the level of significance
The choice of the level of significance (a) is an extremely important consideration in the
establishment of the experimental design. Traditionally, a, the probability of rejecting the null
hypothesis when it is in fact true, is chosen to be 0.05.
4. Sate the number of tails
The number of tails associated with the alternative hypothesis is determined before data
collection. As alternative hypothesis states m ¹ 5.04 ml, it may be less or more than 5.04 ml and the
outcome may be two tailed.
5. Select the most appropriate statistical test
A two tailed parametric, one sample test may be applied to the statistical question. As the
sample size is large, the z test is the most suitable statistical method.
6. Perform the statistical analysis
Step 1 Calculate the test statistic
Formula:
The z statistic is calculated by using the following formula
X - m0
z=
s/ N
Data:
`X = observed sample mean = 5.13
m0 = hypothesised mean= 5.04
s = standard deviation of the population =0.27
N = sample size =100
 Hypothesis Testing for one sample mean 273

Solution:
5.13 - 5.04 0.09
z= = = 3.33
0.27 / 100 0.027
Step 2 Define the critical z statistic
Identification of critical statistic requires knowledge of the chosen level of significance (a)
and the number of tails associated with the experimental design. Therefore, the critical z statistics
associated with the specified level of significance (0.05) and a two tailed design may be derived from
the standardized normal distribution as 1.96. The regions of the z distribution associated with the null
and alternative hypothesis may now be defined in terms of the z statistic as:
H0: - 1.96 < z < + 1.96
Ha: z ³+ 1.96 or z ³ -1.96
7. Decision
As the calculated z value (+3.33) is greater than +1.96 (i.e. calculated z value does not lie in
the region -1.96 < z < +1.96), the null hypothesis is rejected and the alternative hypothesis is accepted.
It is therefore concluded that there is a significant difference in the performance of the new and
existing filling machine, in terms of fill volumes.

Example 18.2
A pharmaceutical company has purchased a new tablet punching machine for use in its
compression division. Existing tableting machine produces mean of 1000 tablets/min with a standard
deviation of 150. A pilot study has been engaged to evaluate whether (or not) the tableting
performance of the new machine is similar to that of previous machine using identical process
conditions. The mean tableting rate for 50 samples were determined and found to be 1050 tablets/min.
Is there a difference between the performance of new and existing filling machines?
Solution:
As the mean tableting rate of one machine is given and the number of samples are more than
30, one sample z test is employed here. The following steps gives the systematic approach of
hypothesis testing.
1. State the null hypothesis
The null hypothesis (H0) states that there is no significant difference between the sample
mean and the population mean.
H0: m = 1000 tablet/min
2. State the alternative hypothesis
The alternative hypothesis states that there is a difference between the sample mean and the
population mean.
Ha: m ¹ 1000 tablet/min
3. State the level of significance
The choice of the level of significance (a) is an extremely important consideration in the
 274 Textbook of Computer Applications and Biostatistics

establishment of the experimental design. Traditionally, a, the probability of rejecting the null
hypothesis when it is in fact true, is chosen to be 0.05.
4. Sate the number of tails
As alternative hypothesis states m ¹ 1000 tablets/min, it may be less or more than 1000
tablets/minl and the outcome may be two tailed.
5. Select the most appropriate statistical test
A two tailed parametric, one sample test may be applied to the statistical question. As the
sample size is large, the z test is the most suitable statistical method.
6. Perform the statistical analysis
Step 1. Calculate the test statistic
Formula:
The z statistic is calculated by using the following formula
X - m0
z=
s/ N
Data:
`X = observed sample mean = 1050
m0 = hypothesised mean= 1000
s = standard deviation of the population = 150
N = sample size =50
Solution:
1050 - 1000 50
z= = = 2.36
150 / 50 21.2

Step 2. Define the critical z statistic

Identification of critical statistic requires knowledge of the chosen level of significance (a)
and the number of tails associated with the experimental design. Therefore, the critical z statistics
associated with the specified level of significance (0.05) and a two tailed design may be derived from
the standardized normal distribution as 1.96. The regions of the z distribution associated with the null
and alternative hypothesis may now be defined in terms of the z statistic as:
H0: - 1.96 < z < + 1.96
Ha: z ³ + 1.96 or z ³ -1.96
7. Decision
As the calculated z value (+2.36) is greater than +1.96, the null hypothesis is rejected and the
alternative hypothesis is accepted. It is therefore concluded that there is a significant difference in
tableting rate of the new and existing tablet machine.
2. One Sample t test
In the use of the z statistic for the interpretation of one sample statistical hypothesis testing, it
 Hypothesis Testing for one sample mean 275

has been assumed that the variance of the population is known. However, in many experiments it is
impossible to obtain knowledge of the relevant population statistics and, accordingly, the only
available estimates of the central tendency and variability of the population are the mean and variance
that are associated with the sample data. Under these circumstances, one sample t test is employed to
calculate the t statistic. The process for calculation of the one sample t test is similar to that for the one
sample z test.
The t statistics is calculated using the following mathematical formula:
X - m0
t ( N -1 ) d f = ...2
s / N

`X = observed sample mean

m0 = hypothesised population mean
s = sample standard deviation
N = sample size
t(N-1)df = t value associated with N-1degrees of freedom.

Example 18. 3
To test a manufacturer claim that his fruit juice contains 60 mg of vitamin C per 100ml, a
quality controller analysts takes six randomly selected samples with the following results: 65, 58, 62,
57, 62, 65. Is the manufacturer’s claim justified (sample mean = 61.5; sample standard deviation
(s)=3.39)?
Answer:
Hypothesis testing of this problem is done by using following steps:
1. State the null hypothesis
The null hypothesis states that there is no difference between the sample mean and
hypothetical mean.
H0: m = 60 mg
2. State the alternative hypothesis
The alternative hypothesis states that there is a difference between the manufacturer’s claim
and observed mean.
Ha :m ¹ 60 mg
3. State the level of significance
It is assumed that the level of significance is 0.05.
4. Sate the number of tails
The number of tails associated with the alternative hypothesis is determined before data
collection. As the sample mean may be either lower or higher than the manufacturers claim, we have
to use two tailed test .
 276 Textbook of Computer Applications and Biostatistics

5. Select the appropriate statistical test

Whenever the population variance is unknown, one sample t test and not one sample z test is
used.Atwo tailed parametric, one sample t test may be applied to the statistical question.
6. Perform the statistical analysis
Step 1. Calculate the t statistics
The t statistic is calculated by using the following formula
Formula:
X - m0
t ( N -1 ) d f =
s / N
Data:
`X = observed sample mean = 61.5
m0 = hypothesised population mean = 60
s = sample standard deviation = 3.391
N = sample size = 6
t(N-1)df = t value associated with N-1degrees of freedom = 5
Solution:
61.5 - 60 1 .5
t5 = = = 1 . 09
3.391/ 6 1 . 38
Step 2. Define the critical ‘t’statistics.
The critical value of ‘t’ statistics for 5 degrees of freedom, two tailed and a = 0.05 is given as
+ 2.57 (Appendix 2). Therefore, the calculated t value should be in the range of - 2.57 < t < + 2.57.
7. Decision
As the calculated t value lies within the region -2.57 < t < +2.57, the null hypothesis is
accepted. So the manufacturer’s claim that his fruit juice contains 60 mg/100ml of vitamin C is
justified.

Example 18. 4
To test a manufacturer claim that his tablet contains 60 mg of drug, 10 tablets were selected
randomly and analysed spectrophotometrically. Is the manufacturer’s claim justified? Analysis
results are given below: (sample mean = 62.7; sample standard deviation (s)=6.255)
53, 56, 57, 60, 61, 64, 67, 68, 70, 71

Solution:
Hypothesis testing of this problem is done by using following steps:
1. State the null hypothesis
The null hypothesis states that there is no difference between the sample mean and
hypothetical mean.
 Hypothesis Testing for one sample mean 277

H0: m = 60 mg
2. State the alternative hypothesis
The alternative hypothesis states that there is a difference between the manufacturer’s claim
and observed mean.
Ha :m ¹ 60 mg
3. State the level of significance
It is assumed that the level of significance is 0.05.
4. Sate the number of tails
The number of tails associated with the alternative hypothesis is determined before data
collection. As the sample mean may be either lower or higher than the manufacturers claim, we have
to use two tailed test .
5. Select the appropriate statistical test
Whenever the population variance is unknown, the one sample t test and not the one sample z
test is used. Atwo tailed parametric, one sample t test may be applied to the statistical question.
6. Perform the statistical analysis
Step 1. Calculate the t statistics
The t statistic is calculated by using the following formula
Formula:
X - m0
t ( N -1 ) d f =
s / N
Data:
`X = observed sample mean = 62.7
m0 = hypothesised population mean = 60
s = sample standard deviation = 6.255
N = sample size = 10
t(N-1)df = t value associated with N-1degrees of freedom = 9
Solution:

62.7 - 60 2 .7
t5 = = = 1 .36
6.255/ 10 1 .978
Step 2. Define the critical ‘t’statistics.
The critical value of ‘t’ statistics for 9 degrees of freedom, two tailed and a = 0.05 is given as
±2.26 (Appendix II). Therefore , the calculated t value should be in the range of - 2.26 < t < + 2.26.
7. Decision
As the calculated t value lies within the region -2.26 < t < +2.26, the null hypothesis is
accepted. So the manufacturer’s claim that his tablet contains 60mg drug is justified.
 278 Textbook of Computer Applications and Biostatistics

Summary
One sample z test
Used to test a single sample mean (`X ) when the population mean (m) and variance (s) is
known.
X - m0
z=
s/ N
One sample t test
Used to test a single sample mean (`X ) when the population variance (s) is unknown or n <
30. X - m0
t (N-1)df =
s/ N

Multiple choice questions

1. One sample z test is used to test a single sample when the ______ is known.
a. population variance b. sample variance
c. a and b d. none of above
2. One sample t test is used to test a single sample when the ______ is unknown.
a. population variance b. sample variance
c. a and b d. none of above
3. If we are testing one sample and sample size is more than 30, which of the following test can be
used?
a. two sample z test b. one sample t test
c. one sample z test d. two sample t test
4. df is given by N-1. df is _______ .
a. degree of freedom b. degree of fame
c. dance of freedom d. degree of frame
5. The non parametric test for one sample is __________ .
a. one sample z test b. one sample t test
c. chi square one sample test d. none of above
6. One sample tests evaluate whether the mean of sample and the mean of ______ are different.
a. other sample b. population
c. subjects d. None of above
7. One sample t test is for ______.
a. Parametric data b. non parametric data
c. nominal data d. ordinal data
8. If the calculated value of z lies in the critical region then null hypothesis is_____.
a. rejected b. accepted
 Hypothesis Testing for one sample mean 279

c. no relation d. None of above

9. If the calculated value of ‘t’statistic lies in the critical region, the alternative hypothesis is _______.
a. rejected b. accepted
c. no relation d. None of above
10. If the sample size is 20, the test used for testing one sample mean is _______.
a. one sample z test b. two sample z test
c. two sample t test d. one sample t test

Exercise
1. During the manufacturer of a 1 L infusion, 50 bottles were drawn at random and the volume of each
container measured to check whether the full volume met the 1 liter claim. The mean volume was
found to be 1008 ml and standard deviation was 65 ml. Is there sufficient evidence to suggest that the
average volume of the infusion was not 1 liter at 5% level of significance?

2. A random sample of 400 items gives the mean 4.45 with a standard deviation of 2. Can it be
regarded as drawn from a normal population with mean 4 at 5% level of significance?

3. A randomly 200 medical shops were taken from Satara district and the average sale per shop was
found to be 520. Population mean sale was 510 per shop with a standard deviation of 40. Is the
difference between sample mean and population mean statistically significant at 5% level of
significance?

4. A random sample of 45 items gives the mean 73.2 with a standard deviation of 8.6. Can it be
regarded as drawn from a normal population with mean 76.7 at 1% level of significance?

5. A random sample of 900 children was found to have a mean fatfold thickness at triceps of 3.4 mm
with SD of 2.3mm. Can it be reasonably regarded as a representative sample of population having a
mean thickness of 3.2mm at 5% level of significance?

6. A batch of 40 L of paracetamol suspension (30 mg/ml) was filled into containers. 25 containers of
product have been removed for analysis of their drug content. Does the mean concentration of drug in
the batch conform to the 30 mg/ml at the 0.05 level of significance.
Concentration of paracetamol (mg/ml) in 25 containers:
31.5 30.5 30.5 29.8 30.1 30.2 30.2 30.6 29.8 31.1
30.1 30.1 30.1 30.2 29.7 28.9 30 30.4 30.3 29.9
32.1 30.1 28.9 30.8 29.5

7. A pharmaceutical manufacturer does a chemical analysis to check the potency of products. The
 280 Textbook of Computer Applications and Biostatistics

standard release potency for cephalothin crystals is 910 and the manufacturer believes this claim may
be too high. An assay of 16 lots gives the following potency data:
Data: 897 918 914 906 913 895 906 893 916 908
918 906 905 907 921 901
Test the manufacturer's claim at the 0.01 level of significance.
8. Ten individuals are chosen at random from a normal population and their weights in kg are found to
be
68 63 66 69 63 67 70 70 71 71
Does this sample adequately represent a population in which the mean weight was found to be 66 kgs?
9. Mean Hb % of the population is 14.3. Can a sample of 15 individuals with a mean of 13.5 and SD
1.5 be from same population?
10. Ten tablets are chosen at random from batch and their content in mg are found to be
47 51 48 49 48 52 47 49 46 47
Manufacturer claims that tablet contains 50 mg of drug. Test the manufacturers claim at the
0.05 level of significance.

Answers:
Multiple Choice Questions
1. a 2. a 3. c 4. a 5. c 6. b 7. a 8. b 9. a 10. d

Exercise
1. z=0.87; Accept null hypothesis because observed z value is between -1.96 and +1.96 for two tailed
test at 0.05% level of significance. It is therefore concluded that the average volume of container was
1 liter.

2. z= 4.5; Accept alternative hypothesis because observed z value is more than +1.96 for two tailed
test at 0.05% level of significance. It is therefore concluded that the sample has not been drawn from a
population with mean 4.

3. z= 3.53; Accept alternative hypothesis because observed z value is more than +1.96 for two tailed
test at 0.05% level of significance. It is therefore concluded that the sample mean and population
mean differ significantly.

4. z= - 2.73; Accept alternative hypothesis because observed z value is more than -2.58 for two tailed
test at 0.01% level of significance. It is therefore concluded that the sample has not been drawn from a
population with mean 76.7.

5. z= 2.61; Accept alternative hypothesis because observed z value is more than +1.96 for two tailed
 Hypothesis Testing for one sample mean 281

test at 0.05% level of significance. It is therefore concluded that the sample is not representative of
population having mean thickness of 3.2 mm at 5% level of significance.

6. Initially determine sample mean (30.216) and standard deviation (0.691). As the population
variance is unknown and also sample size is less than 30 use one sample t test. The critical value of ‘t’
statistics for 24 degrees of freedom at 5% level of significance is given as 2.064. Observed t value is
1.52 which is less than t critical. So it can be concluded that mean concentration of drug in the batch
conform to the 30 mg/ml at the 0.05 level of significance.

7. Initially determine sample mean (907.75) and standard deviation (8.48). As the population
variance is unknown and also sample size is less than 30 use one sample t test. The critical value of ‘t’
statistics for 15 degrees of freedom and two tail at 1% level of significance is given as 2.95. Observed t
value is -1.06 which is in between - 2.95 and +2.95. It can be concluded that sample mean for potency
of cephalothin crystals does not differ significantly with manufacturers claim. So, the manufacturer
claim is false about high potency of cephalothin crystals.

8. Initially determine sample mean (67.8) and standard deviation (3.01). As the population variance is
unknown and also sample size is less than 30 use one sample t test. The critical value of ‘t’statistics for
9 degrees of freedom and two tail at 0.05 level of significance is given as 2.26. Observed t value is
1.89 which is in between - 2.26 and +2.26. It can be concluded that sample adequately represent a
population in which the mean weight was found to be 66 kgs.

9. As the population variance is unknown and also sample size is less than 30 use one sample t test.
The critical value of ‘t’ statistics for 14 degrees of freedom and two tail at 0.05 level of significance is
given as 2.14. Observed t value is -2.58 which is not between - 2.26 and +2.26. It can be concluded
that sample of 15 individuals with a mean of 13.5 and SD 1.5 is not from same population.

10. Initially determine sample mean (48.4) and standard deviation (1.9). As the population variance is
unknown and also sample size is less than 30, use one sample t test. The critical value of ‘t’ statistics
for 9 degrees of freedom and two tail at 0.05 level of significance is given as 2.26. Observed t value is
-2.67 which is not in the range of -2.26 and +2.26. Reject null hypothesis and accept alternative
hypothesis.
 Chapter 19
HYPOTHESIS TESTING FOR TWO SAMPLE MEANS

Learning objectives
When we have finished this chapter, we should be able to;
1. Understand and estimate the z statistics for two independent data sets.
2. Understand and perform the t statistics for two independent data sets.
3. Understand and perform paired t test for matched data sets.

One of the most common statistical testing used in the pharmaceuticals is the examination of
differences between two sets of sample data, i.e. whether or not the two populations whose properties
are estimated by the sample statistics differ from one another.
There are two types of two sample statistical tests; independent and paired, as shown in
following table:
Type Parametric test Non-parametric test

Independent samples z test (Sample > 30) Chi-square test (sample > 20)
t test (sample < 30) Fisher exact test (sample < 20)
Mann Whitney U test (Metric)
Paired samples Paired t test McNemar’s test (Categorical)
Wilcoxon signed rank test (Metric)

In the two sample test, the null hypothesis specifically states that there is no difference
between the mean values of each set of data.

1. The z statistics for two independent samples

The z statistic can be calculated by using following formula

X1 - X2
z= ...1
SE (X1 - X2 )
Where
`X1 & `X2 = sample means of two independent data sets
SE (`X1 -`X2) is the standard error of difference between two means
The standard error of difference of two means can be calculated as

282
 Hypothesis Testing for two sample means 283

s 12 s 22
SE(X1 - X 2 ) = + ...2
N1 N 2
Where
s1, s2 = population standard deviations of samples N1 and N2 respectively

The following steps are used for calculating z statistic for two sample means:
1. Calculate the mean `X1 and `X2 and standard deviation (s1) and (s2) of each population.
2. Calculate the standard error of difference between two means SE(`X1 -`X2 ).
3. Calculate z statistics using formula given above.
4. If the observed difference between the two means is greater than 1.96 times the standard error of
difference, it is significant at 5% level of significance.
5. If the observed difference is greater than 3 times the SE, it is real variability in more than 99% cases,
and biological or due to chance in less than 1% cases

Example 19.1
To determine whether two drugs affected human mental concentration equally, 50 students
were given one drug and 50 others the second drug.All the students were then given an examination to
measure their mental concentration index. The mean scores for the two groups were 65 and 70, and
the respective standard deviations were 15 and 18. Is there sufficient evidence to suggest that the
drugs affected mental concentration differently?
Solution: Let us test the difference between two groups systematically.
1. State the null hypothesis
The null hypothesis states that there is no difference between the effect of two drugs on
mental concentration.
H0: m1 = m2
2. State the alternative hypothesis
The alternative hypothesis states that there is a difference between the effect of two drugs on
mental concentration.
Ha :m1 ¹ m2
3. State the level of significance
It is assumed that the level of significance is 0.05.
4. Sate the number of tails
The number of tails associated with the alternative hypothesis is determined before data
collection. As the effect produced by two drugs on mental concentration may differ to any side, we
have to use two tailed test .
 284 Textbook of Computer Applications and Biostatistics

5. Select the appropriate statistical test

The most relevant statistical method to assess the validity of the null hypothesis, is a two-
tailed z test for two independent samples. As the sample size is more than 30, the z test is the most
suitable statistical method.
6. Perform the statistical analysis
Step 1. Calculate z statistic
The following steps are used for calculating z statistic
1. Calculate the mean `X1 and `X2 and standard deviation (s1) and (s2) for each population.
Here they have given.
`X1 = 65 s1= 15 `X2 = 70 s2=18
2. Calculate the standard error of difference between two means SE(`X1 -`X2 ).

s 12 s 22 152 182
SE(X1 - X 2 ) = + = + = 10.98 = 3.31
N1 N 2 50 50
3. Calculate z statistics using formula given above.

X1 - X 2 65 - 70 -5
z= = = = - 1 . 51
SE ( X 1 - X 2 ) 3 . 31 3 . 31

Step 2. Find critical value of z statistics

The critical value of z statistics for two tailed test and at 0.05 level of significance, is given as
±1.96. Therefore it should lie in critical region -1.96 < z < + 1.96 (Appendix I).
7. Decision
The observed z value (-1.51) is less than z critical (-1.96). It lies in critical region. Therefore,
there is no reason to reject H0. The evidence would suggest that both drugs have the same effect on
mental concentration.

Example 19.2
In a group of 196 adults the mean serum cholesterol was 180 mg/dl with a standard deviation
of 42 mg/dl. In a comparable group of 144 adults, the mean serum cholesterol was 150 mg/dl with a
standard deviation of 48 mg/dl. Is the difference in cholesterol level of the two classes statistically
significant? Solution: Let us test the difference between two groups systematically.
1. State the null hypothesis
The null hypothesis states that there is no difference between the cholesterol level of the two
groups.
H0: m1 = m2
2. State the alternative hypothesis
The alternative hypothesis states that there is a difference between the cholesterol levels of
 Hypothesis Testing for two sample means 285

two classes.
Ha :m1 ¹ m2
3. State the level of significance
It is assumed that the level of significance is 0.05.
4. Sate the number of tails
The number of tails associated with the alternative hypothesis is determined before data
collection.As the cholesterol levels may differ to any side, we have to use two tailed test.
5. Select the appropriate statistical test
The most relevant statistical method to assess the validity of the null hypothesis, is a two-
tailed z test for two independent samples. As the sample size is more than 30, the z test is the most
suitable statistical method.
6. Perform the statistical analysis
Step 1. Calculate z statistic
The following steps are used for calculating z statistic
1. Calculate the mean `X1 and `X2 and standard deviation (s1) and (s2) for each population.
The data is already provided.
`X1 = 180 s1= 42 `X2 = 150 s2=48
2. Calculate the standard error of difference between two means SE(`X1 -`X2 ).

s 12 s 22 42 2 48 2
SE( X 1 - X 2 ) = + = + =5
N1 N 2 196 144

3. Calculate z statistics using formula given above.

X1 - X 2 180 - 150 30
z= = = =6
SE ( X 1 - X 2 ) 5 5

Step 2. Find critical value of z statistics

The critical value of z statistics for two tailed test and at 0.05 level of significance, is given as
±1.96 (Appendix I). Therefore it should lie in the critical region -1.96< z < +1.96.

7. Decision
As observed z value (6) is more than z critical (1.96), reject null hypothesis and accept
alternative hypothesis. The evidence would suggest that difference in cholesterol level of the two
groups is statistically significant.
2. The t test for two independent samples
In the equation for the calculation of the z statistic the population variance is used, indicating
that there is prior knowledge of this parameter. Conversely, in the calculation of the t statistic the
variances of the populations from which the samples were drawn are estimated from the variances of
 286 Textbook of Computer Applications and Biostatistics

the sample data.

The t statistic can be calculated by using following formula,
(X1 - X 2 )
t= ...3
SE (X1 - X 2 )
Where,
`X1 & `X2 = mean values of the two sets of sample data.
SE (`X1 -`X2) = the standard error of difference between two means.
The standard error of difference of two means can be calculated as,

S 2p S 2p
SE (X1 - X 2 ) = + ...4
N1 N2
Where,
2
Sp is calculated as pooled variance.

( N 1 - 1) s 12 + ( N 2 - 1) s 22
S 2p = ...5
N1 + N 2 - 2
The following steps are used for calculating t statistics:
1. Calculate mean and standard deviation of each group.
2. Calculate the pooled variance from sample size and sample variances using following formula.
( N 1 - 1) s 12 + ( N 2 - 1) s 22
S 2p =
N1 + N 2 - 2
Where,
S1 and S2 are standard deviations of two sample groups.
3. Calculate the standard error of difference between two means by using formula.

S2p S2p
SE (X1 - X2 ) = +
N1 N2
4. Calculate t statistic using formula.
(X 1 - X 2 ) (X 1 - X 2 )
t = =
SE (X 1 - X 2 ) S 2p S 2p
+
N1 N 2
5. Find the degrees of freedom, which is N1+N2 - 2.
6. For given degrees of freedom, find critical value from t table. If the calculated value is less than
critical value, then the null hypothesis is accepted while if calculated t statistic exceeds then the null
hypothesis is rejected.
 Hypothesis Testing for two sample means 287

Example 19.3
Two formulations of same drug was placed for stability testing under controlled storage
o
conditions at 37 C. After 3 months, samples of each formulation were removed and assayed
individually. The analytical results are given in following table. Determine whether there is a
difference in the mean assay of drug in each formulation following the period of storage.
Data:
Formulation 1 104.1, 108.2, 108.6, 100.8, 106.5, 101.0, 102.6, 99.2, 95.2, 100.8
Formulation 2 102.9, 99.6, 98.1, 104.2, 90.2, 101.0, 99.9, 89.5, 95.5, 98.6
Solution: This problem can be solved in following steps:
1. State the null hypothesis
The null hypothesis states that there is no difference between the mean of assay of two
formulations.
H0: m1 = m2
2. State the alternative hypothesis
The alternative hypothesis states that there is a difference between the mean of assay of two
formulation
Ha :m1 ¹ m2
3. State the level of significance
It is assumed that the level of significance is 0.05.
4. Sate the number of tails
The number of tails associated with the alternative hypothesis is determined before data
collection.As there are two possible outcomes to the study, this is a two tailed experimental design.
5. Select the appropriate statistical test
As the population variances are not known and the sample size is less than 30, two sample t
test is the appropriate choice.
6. Perform the statistical analysis
1. Calculate mean and standard deviation of each group.
Calculation of mean
å (X )
X =
N
Mean of formulation 1= `X1 = 102.7
Mean of formulation 2= `X2 = 97.95
Calculation of standard deviation
(å X) 2
å X2 -
SD = N
N -1
Standard deviation of formulation 1 = s1 = 4.21
 288 Textbook of Computer Applications and Biostatistics

Standard deviation of formulation 2 = s2 = 4.91

2. Calculate the pooled standard deviation from sample size and sample SD of two groups using
following formula
( N1 - 1)s1 + ( N 2 - 1)s 2 (9 ´ 4.21) + (9 ´ 4.91)
Sp = = = 4.56
N1 + N 2 - 2 20 - 2

3. Calculate the standard error of difference between two means by using formula

S2p S2p 1 1 1 1
SE ( X1 - X 2 ) = + = Sp + = 4.56 + = 2.04
N1 N2 N1 N 2 10 10
4. Calculate t statistic using formula

(X1 - X2 ) 102.7 - 97.95

t= = = 2.32
SE (X1 - X2 ) 2.04
5. Find the critical t values
The critical t values for 18 degrees of freedom, two tailed design and at 0.05 level of
significance is given as ±2.1 (Appendix II). Therefore the critical region of acceptance is -2.1< t < +
2.1.
5. Decision
The calculated t value more than critical value suggests rejection of null hypothesis.
Therefore it can be concluded that there was significant difference in the assay of drug in two
o
formulations after storage for three months at 37 C.

Example 19.4
Weight in kg of 10 boys and 10 girls aged between 15 to 20 are given below. Determine
whether there is a significant difference in the mean weight of two groups.
Data:
Boys (Wt in kg) 42 46 50 48 50 52 41 49 51 56
Girls (Wt in kg) 38 41 36 35 30 42 31 29 31 35

Solution: This problem can be solved in following steps:

1. State the null hypothesis
The null hypothesis states that there is no difference between the mean weight of two
groups.
H0: m1 = m2
2. State the alternative hypothesis
The alternative hypothesis states that there is a difference between the mean weight of two
 Hypothesis Testing for two sample means 289

groups.
Ha :m1 ¹ m2
3. State the level of significance
It is assumed that the level of significance is 0.05.
4. Sate the number of tails
The number of tails associated with the alternative hypothesis is determined before data
collection.As there are two possible outcomes to the study, this is a two tailed experimental design.
5. Select the appropriate statistical test
As the population variances are not known and the sample size is less than 30, two sample t
test is the appropriate choice.
6. Perform the statistical analysis
1. Calculate mean and standard deviation of each group.
Calculation of mean
å(X )
X=
N
Mean of weight of Boys= `X1 = 48.5
Mean of weight of Girls= `X2 = 34.8
Calculation of standard deviation

(å X) 2
å X2 -
SD = N
N -1

Standard deviation of weight of Boys= s1 = 4.53

Standard deviation of weight of Girls = s2 = 4.56
2. Calculate the pooled standard deviation from sample size and sample SD of two groups using
following formula

( N1 - 1)s1 + ( N 2 - 1)s 2 (9 ´ 4.53) + (9 ´ 4.56)

Sp = = = 4.54
N1 + N 2 - 2 20 - 2
3. Calculate the standard error of difference between two means by using formula

S2p S2p 1 1 1 1
SE (X1 - X 2 ) = + = Sp + = 4.54 + = 2.03
N1 N2 N1 N 2 10 10

4. Calculate t statistic using formula

(X 1 - X 2 ) 48 .5 - 34 .8
t= = = 6 .7
SE ( X 1 - X 2 ) 2 .03
 290 Textbook of Computer Applications and Biostatistics

5. Find the critical t values

The critical t values for 18 degrees of freedom, two tailed design and at 0.05 level of
significance is given as ±2.1 (Appendix II). Therefore calculated value should lie in acceptance
region of -2.1< t < +2.1.
5. Decision
The calculated t value more than critical value suggests rejection of null hypothesis.
Therefore it may be concluded that there was significant difference between the mean weight of two
groups.

3. Paired t test for two matched samples

The paired t test is a parametric statistical method that examines the significance of the mean
of the differences between the pairs of data and a fixed value, determined from the null hypothesis.
It is applied to paired data of independent observations from one sample only when each
individual gives a pair of observations. Testing by this method eliminates individual sampling
variations because the sample is one and the observations on each person in the sample are taken
before and after the experiment.
The following steps are used for calculating paired t statistics.
1. Find the difference in each set of paired observations before and after (X1-X2 = x) and its squares.
2. Calculate the mean of the difference (`x).
3. Calculate SD of differences by using formula
(å X ) 2
å X2 - ...6
SD = N
N -1
4. Determine ‘t” value by using following formula:
x -O x
t= = ...7
SE SD/ N
As per null hypothesis, there should be no real difference in means of two sets of
observations, i.e. theoretically it should be 0.
5. Find the degrees of freedom, which is N-1.
6. For given degrees of freedom, find critical value from t table. If the calculated value is less than
critical value, then the null hypothesis is accepted while if calculated t statistic exceeds then the null
hypothesis is rejected.

Example 19.5
Systolic blood pressure of 9 normal individuals was taken. Then a known hypotensive drug
was given and blood pressure was again recorded. Did the hypotensive drug lowers the systolic blood
pressure?
 Hypothesis Testing for two sample means 291

Data: Blood pressure of nine healthy volunteers before and after injection of hypotensive drug.
BP Before (X1) 122 121 120 115 126 130 120 125 128
BPAfter (X2) 120 118 115 110 122 130 116 124 125
Solution:
Let us test the hypothesis using following steps:
1. State the null hypothesis
The null hypothesis states that there is no difference between the mean of systolic BP of
healthy volunteers before and after injection of hypotensive drug treatment.
H0: m1 before = m2 after
2. State the alternative hypothesis
The alternative hypothesis states that the mean of systolic BP of healthy volunteers before
injection of hypotensive drug treatment is greater than mean of systolic BP after injection of drug.
Ha :m1before > m2 after
3. State the level of significance
It is assumed that the level of significance is 0.05.
4. Sate the number of tails
The number of tails associated with the alternative hypothesis is determined before data
collection. As there is only one possible outcome to the study i.e. injection of drug lowers the blood
pressure, this is a one tailed experimental design.
5. Select the appropriate statistical test
As the given experiment consists of data of independent observations from same sample
giving pair of observations, paired t test is applied.
6. Perform the statistical analysis
1. Find the difference in each set of paired observations before and after (X1-X2 = x) and its squares
BP Before (X1) 122 121 120 115 126 130 120 125 128
BPAfter (X2) 120 118 115 110 122 130 116 124 125
Difference (x) 2 3 5 5 4 0 4 1 3 Sx = 27
Sx = 105
2 2
Squares (x ) 4 9 25 25 16 0 16 1 9
2. Calculate the mean of the difference (`x).
Formula:
åx
x=
N
Data: å x = 27; N=9
Mean of the difference (`x) = 27/9 = 3
3. Calculate SD of differences by using formula
 292 Textbook of Computer Applications and Biostatistics

Formula:
(å X) 2
å X2 -
SD = N
N -1
Data: Sx2 = 105; Sx = 27; N = 9

( å X) 2 (27) 2
å X2 - 105 -
SD = N = 9 = 24 = 1.73
N -1 9 -1 8
SD of differences = 1.73
4. Determine ‘t” value by using following formula:
x x 3 3
t= = = = = 5.17
SE SD/ N 1.73 / 9 0.58

5. Find critical t value

For 8 degrees of freedom and at level of significance of 0.05 one tailed experiment value
from appendix II gives ± 1.86. Therefore, calculated t value should be in critical region -1.86 < t < +
1.86.

7. Decision
The calculated t value is more than critical value and therefore it may be concluded that there
was significant difference in blood pressure before and after administration of hypotensive drug.
Reject the null hypothesis and accept the alternative hypothesis i.e. the mean of systolic BP of healthy
volunteers before injection of hypotensive drug treatment is greater than mean of systolic BP after
injection of drug.

Example 19.6
Serum digoxin levels were determined for nine healthy males following rapid intravenous
injection of the drug. The measurements were made 4h after the injection and again at the end of an 8h
period. Is the difference in the serum digoxin concentration at the end of 4h and at the end of 8h
statistically significant?
Data: Serum digoxin concentration mg/ml after 4h and 8h.
After 4h (X1) 1.0 1.3 0.9 1.0 1.0 0.9 1.3 1.1 1.0
After 8h (X2) 1.0 1.3 0.7 1.0 0.9 0.8 1.2 1.0 1.0

Answer: Let us test the hypothesis using following steps:

1. State the null hypothesis
The null hypothesis states that there is no difference between the mean serum concentration
of digoxin after 4h and 8h.
 Hypothesis Testing for two sample means 293

H0: m1 4h = m2 8h
2. State the alternative hypothesis
The alternative hypothesis states that the mean serum concentration of digoxin after 4h is
greater than mean serum concentration of digoxin after 8h .
Ha :m1 4h > m2 8h
3. State the level of significance
It is assumed that the level of significance is 0.05.
4. Sate the number of tails
The number of tails associated with the alternative hypothesis is determined before data
collection.As there is only one possible outcome to the study, this is one tailed experiment.
5. Select the appropriate statistical test
As the given experiment consists of data of independent observations from same sample
giving pair of observations, paired t test is applied.
6. Perform the statistical analysis
1. Find the difference in each set of paired observations before and after (X1-X2 = x) and its squares
BP Before (X1) 1.0 1.3 0.9 1.0 1.0 0.9 1.3 1.1 1.0
BPAfter (X2) 1.0 1.3 0.7 1.0 0.9 0.8 1.2 1.0 1.0
Difference (x) 0 0 0.2 0 0.1 0.1 0.1 0.1 0 Sx = 0.6
2
Squares (x ) 0 0 0.04 0 0.01 0.01 0.01 0.01 0 Sx2 = 0.08
2. Calculate the mean of the difference (`x).
Formula:
åx
x=
N
Data: å x = 0.6; N=9
Mean of the difference (`x) = 0.6/9 = 0.067

3. Calculate SD of differences by using formula

Formula:
( å X) 2
å X2 -
SD = N
N -1

Sx = 0.08; Sx = 0.6;
2
Data: N=9
(å X ) 2 ( 0 .6 ) 2
å X2 - 0.08 -
N 9 0.04
SD = = = = 0.07
N -1 9 -1 8
SD of differences = 0.07
 294 Textbook of Computer Applications and Biostatistics

4. Determine ‘t” value by using following formula:

x x 0 .0 6 7 0 .0 6 7
t = = = = = 2 .9 1
SE SD/ N 0 .0 7 / 9 0 .0 2 3

5. Find critical t value

For 8 degrees of freedom and at level of significance of 0.05, one tailed experiment value of t
is given in appendix II as ±1.86. Therefore, calculated t values should be within -1.86 <t<+1.86.
7. Decision
The calculated t value is more than critical value and therefore it may be concluded that there
was significant difference in mean serum digoxin concentration at the end of 4h and 8h. Reject the
null hypothesis and accept the alternative hypothesis i.e. mean serum concentration of digoxin after
4h is greater than mean serum concentration of digoxin after 8h

Use of Microsoft Excel in Hypothesis Testing of Two Sample Means

1. Performing z test for two independent samples using Excel
Excel Solution: The systematic steps for calculating z test are given below:
Step I
1. Open new MS-Excel file from MS-office.
2. Enter data into Sheet1.
3. In first row put the labels for variables (i.e. formulation 1 in A1 and formulation 2 in B1
cell).
4. Enter data for formulation 1 in column ‘A’(FromA2 cell).
5. Enter data for formulation 2 into column ‘B’(From B2 cell).
Step II
1. After entering data into the Sheet1, select Tools/Data Analysis/z-Test: Two Sample for
means.
a. Select Tools menu from the Menu bar of MS-Excel, Instantly, it will display pull down
menus.
b. Then, click on the DataAnalysis option from pull down menus.
c. When the Data Analysis dialog box appears: Choose z-Test: Two Sample for means and
click on OK button.
d. ‘z-Test: Two Sample for means’ dialog box will appear as shown below.
 Hypothesis Testing for two sample means 295

Figure 19.1 Window of z-Test: Two samples for means

2. Variable 1 Range - Enter the cell reference for the first range of data we want to analyze.
The range must consist of a single column or row of data.
3. Variable 2 Range - Enter the cell reference for the second range of data we want to analyze.
The range must consist of a single column or row of data.
4. Hypothesized Mean Difference - Enter the number we want for the shift in sample means.
Avalue of 0 (zero) indicates that the sample means are hypothesized to be equal.
5. Variable 1 Variance (known) - Enter the known population variance for the Variable 1
input range.
6. Variable 2 Variance (known) - Enter the known population variance for the Variable 2
input range.
7. Labels - Select if the first row or column of our input ranges contains labels. Clear this
check box if our input ranges have no labels; Microsoft Excel generates appropriate data
labels for the output table.
9.Alpha - Enter the confidence level for the test. Generally it is 0.05.
10. Leave the other items at their default selections. Click OK.
This will give the result of z test for two independent samples.

2. Performing t test for two independent samples using Excel

Example
Two formulations of same drug was placed for stability testing under controlled storage
o
conditions at 37 C. After 3 months, samples of each formulation were removed and assayed
individually. The analytical results are given in following table. Determine whether there is a
difference in the mean assay of drug in each formulation following the period of storage.
 296 Textbook of Computer Applications and Biostatistics

Data:
Formulation 1 104.1, 108.2, 108.6, 100.8, 106.5, 101.0, 102.6, 99.2, 95.2, 100.8
Formulation 2 102.9, 99.6, 98.1, 104.2, 90.2, 101.0, 99.9, 89.5, 95.5, 98.6

Excel Solution:
Step I
1. Open new MS-Excel file from MS-office.
2. Enter data into Sheet1.
3 . In first row put the labels for variables (i.e. inA1 and B1 cell ).
4. Enter data for variable 1 i.e formulation 1 in column ‘A’
5. Enter data for variable 2 i.e. formulation 2 into column ‘B’.

Figure 19.2 Worksheet after data entry

Step II
1. After entering data into the Sheet 1 Select: Tools/ Data Analysis/ t-Test: Two Sample
assuming Unequal Variances.
2. Then click on OK button.
 Hypothesis Testing for two sample means 297

Figure 19.3 Window of Data Analysis

Step III
1. For the Input Range for Variable 1, SelectA1:A11 (Label along with values).
2. For the input range for Variable 2, Select B1:B11 (Label along with values).
3. Click on Labels. Leave the other items at their default selections.
This dialog box is shown below. Click on OK button.

Figure 19.4 Window of t-Test: Two samples assuming unequal variances

Step IV
1. The following output is created in Sheet 4
 298 Textbook of Computer Applications and Biostatistics

t-Test: Two-Sample Assuming Unequal Variances

Formulation 1 Formulation 2
Mean 102.7 97.95
Variance 17.78666667 24.145
Observations 10 10
Hypothesized Mean Difference 0
df 18
t Stat 2.319650459
P(T<=t) one-tail 0.016157567
t Critical one-tail 1.734063592
P(T<=t) two-tail 0.032315134
t Critical two-tail 2.100922037

Note: Always assume unequal variance.

Interpretation of results
Depending on the hypothesis, level of significance and number of tails associated with
design, one can interpret above obtained result for t test.
At 18 degrees of freedom two tail critical value is 2.1 which is less than observed value i.e.
2.32. Therefore, it can be concluded that there is a significant difference in the mean assay of drug in
each formulation following the period of storage.

3. Performing Paired t-test for two sample means using Excel

Example
Systolic blood pressure of 9 normal individuals was taken. Then a known hypotensive drug was
given and blood pressure again recorded. Did the hypotensive drug lowers the systolic blood
pressure?
Data: Blood pressure of nine healthy volunteers before and after injection of hypotensive drug.
BP Before (X1) 122 121 120 115 126 130 120 125 128
BPAfter (X2) 120 118 115 110 122 130 116 124 125

Excel Solution:
Step I
1. Open new MS-Excel file from MS-office.
2. Enter data into Sheet1.
3 . In first row put the labels for variables (i.e. inA1 and B1 cell ).
4. Enter data for variable 1 i.e. ‘BP Before’into column ‘A’.
5. Enter data for variable 2 i.e. ‘BPAfter’into column ‘B’.
 Hypothesis Testing for two sample means 299

Figure 19.5 Data entry

Step II
1. To perform a paired t-test, select Tools/ DataAnalysis / t-test: Paired two sample for means.
2. In the dialog box t-test: Paired two sample for means, we have to provide necessary information.
a. The Input Range for Variable 1, i.e BP Before, is given asA1:A10.
b. The input range for Variable 2, i.e BP Before, is given as B1:B10.
c. If selection includes labels in first row: Tick mark Labels.
d. For now, leave the other items at their default selections.
3. The dialog box is shown below. Click OK button.

Figure 19.6 Window of t-Test: Paired Two Sample for means

Step III
1. The following output is created in Sheet 4.
 300 Textbook of Computer Applications and Biostatistics

t-Test: Paired Two Sample for Means

BP Before BP After
Mean 123 120
Variance 21.75 36.25
Observations 9 9
Pearson Correlation 0.979375
Hypothesized Mean Difference 0
df 8
t Stat 5.196152
P(T<=t) one-tail 0.000413
t Critical one-tail 1.859548
P(T<=t) two-tail 0.000826
t Critical two-tail 2.306004

Interpretation of Results
Depending on the hypothesis, level of significance and number of tails associated with
design, one can interpret above obtained result for paired t test.
At (N-1) = 8 degrees of freedom, 0.05 level of significance critical t value for one tail is 1.86.
The calculated t value (5.19) is more than critical value (1.86) and therefore it may be concluded that
there was significant difference in blood pressure before and after administration of hypotensive
drug. Therefore, reject the null hypothesis and accept the alternative hypothesis that the mean systolic
BP of healthy volunteers decreases after hypotensive drug injection.

Summary
Statistical tests for two samples

Type Parametric test Non-parametric test

Independent samples z test (Sample > 30) Chi-square test (sample > 20)
t test (sample < 30) Fisher exact test (sample < 20)
Mann Whitney U test (Metric)
Paired samples Paired t test McNemar’s test (Categorical)
Wilcoxon signed rank test (Metric)

z test for two independent samples

X1 - X 2 s 12 s 22
z= SE(X1 - X 2 ) = +
SE (X1 - X 2 ) N1 N 2

t test for two independent samples

(X1 - X 2 ) S2p S2p ( N1 - 1)s12 + ( N 2 - 1)s 22

t= SE (X1 - X 2 ) = + S2p =
SE (X1 - X 2 ) N1 N2 N1 + N 2 - 2
 Hypothesis Testing for two sample means 301

Paired t test for two matched samples

(å X) 2
x -O x -O å X2 -
t= = SD = N
SE SD/ N N -1

Multiple Choice Questions

1. Which of the following is used to test significance between two independent samples?
a. z test b. t test
c. z and t test d. None of above
2. For sample less than 30, which of the following is used to test significance between two
independent samples?
a. z test b. t test
c. z and t test d. Paired t test
3. For sample more than 30, which of the following is used to test significance between two
independent samples?
a. z test b. t test
c. z and t test d. Paired t test
4. Which of the following is used to test significance between paired samples?
a. z test b. t test
c. z and t test d. Paired t test
5. For categorical data, non-parametric test equivalent to paired t test is_______.
a. Mann Whitney U test b. Wilcoxon signed rank test
c. Chi-square test d. McNemar’s test
6. In t test, the degree of freedom is given by _______.
a. N -1 b. N1 - N2 + 2
c. N1 + N2 - 2 d. N1 + N2 + 2
7. In paired t test, the degree of freedom is given by ______ .
a. N -1 b. N1 - N2 + 2
c. N1 + N2 - 2 d. N1 + N2 + 2
8.Acritical region defined by z statistic is _______ region for null hypothesis.
a. acceptance b. rejection
c. no relation d. can’t say
9. Non parametric test used for comparing two independent metric data sets is _________ .
a. Chi-square test b. Fisher exact test
c. Mann Whitney U test d. McNemar’s test
10. Non parametric test used for comparing two independent categorical samples (n < 20) is ______ .
a. Chi-square test b. Fisher exact test
 302 Textbook of Computer Applications and Biostatistics

c. Mann Whitney U test d. McNemar’s test

Exercise
1. In a group of 169 boys in the age group of 12-20 years the mean height was 168 cm with a standard
deviation of 14 cm. In a comparable group of 54 girls the mean height was 153 cm with a standard
deviation of 8 cm. Is the height differs with sex?

2. In a group of 100 infants of 8 month age, the mean weight was 6.9 kg with a standard deviation of
1.1 kg. In a comparable group of 169 infants the mean weight was 7.3 kg with a standard deviation of
0.91 kg. Test whether mean weights are significantly different.

3. The mean plasma potassium level for 50 adult males with a disease was found to be 3.35 mEq/l with
a standard deviation of 0.5 mEq/l. The normal 50 adult male value for plasma potassium is 4.6 mEq/l
with a standard deviation of 0.01 mEq/l. Based on above data, can it be concluded that males with
disease have lower plasma potassium levels than normal males?

4. Mean systolic blood pressure of 54 normal adults was 75 with a standard deviation of 6. In 31
diseased adults, mean systolic blood pressure was 69 with a standard deviation of 5. Test whether
mean systolic blood pressure of two groups differ significantly?

5. In a nutritional study, 13 children (group A) were given a usual diet plus vitamin A and D tablets
while the second group (B) of 12 children was taking the usual diet. After 24 months, the gain in
weight in kg was noted as given in table below. Can we say that vitamins A and D were responsible for
this difference.
GroupA 5 3 4 3 2 6 3 2 3 6 7 5 3
Group B 1 3 2 4 2 1 3 4 3 2 2 3

6. In the experiment, there are two groups of 15 subjects. For the first group, each individual was pre-
treated with oral rifampicin (600 mg daily for 10 days). The other group acted as a control, receiving
only placebo pre-treatment. All subjects then received an intravenous injection of theophylline (3
mg/kg). A series of blood samples was obtained after the theophylline injections, and analysed for
drug content. The efficiency of removal of theophylline was reported as a clearance value. Is increase
in clearance of theophylline would be due to the fact that rifampicin may increase the ability of the
liver to eliminate this drug? Clearance of theophylline (ml/min/kg) for control subjects and for those
pre-treated with rifampicin are given below.
Control 0.81 1.06 0.43 0.54 0.68 0.56 0.45 0.88 0.73 0.43
Treated 1.15 1.28 1.00 0.95 1.06 1.15 0.72 0.79 0.67 1.21
Control 0.46 0.43 0.37 0.73 0.93
Treated 0.92 0.67 0.76 0.82 0.82
 Hypothesis Testing for two sample means 303

7. Two groups of hypertensive patients are subjected to two different treatment regimens and systolic
blood pressure was recorded after specific time period. Do the results of systolic BP listed below
indicate a significant difference between the two therapies at 95% confidence level?
Group I 78 87 75 88 91 82 87 65 80
Group II 75 88 93 86 84 71 91 79 81 86 89

8. A new natural product of company reported to promote weight loss. To ensure the claim, a clinical
study of developed tablet was conducted on ten obese volunteers. Initial weight of volunteers before
clinical trial and weight after receiving therapy for 2 months were recorded. Whether the natural
product was clinically successful?
Weight before (kg) 141 124 153 120 116 155 151 155 132 116
Weight after (kg) 130 120 149 120 109 150 151 154 125 110

9. The systolic blood pressures of 12 women between the ages of 20 and 35 were measured before and
after administration of a newly developed oral contraceptive. Is this increase in systolic blood
pressure due to oral contraceptive?
Before 122 126 132 120 142 130 142 137 128 132 128 129
After 127 128 140 119 145 130 148 135 129 137 128 133

10. The extent to which an infant’s health is affected by parental smoking is an important public health
concern. The following data are the urinary concentrations of cotinine (a metabolite of nicotine);
measurements were taken both from a sample of infants who had been exposed to household smoke
and from a sample of unexposed infants. Comment.
Unexposed 8 11 12 14 20 43 111
Exposed 35 56 83 92 128 150 176 208

Answers:
Multiple Choice Questions
1. c 2. b 3. a 4. d 5. d 6. c 7. a 8. a 9. c 10. b

Exercise
1. Yes, height differs with sex. (SE= 1.53; z=9.8)
2. Yes, mean weights are significantly different. (SE=0.13; z=-3.07)
3. It be concluded that males with disease have significantly lower plasma potassium levels than
normal males. (SE=0.07; z=-17.67)
4. Yes, mean systolic blood pressure of two groups differ significantly. (SE=1.21; z= 4.94)
5. The calculated t value (2.74) is more than critical value (At 23 df, t critical for two tailed test at 5%
level of significance is 2.07) and therefore it can be concluded that vitamins A and D were responsible
 304 Textbook of Computer Applications and Biostatistics

for the difference in increase of weight in two groups.

6. The calculated t value (-3.09) is more than critical value (At 28 df, t critical for two tailed test at 5%
level of significance is 2.05) and therefore it can be concluded that increase in clearance of
theophylline may be due to rifampicin increasing the ability of the liver to eliminate this drug.
7. The calculated t value (-0.74) is less than critical value (At 18 df, t critical for two tailed test at 5%
level of significance is 2.1) and therefore it can be concluded that two treatment regimes do not differ
significantly.
8. The calculated t value (4.06 is more than critical value (At 9 df, t critical for two tailed test at 5%
level of significance is 2.26) and therefore it can be concluded that natural product was clinically
successful.
9. The calculated t value (-2.89) is more than critical value (At 11 df, t critical for two tailed test at 5%
level of significance is 2.2) and therefore it can be concluded that increase in systolic blood pressure is
due to oral contraceptive.
10. The calculated t value (-3.228) is more than critical value (At 13 df, t critical for two tailed test at
5% level of significance is 2.16). Hence, it can be concluded that the infants exposed to smoking are
more affected.
 Chapter 20
ONE WAY ANALYSIS OF VARIANCE (ANOVA)

Learning objectives
When we have finished this chapter, we should be able to:
1. Understand meaning ofANOVA.
2. Calculate ANOVAby definitional formula.
3. Calculate ANOVAby computational formula.

What isANOVA?
The ANOVA is used to identify and measure sources of variation within a collection of
observations, hence the name analysis of variance. Analysis of variance is a parametric statistical
technique that has found extensive applications in scientific research, mainly because of its
flexibility. This method may be employed to analyse both paired and independent data and also is
used to simultaneously compare large number of variables.
The one-way ANOVA is nothing more than an expansion of the t-test to more than two
groups of sample. The analysis of variance involves determining if the observed values belong to the
same population, regardless of the group, or whether the observations in at least one of these groups
come from a different population.
H0 : m1 = m2 = m3 .... = mk = m
To obtain a F value we need two estimates of the population variance. It is necessary to
examine the variability (analysis of variance) of observations within groups as well as between
groups. The F statistic is computed using a simplified ratio similar to the t-test.
Mean squared between (MSB)
F= ...1
Mean squared within (MSW)

To calculate the F-statistic for the decision rule either the definitional or computational
formulas may be used. With the exception of rounding errors, both methods will produce the same
results. In the former case the sample means and standard deviations are used.

Calculations ofANOVAusing definitional formula

The following steps are utilised for calculatingANOVA
1. Calculate the denominator of the F statistics, the mean squared within (MSW) in the same way as
the pooled variance is calculated for t test.
(n 1 - 1)s12 + (n 2 - 1)s 22 + (n 3 - 1)s 32 . . . (n k - 1)s 2k
M SW =
N ...2
305
 306 Textbook of Computer Applications and Biostatistics

Where,
n1, n2, n3 =number of observations in group 1, 2 and 3.
k = number of groups.
s1, s2, s3 = standard deviation of group 1, 2 and 3.
N = total number of observations in all groups.
2. Determine the grand mean or pooled mean by using following formula,
(n 1 X 1 ) + (n 2 X 2 ) + (n 3 X 3 ) . . . (n k X k )
XG =
N ...3
`X= mean
3. Now, the mean squared between (MSB) is calculated similar to a sample variance by squaring the
difference between each sample mean and the grand mean, and multiplying by the number of
observations associated with each sample mean
n 1 ( X1 - X G ) 2 + n 2 ( X 2 - X G ) 2 + n 3 ( X 3 - X G ) 2 . . . n k ( X k - X G ) 2
MSB =
K -1 ...4
4. Finally, F statistics is calculated by formula
M SB
F =
M SW

5. Compare the calculated F value with critical value from F table (Appendix III ) and take decision.
6. The greater the spread of the sample observations, the larger the denominator and the smaller the
calculated statistic, and thus the lesser the likelihood of rejecting H0. The greater the differences
between the means, the larger the numerator, the larger the calculated statistic, and the greater the
likelihood of rejecting Ho in favor of Ha.

Example 20.1
The rate of release of three controlled release formulation of Diclofenac sodium after 2hr in
% are given in following table. Is there a difference between the release kinetics of these three
formulations?
Formulation 1 4.21 5.23 4.01 6.00 5.25 6.41 4.52 4.18 6.05 4.66
Formulation 2 3.69 3.99 4.25 4.08 5.22 2.99 5.66 4.25
Formulation 3 5.00 6.55 6.02 6.11 4.88 4.29 6.00 5.45 5.04
Answer: Let us performANOVAfollowing the steps given below
1. State the null hypothesis
In this experiment null hypothesis states that the mean rate of release of diclofenac sodium
from three formulations is identical.
H0 : m1 = m2 = m3
2. State the alternative hypothesis
The mean rate of release of diclofenac sodium from the three formulations is not identical.
 One way analysis of Variance (ANOVA) 307

Ha : m1 ¹ m2 ¹ m3
3. State the level of significance
It is assumed that the level of significance is 0.05.
4. State the number of tails
As experiment involves a multiple hypothesis test, so the outcome is two tailed.
5. Select the appropriate statistical test
This is a case where more than two means will be simultaneously compared and hence
multiple hypothesis test is required.
There is a one factor i.e. formulation of which there are three sub categories F1, F2 and F3, so
one wayANOVAis required.
6. Perform the statistical test
1. Calculate the denominator of the F statistics, the mean squared within (MSW) in the same way as
the pooled variance is calculated for t test.
Formula:
(n1 - 1)s12 + (n 2 - 1)s22 + (n3 - 1)s32 . . . (n k - 1)s2k
MSW =
N
Data: n1 = 10, n2 =8, n3 = 9 k = number of groups= 3
Calculation for standard deviation
Formula:
( å X) 2
å X2 -
SD = n
n -1
Data: s1= standard deviation of group 1= 0.87
s2= standard deviation of group 2= 0.84
s3= standard deviation of group 3 =0.73
(10 - 1)0.87 2 + (8 - 1)0.84 2 + (9 - 1)0.73 2
MSW = = 0 .57
27
2. Determine the grand mean or pooled mean by using following formula
Formula:
(n 1 X 1 ) + (n 2 X 2 ) + (n 3 X 3 ) . . . (n k X k )
XG =
N
Calculation of mean
åX
X =
n
Data: `X1= mean = 5.05; `X2= mean = 4.27; `X3= mean = 5.48
(10 x 5.05) + (8 x 4.27) + (9 x 5.48) 133.99
XG = = = 4.96
27 27
 308 Textbook of Computer Applications and Biostatistics

3. Now, the mean squared between (MSB) is calculated similar to a sample variance by squaring the
difference between each sample mean and the grand mean, and multiplying by the number of
observations associated with each sample mean
Formula:
n ( X1 - X G ) 2 + n 2 ( X 2 - X G ) 2 + n 3 ( X 3 - X G ) 2 . . . n k ( X k - X G ) 2
MSB = 1
K -1

10(5.05 - 4.96 ) 2 + 8(4.27 - 4.96 ) 2 + 9(5.48 - 4.96 ) 2

MSB = = 3 .2
3 -1
4. Finally, F statistics can be calculated by using formula
M SB 3 .2
F= = = 5 . 35
M SW 0 . 59
5. Compare the calculated F value with critical value from F table (Appendix III ) and take decision.
The level of significance is 0.05
The numerator degrees of freedom = number of groups -1 = K-1= 3-1= 2
The number of degrees of freedom of denominator =N - K = 27- 3= 24
The critical F value for given degrees of freedom is found to be 3.40 (Appendix III).
7. Decision
If calculated F value is less than the critical F value the null hypothesis is accepted, whereas if
the calculated F value is more than or equal to critical F value, the null hypothesis is rejected in favour
of alternative hypothesis. The observed F value is greater than the critical F value. It is concluded that
there is a significant difference between the rates of release of diclofenac sodium from the three
controlled release formulations.

Example 20.2
Four brands of cereal are compared to see if they produce significant weight gain in rats. Four
groups of seven rats each were given a diet of the respective cereal brand. At the end of the
experimental period, the rats were weighed and the weight was compared to the weight just prior to
the start of the cereal diet. Determine whether each brand has a statistically significant effect on the
amount of weight gain. The data are provided in the table below.
BrandA 9 7 8 8 7 8 8
Brand B 5 4 6 4 5 7 3
Brand C 2 1 1 2 2 3 2
Brand D 3 8 5 9 2 7 8
Answer:
Let us performANOVAfollowing the steps given below
1. State the null hypothesis
In this experiment null hypothesis is the mean weight gain of rats from four brands of cereal
 One way analysis of Variance (ANOVA) 309

is identical.
H0 : m1 = m2 = m3
2. State the alternative hypothesis
The mean weight gain of rats from the four brands of cereal is not identical.
Ha : m1 ¹ m2 ¹ m3
3. State the level of significance
It is assumed that the level of significance is 0.05.
4. State the number of tails
As experiment involves a multiple hypothesis test, so the outcome is two tailed.
5. Select the appropriate statistical test
This is a case where more than two means will be simultaneously compared and hence
multiple hypothesis test is required.
There is a one factor i.e. brands of cereal, of which there are four sub categories A, B, C and
D, so one wayANOVA is required.
6. Perform the statistical test
1. Calculate the denominator of the F statistics, the mean squared within (MSW) in the same way as
the pooled variance is calculated for t test.
Formula:
(n - 1)s 12 + (n 2 - 1)s 22 + (n 3 - 1)s 32 . . . (n k - 1)s 2k
MSW = 1
N
Data: n1 = 7, n2 =7, n3 = 7, n4 = 7
k = number of groups= 4
Calculation for standard deviation
Formula:
(å X) 2
å X2 -
SD = n
n -1

Data: s1= standard deviation of groupA= 0.69

s2= standard deviation of group B= 1.34
s3= standard deviation of group C =0.69
s4= standard deviation of group D =2.7

(7 - 1)0.69 2 + (7 - 1)1.34 2 + (7 - 1)0.69 2 + (7 - 1)2.7 2

MSW = = 2.16
28

2. Determine the grand mean or pooled mean by using following formula

 310 Textbook of Computer Applications and Biostatistics

Formula:
(n1 X1 ) + (n 2 X 2 ) + (n 3 X 3 ) . . . (n k X k )
XG =
N
Calculation of mean
åX
X=
n

Data: `X1= mean =7.86 ; `X2= mean = 4.86; `X3= mean =1.86; X4= mean = 6.0

(7 x 7.86 ) + (7 x 4.86 ) + (7 x1.86) + (7 x6)

XG = = 5 . 14
28

3. Now, the mean squared between (MSB) is calculated similar to a sample variance by squaring the
difference between each sample mean and the grand mean, and multiplying by the number of
observations associated with each sample mean
Formula:
n ( X1 - X G ) 2 + n 2 ( X 2 - X G ) 2 + n 3 ( X 3 - X G ) 2 . . . n k ( X k - X G ) 2
MSB = 1
K -1

7(7.86 - 5.14) 2 + 7(4.86 - 5.14) 2 + 7(1.86 - 5.14) 2 + 7(6 - 5.14) 2

MSB = = 44.28
4 -1
4. Finally, F statistics can be calculated by using formula
M SB 4 4 .2 8
F= = = 2 0 .4 7
M SW 2 .1 6
5. Compare the calculated F value with critical value from F table (Appendix III ) and take decision.
The level of significance is 0.05
The numerator degrees of freedom = number of groups -1 = K-1= 4-1= 3
The number of degrees of freedom of denominator =N - K = 28- 4= 24
The critical F value for given degrees of freedom is found to be 3.01 (Appendix III).
7. Decision
If calculated F value is less than the critical F value the null hypothesis is accepted, whereas if
the calculated F value is more than or equal to critical F value, the null hypothesis is rejected in favour
of alternative hypothesis. The observed F value is greater than the critical F value. It is concluded that
there is a significant difference between the mean weight gain by rats fed on four brands of cereal.

Calculation ofANOVAusing computational formula

1. Calculate total sum of squares (SST)
SST can be calculated using the formula
 One way analysis of Variance (ANOVA) 311

(å X) 2
SST = å X 2 -
N ...5
Where,
å X = sum of squares of all observations.
2

(å X) = square of summation of observations.

2

N = total number of observations.

2. Calculate between sum of square (SSB)
The between sum of squares is calculated using the following equation

(å X 1 ) 2 (å X 2 ) 2 (å X 3 ) 2 (å X T ) 2
SSB = + + ..... - ...6
n1 n2 n3 N

(å X1) = Square of summation of observations in group 1

2

(å X2) = Square of summation of observations in group 2

2

(å X3) = Square of summation of observations in group 3

2

(å XT) = Square of summation of total observations

2

n1, n2 and n3 = Number of observations in group 1, 2 and 3 respectively

3. Calculate within sum of squares (SSW)
The within group sum of squares is calculated as difference between the total sum of squares
and the between sum of squares.
SSW = SST - SSB
4. Calculate between mean sum of squares (MSB) and within mean sum of squares (MSW)
Between mean sum of squares groups, MSB = SSB/Df
Where,
Df = degree of freedom = k -1= (No. of groups - 1)
Within mean sum of squares group, MSW= SSW/Df
Where
Df = degree of freedom = No. of observations - No. of groups = N-k
SSW = within group sum of squares
5. Calculate F statistics
F statistics can be calculated by using formula
MSB
F=
MSW
 312 Textbook of Computer Applications and Biostatistics

6. PrepareANOVATable

Source Degree of Sum of Mean F

freedom squares squares
Between groups k-1 SSB MSB MSB/MSW

Within groups N-k SSW MSW

Total N-1 SST

Example 20.3
The rate of release of three controlled release formulation of Diclofenac sodium after 2 hr in
% are given in following table. Is there a difference between the release kinetics of the three
formulations?
Formulation 1 4.21 5.23 4.01 6.00 5.25 6.41 4.52 4.18 6.05 4.66
Formulation 2 3.69 3.99 4.25 4.08 5.22 2.99 5.66 4.25
Formulation 3 5.00 6.55 6.02 6.11 4.88 4.29 6.00 5.45 5.04
Answer: Let us performANOVAfollowing the steps given below:
1. State the null hypothesis
In this experiment null hypothesis is the mean rate of release of diclofenac sodium from three
formulation is identical.
H 0 : m1 = m2 = m3
2. State the alternative hypothesis
The mean rate of release of diclofenac sodium from the three formulations is not identical.
Ha : m1 ¹ m2 ¹ m3
3. State the level of significance
It is assumed that the level of significance is 0.05.
4. State the number of tails
As experiment involves a multiple hypothesis test, so the outcome is two tailed.
5. Select the appropriate statistical test
This is a case where more than two means will be simultaneously compared and hence
multiple hypothesis test is required.
There is a one factor i.e. formulation, of which there are three sub categories F1, F2 and F3,
so one wayANOVAis required.
6. Perform the statistical test
1. Calculate total sum of squares (SST)
SST can be calculated using the formula
 One way analysis of Variance (ANOVA) 313

Formula:
(å X) 2
SST = å X 2 -
N
å X = sum of squares of all observations
2
Data:
(åX) = square of summation of observations
2

N = total number of observations = 27

Squared values
F1 F2 F3 F1 F2 F3
4.21 3.69 5 17.72 13.62 25.00
5.23 3.99 6.55 27.35 15.92 42.90
4.01 4.25 6.02 16.08 18.06 36.24
6 4.08 6.11 36.00 16.65 37.33
5.25 5.22 4.88 27.56 27.25 23.81
6.41 2.99 4.29 41.09 8.94 18.40
4.52 5.66 6 20.43 32.04 36.00
4.18 4.25 5.45 17.47 18.06 29.70
6.05 5.04 36.60 25.40
4.66 21.72
50.52 34.13 49.34 262.03 150.53 274.80

(å X) = (50.52+34.13+49.34) å X = (262.03+150.53+274.8)
2 2 2

(å X) = (133.99) = 17953.3 å X = 687.36

2 2 2

Therefore, the total sum of squares is

( å X) 2 17953 .3
SST = å X 2 - = 687 .36 - = 687 .36 - 664 .94 = 22 .42
N 27

2. Calculate between sum of squares (SSB)

Formula:
å(X1 ) 2 å(X 2 ) 2 å(X 3 ) 2 å(X T ) 2
SSB = + + -
n1 n2 n3 N
(å X1) = Square of summation of observations in group 1
2

(å X2) = Square of summation of observations in group 2

2

(å X3) = Square of summation of observations in group 3

2

(å XT) = Square of summation of total observations

2

(å X1) = (50.52) =2552.27

2 2

(å(X2) = (34.13) = 1164.86

2 2
 314 Textbook of Computer Applications and Biostatistics

(å X3) = (49.34) = 2434.44

2 2

(å XT) = 17953.3
2

n1 = 10; n2 =8; n3=9,

N =27
Therefore, the between sum of squares is

2552.27 1164.86 2434.44 17953

SSB = + + - = 671.33 - 664.94 = 6.39
10 8 9 27

3. Calculate within group sum of squares (SSW)

SSW = SST - SSB
SSW= 22.42-6.39 =16.03

4. Calculate between mean sum of squares (MSB) and within mean sum of squares (MSW)
Mean sum of squares between groups,
MSB = SSB/Df
Df = degree of freedom =No. of groups-1 = k-1 =3-1 =2
MSB= SSB/Df = 6.39/2 = 3.2
Mean sum of squares within group,
MSW= SSW/Df
Df = degree of freedom = No. of observations - k= 27 -3=24
MSW= SSW/Df = 16.03/24= 0.67

5. Calculate F statistics
F statistics can be calculated by using formula
F = Mean sum of squares between groups/ Mean sum of squares within group
F=MSB/MSW
F = 3.2/0.67 = 4.78

6. PrepareANOVATable

Source Degree of Sum of Mean F

freedom squares squares
Between groups 2 6.39 3.20 4.78

Within groups 24 16.03 0.67

Total 26 22.42
 One way analysis of Variance (ANOVA) 315

7. Decision
If calculated F value is less than the critical F value the null hypothesis is accepted, whereas if
the calculated F value is more than or equal to critical F value, the null hypothesis is rejected in favour
of alternative hypothesis. The observed F value is greater than the critical F value (3.40). It is
concluded that there is a significant difference between the rates of release of diclofenac sodium from
the three controlled release formulations.

Example 20.4
Four brands of cereal are compared to see if they produce significant weight gain in rats. Four
groups of seven rats each were given a diet of the respective cereal brand. At the end of the
experimental period, the rats were weighed and the weight was compared to the weight just prior to
the start of the cereal diet. Determine whether each brand has a statistically significant effect on the
amount of weight gain. The data are provided in the table below.
BrandA 9 7 8 8 7 8 8
Brand B 5 4 6 4 5 7 3
Brand C 2 1 1 2 2 3 2
Brand D 3 8 5 9 2 7 8
Answer:
Let us performANOVAfollowing the steps given below
1. State the null hypothesis
In this experiment null hypothesis is the mean weight gain of rats from four brands of cereal
is identical.
H 0 : m1 = m2 = m3
2. State the alternative hypothesis
The mean weight gain of rats from the four brands of cereal is not identical.
Ha : m1 ¹ m2 ¹ m3
3. State the level of significance
It is assumed that th level of significance is 0.05
4. State the number of tails
As experiment involves a multiple hypothesis test, so the outcome is two tailed.
5. Select the appropriate statistical test
This is a case where more than two means will be simultaneously compared and hence
multiple hypothesis test is required.
There is a one factor i.e. brands of cereal of, which there are four sub categories A, B, C and
D, so one wayANOVA is required.
6. Perform the statistical test
1. Calculate total sum of squares (SST)
SST can be calculated using the formula
 316 Textbook of Computer Applications and Biostatistics

Formula:
(å X) 2
SST = å X 2 -
N
å X = sum of squares of all observations
2
Data:
(åX) = square of summation of observations
2

N = total number of observations = 28

Squared values
A B C D A B C D
9 5 2 3 81 25 4 9
7 4 1 8 49 16 1 64
8 6 1 5 64 36 1 25
8 4 2 9 64 16 4 81
7 5 2 2 49 25 4 4
8 7 3 7 64 49 9 49
8 3 2 8 64 9 4 64
55 34 13 42 435 176 27 296

(å X) = (55+34+13+42) å X = (435+176+27+296)
2 2 2

(å X) = (144) = 20736 å X = 934

2 2 2

Therefore, the total sum of squares is

( å X) 2 20736
SST = å X 2 - = 934 - = 934 - 740 .57 = 193 .43
N 28
2. Calculate between sum of squares (SSB)
Formula:
å(X1 ) 2 å(X2 ) 2 å(X3 ) 2 å(X4 ) 2 å(XT ) 2
SSB = + + + -
n1 n2 n3 n4 N
(å X1) = Square of summation of observations in BrandA
2

(å X2) = Square of summation of observations in Brand B

2

(å X3) = Square of summation of observations in Brand C

2

(å X4) = Square of summation of observations in Brand D

2

(å XT) = Square of summation of total observations

2

(å X1) = (9+7+8+8+7+8+8) = (55) =3025

2 2 2

(å(X2) = (5+4+6+4+5+7+3) = (34) = 1156

2 2 2

(å X3) = (2+1+1+2+2+3+2) = (13) = 169

2 2 2

(å X4) = (3+8+5+9+2+7+8) = (42) = 1764

2 2 2

(å XT) = 20736
2

n1 = 7, n2 =7, n3=7, n4=7

 One way analysis of Variance (ANOVA) 317

N =28
Therefore, the between sum of squares is

3025 1156 169 1764 20736

SSB = + + + - = 873.43 - 740.57 = 132.86
7 7 7 7 28

3. Calculate within group sum of squares (SSW)

SSW = SST - SSB
SSW= 193.43-132.86 =60.47

4. Calculate between mean sum of squares (MSB) and within mean sum of squares (MSW)
Mean sum of squares between groups,
MSB = SSB/Df
Df = degree of freedom =No. of groups-1 = k-1 =4-1 =3
MSB= SSB/Df = 132.86/3 = 44.28
Mean sum of squares within group,
MSW= SSW/Df
Df = degree of freedom = No. of observations - k= 28 -4=24
MSW= SSW/Df = 60.47/24= 2.52

5. Calculate F statistics
F statistics can be calculated by using formula
F = Mean sum of squares between groups/ Mean sum of squares within group
F=MSB/MSW
F =44.28/2.52 = 17.57

6. PrepareANOVATable

Source Degree of Sum of Mean F

freedom squares squares
Between groups 3 132.86 44.28 17.57

Within groups 24 60.57 2.52

Total 27 22.42
7. Decision
If calculated F value is less than the critical F value the null hypothesis is accepted, whereas if
the calculated F value is more than or equal to critical F value, the null hypothesis is rejected in favour
of alternative hypothesis. The observed F value is greater than the critical F value (3.01). Therefore, it
 318 Textbook of Computer Applications and Biostatistics

is concluded that there is a significant difference between the mean weight gain by rats fed from four
brands of cereal.

One WayANOVAUsing Microsoft Excel

The rate of release of three controlled release formulation of Diclofenac sodium after 2 hr in
% are given in following table. Is there a difference between the release kinetics of the three
formulations?
Formulation 1 4.21 5.23 4.01 6.00 5.25 6.41 4.52 4.18 6.05 4.66
Formulation 2 3.69 3.99 4.25 4.08 5.22 2.99 5.66 4.25
Formulation 3 5.00 6.55 6.02 6.11 4.88 4.29 6.00 5.45 5.04
Excel Solution
Step 1:
Open new file in MS-Excel as Book 1. Enter the data into an Excel datasheet (Sheet 1).
Worksheet will appear as follows:

Figure 20.1 Worksheet after data entry

Step 2:
In MS-Excel, select Tools menu from Menu bar. Then, it will display pull down menus. From
pull down menus, select DataAnalysis option. Instantly, DataAnalysis dialog box will appear.
Step 3:
SelectAnova: Single Factor option fromAnalysis Tools.

Figure 20.2 Window of Data Analysis

 One way analysis of Variance (ANOVA) 319

Step 4:
In the following Dialog box, enter the input range that corresponds to the data columns
($A$1:$C$11) and click OK. Check the option "Labels in First Row".

Figure 20.3 Window of Anova: Single Factor

The results appear in a new worksheet, as shown here:

Anova: Single Factor

SUMMARY
Groups Count Sum Average Variance
Formulation 1 10 50.52 5.052 0.755729
Formulation 2 8 34.13 4.26625 0.703513
Formulation 3 9 49.34 5.482222 0.538094

ANOVA
Source of Variation SS df MS F P-value F crit
Between Groups 6.38922 2 3.194608 4.782674 0.01786 3.40283
Within Groups 16.0309 24 0.667954

Total 22.4201 26

In this output, the test statistic, F, is reported in the analysis of variance table, F(2, 24) = 4.78 .
The p-value for this statistics is reported in the table as 0.017. As observed F value is greater than
 320 Textbook of Computer Applications and Biostatistics

critical F value at 5 % level of significance, it can be concluded that there is a significant difference
between the rates of release of diclofenac sodium from the three controlled release formulations.

Summary
An independent group ANOVA is an extension of the independent group t-test where we
have more than two groups. This test is used to compare the means of more than two independent
groups and is also called a One WayAnalysis of Variance.
ANOVAusing definitional formula
(n 1 - 1)s 12 + (n 2 - 1)s 22 + (n 3 - 1)s 32 . . . (n k - 1)s 2k
M SW =
N

(n1 X1 ) + (n 2 X 2 ) + (n 3 X 3 ) . . . (n k X k )
XG =
N
n 1 ( X1 - X G ) 2 + n 2 ( X 2 - X G ) 2 + n 3 ( X 3 - X G ) 2 . . . n k ( X k - X G ) 2
MSB =
K -1
MSB
F=
MSW

ANOVA using computational formula

(å X) 2
SST = å X 2 -
N
(å X1)2 (å X2 )2 (å X3 )2 (å XT )2
SSB = + + -
n1 n2 n3 N
SSW = SST - SSB
MSB = SSB/Df
MSW= SSW/Df
F = Mean sum of squares between groups (MSB)/ Mean sum of squares within group (MSW)

Multiple Choice Questions

1. Which of the following is not a necessary assumption underlying the use of the Analysis of
Variance technique?
a. The samples are independent and randomly selected.
b. The populations are normally distributed.
c. The variances of the populations are the same.
d. The means of the populations are equal.
 One way analysis of Variance (ANOVA) 321

2.Astatistical test used to compare 2 or more group means is known as _____.

a. One-way analysis of variance b. Post hoc test
c. t-test for correlation coefficients d. Simple regression
3. One wayANOVAis extension of _______ to more than 2 groups
a. Chi square b. t test
c. paired t test d. z test
4. One wayANOVAis _________ test.
a. parametric b. non-parametric
c. categorical d. none of these
5.ANOVAis used to identify and measure ________.
a. sources of variation b. location of mean
c. weighted mean of groups d. none of these
6. The greater the spread of sample observations, _______ will be the denominator and smaller the
calculated statistic.
a. smaller b. larger
c. lesser d. none of above
7. The greater the differences between the means, _______ will be the numerator and larger the
calculated statistic.
a. smaller b. larger
c. lesser d. none of above
8. The greater the spread of sample observations, _______ the likelihood of rejecting null hypothesis.
a. greater b. lesser
c. no relation d. none of above
9. The greater the differences between the means, ______ the likelihood of rejecting null hypothesis.
a. greater b. lesser
c. no relation d. none of above
10. F statistic is also called as_______.
a. deviation ratio b. mean ratio
c. mean squared d. variance ratio

Exercise
1. During the manufacture of a diclofenac coated tablet, samples were periodically selected from
production lines at three different facilities. Weights were taken for 10 tablets and their average
weights listed in table. Is there any significant difference in weights of the tablets between the three
facilities? F Critical at 2 degrees of freedom in numerator and 27 degrees of freedom in denominator
is 3.35 at 5% level of significance.
 322 Textbook of Computer Applications and Biostatistics

FacilityA 77.3 80.3 79.1 75.2 73.6 76.7 81.7 78.7 78.4 72.9
Facility B 71.6 74.8 71.2 77.6 74.5 75.7 76.1 75.9 75.5 74.0
Facility C 75.5 74.2 67.5 74.2 70.5 84.4 75.6 77.1 72.2 73.4

2. Four brands of diclofenac sodium were selected and assayed according to IP. Results of assay in
terms of percent labeled amount of drug are listed in table. Was there any significant difference based
on brands in terms of labeled claim? F Critical at 3 degrees of freedom in numerator and 36 degrees of
freedom in denominator is 2.86 at 5% level of significance.
BrandA 100 99.8 99.5 100.1 99.7 99.9 100.4 100 99.7 99.9
Brand B 99.5 100 99.3 99.9 100.3 99.5 99.6 98.9 99.8 100.1
Brand C 99.6 99.3 99.5 99.1 99.7 99.6 99.4 99.5 99.5 99.9
Brand D 99.8 100.5 100 100.1 99.4 99.6 100.2 99.9 100.4 100.1

3. Four brands of Atenolol were under investigation for their antihypertensive efficacy. Four groups
were selected comprising of 10 volunteers in each group. According to designed protocol drug was
given to each group and mean systolic blood pressure was listed in table. Determine if there is
significant difference in mean systolic blood pressure in order to assess the role of different brands of
Atenolol. F Critical at 3 degrees of freedom in numerator and 36 degrees of freedom in denominator is
2.86 at 5% level of significance.
BrandA 125 130 135 120 115 120 130 135 140 135
Brand B 120 122 115 110 125 122 120 120 126 120
Brand C 120 115 115 130 120 125 122 115 126 118
Brand D 118 120 118 120 120 115 125 125 120 115

4. Four treatments are given to four groups of patients with anaemia. Increase in Hb% level was noted
after one month . Find whether the difference in improvement in four groups is significant or not.
GroupA 3 1 2 0 1 2 2
Group B 3 2 2 3 1 3 2
Group C 3 4 5 4 2 2 4
Group D 4 2 3 1 4 5 1
F Critical at 3 degrees of freedom in numerator and 24 degrees of freedom in denominator is 3.01 at
5% level of significance.

5. Pharmaceutical company suspected that four filling machines were not filling the bottles in a
uniform way. An experiment on four machines were performed and recorded in table. Find whether
there is a significant difference in the filling performance of four machines? F Critical at 3 degrees of
freedom in numerator and 19 degrees of freedom in denominator is 3.24 at 5% level of significance.
 One way analysis of Variance (ANOVA) 323

MachineA 52.05 52.07 52.04 52.04 51.99

Machine B 51.98 52.05 52.06 52.02 51.99
Machine C 52.04 52.03 52.03 52.00 51.96
Machine D 52.00 51.97 52.98 51.99 51.96

6. Dissolution test was performed for different brands of paracetamol available in India. Dissolution
efficiencies were calculated and reported in table. Find whether there is a significant difference in the
dissolution efficiencies of six brands of paracetamol? F Critical at 5 degrees of freedom in numerator
and 30 degrees of freedom in denominator is 2.53 at 5% level of significance.
BrandA 63.33 56.67 56.67 60.00 56.67 53.33
Brand B 63.33 63.33 60.00 66.67 60.00 73.33
Brand C 50.00 46.67 53.33 50.00 46.67 53.33
Brand D 53.33 56.67 46.67 50.00 45.00 46.67
Brand E 45.00 46.67 48.33 45.00 46.67 48.33
Brand F 50.00 53.33 50.00 46.67 56.67 60.00

7. The students are doing titrimetric estimation of certain compound in the laboratory. Three groups
have reported their analysis results as given below.
GroupA 18.0 16.4 15.7 19.6 16.5 18.2
Group B 21.1 17.8 18.6 20.8 17.9 19.0
Group C 16.1 17.8 16.5
Perform ANOVA at 0.05 level of significance to test whether the differences among the
sample means are significant.

8. Three chemicals A, B and C show the cleaning efficiency as given below. Find whether the
differences among them are significant at 5% significance level.
ChemicalA 80 77 76 81 71
Chemical B 70 58 72 66 74
Chemical C 77 80 82 85 76

9. If five different liquid filling machines fill 30 ml medicines into a container as shown below, find
whether the differences amongst the five machines is significant at 0.01 significance level by
performingANOVA.
MachineA 30 25 27 26
Machine B 29 28 26 29
Machine C 37 32 32 35
Machine D 32 33 34 29
Machine E 31 26 27 32
 324 Textbook of Computer Applications and Biostatistics

10. In a pharmaceutical company, three different brands of lubricants were used during tableting. The
quantity required for three brands is given below. Test at the 0.01 level of significance whether the
differences among the three sample means are significant.
LubricantA 8 14 10 10 13 12 13 12
Lubricant B 11 7 6 8 9 11 8 9 12 8 9
Lubricant C 7 7 5 6 10 4 8 9

Answers:
Multiple Choice Questions
1. d 2. a 3. b 4. a 5. a 6. b 7. b 8. b 9. a 10. d

Exercise
1. The observed F value is less than the critical F value (3.35). It is concluded that there is a no
significant difference in weights of the tablets between the three facilities.

Source Degree of Sum of Mean F

freedom squares squares
Between groups 2 53.09 26.55 2.49
Within groups 27 287 10.63
Total 29 340.1

2. The observed F value is more than the critical F value (2.86). It is concluded that there is a
significant difference in terms of labeled claim of four brands.

Source Degree of Sum of Mean F

freedom squares squares
Between groups 3 1.437 0.479 4.81
Within groups 36 3.578 0.099
Total 39 5.015

3. The observed F value is more than the critical F value (2.86). It is concluded that there is a
significant difference in mean systolic blood pressure produced by different brands ofAtenolol.

Source Degree of Sum of Mean F

freedom squares squares
Between groups 3 535.47 179.49 5.65
Within groups 36 1143.3 31.75
Total 39 1681.77
 One way analysis of Variance (ANOVA) 325

4. The observed F value is more than the critical F value (3.01). It is concluded that there is a
significant difference in mean %Hb in four groups.
Source Degree of Sum of Mean F
freedom squares squares
Between groups 3 13.25 4.41 3.34
Within groups 24 31.71 1.32
Total 27 44.96

5.The observed F value is less than the critical F value (3.24). It is concluded that there is a no
significant difference in the filling performance of four machines.

Source Degree of Sum of Mean F

freedom squares squares
Between groups 3 0.094 0.031 0.615
Within groups 16 0.81 0.051
Total 19 0.91

6. The observed F value is more than the critical F value (2.53). It is concluded that there is a
significant difference in the dissolution efficiencies of six brands of paracetamol.

Source Degree of Sum of Mean F

freedom squares squares
Between groups 5 1199.83 239.96 15.43
Within groups 30 466.47 15.54
Total 35 1666.3

7. The observed F value is more than the critical F value (3.89). It is concluded that there is a
significant difference among sample means of three groups.

Source Degree of Sum of Mean F

freedom squares squares
Between groups 2 15.12 7.58 4.06
Within groups 12 22.34 1.86
Total 14 37.46

8.The observed F value is more than the critical F value (3.89). It is concluded that there is significant
difference in the cleaning efficiency of chemicalsA, B and C.

Source Degree of Sum of Mean F

freedom squares squares
Between groups 2 390 195 8.48
Within groups 12 276 23
Total 14 666
 326 Textbook of Computer Applications and Biostatistics

9. The observed F value is more than the critical F value (6.42). It is concluded that there is a
significant difference in filling performance of five liquid filling machines at 0.01 level of
significance.

Source Degree of Sum of Mean F

freedom squares squares
Between groups 4 136 34 6.54
Within groups 15 78 5.2
Total 19 214

10. The observed F value is more than the critical F value (5.61). It is concluded that there is a
significant difference in three brands of lubricants at 0.01 level of significance.

Source Degree of Sum of Mean F

freedom squares squares
Between groups 2 82 41 11.05
Within groups 24 89 3.70
Total 26 171
 Chapter 21
CORRELATION

Learning objectives
When we have finished this chapter, we should be able to:
1. Understand meaning of correlation.
2. Understand types of correlation.
3. Calculate the correlation coefficient.

What is correlation?
The relationship between two metric continuous variables is called correlation. The easiest
way to visualise the relationship between two continuous variables is graphically, using a scatter plot.
One variable is labeled X and plotted on the x-axis of the graph or the abscissa. The second variable, Y
is plotted on the vertical y-axis or the ordinate. The first role of correlation is to determine the strength
of the relationship between the two variables represented on the x-axis and the y-axis. The measure of
this magnitude is called the correlation coefficient. This index measures both the magnitude and the
direction of the relationships:
+ 1.0 perfect positive correlation
0.0 no correlation
- 1.0 perfect negative correlation
Types of correlation
1. Perfectly Positive Correlation
2. Perfectly Negative Correlation
3. Moderately Positive Correlation
4. Moderately Negative Correlation
5.Absolutely no correlation

1. Perfectly Positive Correlation

In this the variables are directly proportional to each other. If one variable rises another also
rises of if one variable falls another variable also falls. In a perfect +ve relationship (coefficient of +1),
all of the data points would fall on a straight line. Here this straight line runs from the lower left of the
scatter plot to upper right.

2. Perfectly Negative Correlation

If one variable increases, other variable decreases but the fall is directly proportional to each
other. Here also the points fall along a straight line, but the straight line runs from upper left side to

327
 328 Textbook of Computer Applications and Biostatistics

lower right of the scatter plot. This is called as negative correlation.

Positive Correlation Negative Correlation

3. Moderately Positive Correlation

If the points do not fall on straight line but an orientation of points is from the lower left to the
upper right, as shown in figure then the moderately positive correlation exists. Here the correlation
coefficient ranges from 0 to +1.

4. Moderately Negative Correlation

If the points do not fall on straight line but an orientation of points is from the upper left to the
lower right, as shown in figure, then the moderately negative correlation exists. Here the correlation
coefficient ranges from 0 to - 1.

5.Absolutely No Correlation
If the points fall within the circle there in no correlation. Here the two variables are not
related to one another.

Moderately Negative Correlation No Correlation Moderately positive Correlation

 Correlation 329

Correlation Coefficient
The strength of the relationship (correlation coefficient) can be calculated by using Pearson
correlation coefficient for parametric data while Spearman rank correlation is used for non parametric
procedures.
Significance of a correlation coefficient
A positive or negative correlation between two variables shows that a relationship exists.
Whether it is strong or weak correlation, it can be roughly interpreted as given in following
table.
< 20 Slight correlation
0.20 - 0.40 Low correlation
0.40-0.70 Moderate correlation
0.70-0.90 High correlation
> 90 Very high correlation

Calculation of Pearson Product Moment Coefficient (r) by definitional formula

The following mathematical formula is used for determining Pearson Correlation coefficient (r)

å(x - X) (y - Y)
r= ...1
å(x - X) 2 (y - Y) 2
Where
x = value of each measurement on x axis
y = value of each measurement on y axis
`X = Mean for variables on x axis
`Y = Mean for variables on y axis
1. Calculate the mean for both x variable (`X) and y variable (`Y).
åx
X = ...2
N
å y
Y =
N ...3
Where
x = value of each measurement on x axis
y = value of each measurement on y axis
N = number of observations
2. Now, calculate deviations of individual scores x and y about means; (x-`X) and (y-`Y).
3. Then, multiply these deviations (x-`X) (y-`Y) and find the sum of the product of these deviations
S (x-`X) (y-`Y). This is used in the numerator of the equation.
4. The deviations (x-`X) of x variables and deviations (y-`Y) of y variables are squared, which
 330 Textbook of Computer Applications and Biostatistics

gives S (x-`X) (y-`Y)

2 2

5. The table shown below is developed for computation of correlation coefficient.

Data layout for computation for Pearson Product Moment Correlation Coefficient formula

x y (x-`X) (y-`Y) (x-`X) (y-`Y) (x-`X)2 (y-`Y)2

x1 y1 ... ... ... ... ...
x2 y2 ... ... ... ... ...
x3 y3 ... ... ... ... ...
... ... ... ... ... ... ...
xn yn ... ... ... ... ...
S (x-`X) (y-`Y) S (x-`X) S (y-`Y)
2 2

6. Calculate, correlation coefficient, r by using following formula

å(x - X) (y - Y)
r=
å(x - X) 2 (y - Y) 2

Example 21.1
Samples of drug products are stored in their original containers under normal conditions and
sampled periodically to analyse the content of the medication. Determine correlation coefficient.
Data:
Time (months) 6 12 18 24 36 48
Content (mg) 995 984 973 960 952 948
Solution:
1. Construct the following table
(x-`X) (y-`Y) (x-`X) (y-`Y) (x-`X) (y-`Y)
2 2
x y
6 995 -18 26.33 -474 324 693.44
12 984 -12 15.33 -184 144 235.11
18 973 -6 4.33 -26 36 18.78
24 960 0 -8.67 0 0 75.11
36 952 12 -16.67 -200 144 277.78
48 948 24 -20.67 -496 576 427.11
144 5812 0 00 -1380 1224 1727.33
Sx Sy S (x-`X) S (y-`Y) S (x-`X) (y-`Y) S (x-`X) S (y-`Y)
2 2

2. Calculate the mean for both x variable (`X) and y variable (`Y).
`X = 144/6 = 24
 Correlation 331

`Y = 5812/6 = 968.66
S (x-`X) (y-`Y) = -1380
S (x-`X)2 = 1224
S (y-`Y)2 = 1727.33
3. Calculate, correlation coefficient, r by using following formula
å(x - X) (y - Y)
r=
å(x - X) 2 (y - Y) 2
-1380
r= = -0.9490
1224 ´1727.33
Correlation coefficient is -0.949.

Calculation of Pearson Correlation Coefficient, r by Computational Formula

Following mathematical formula is used for calculation of Pearson Correlation
n åxy - åx å y
r= ...4
n åx - (åx)2 n å y2 - (å y)2
2

Where
åx = Summation of x
åy =Summation of y
åxy = Summation of product of x and y
(åx)2 = Square of summation of x
åx2 = Summation of square of x
(åy)2 = Square of summation of y
åy = Summation of square of y
2

n = number of pairs of data

The following steps should be used for calculating correlation coefficient by computational formula
1. Develop the table as shown below

2 2
x y x y xy
x1 y1 ... ... ...
x2 y2 ... ... ...
x3 y3 ... ... ...
... ... ... ... ...
xn yn ... ... ...
Sx Sy Sx2 Sy2 Sxy
 332 Textbook of Computer Applications and Biostatistics

2. Write the values of x variable in column 1of the table while the values of y variable in column 2 of
the table.
3. Write the individual x values squared (x2) in column 3 while individual y values squared (y2) in
column 4 of the table.
4. Write the product of x and y for each data point in column 5 of the table.
5. Now using the formula for Pearson correlation, find the value of r.
n å xy - å x å y
r=
n å x - (å x) 2 n å y 2 - (å y)2
2

Example 21.2
Samples of drug products are stored in their original containers under normal conditions and
sampled periodically to analyse the content of the medication. Determine correlation coefficient.
Data:
Time (months) 6 12 18 24 36 48
Content (mg) 995 984 973 960 952 948
Solution:
1. Develop the table as shown below
2 2
x y x y xy
6 995 36 990025 5970
12 984 144 968256 11808
18 973 324 946729 17514
24 960 576 921600 23040
36 952 1296 906304 34272
48 948 2304 898704 45504

Sx Sy Sx2 Sy2 Sxy

144 5812 4680 5631618 138108
2. Calculate, correlation coefficient, r by using following formula
n å xy - å x å y
r=
n å x - ( å x) 2
2
n å y 2 - ( å y) 2

(6 ´ 138108) - (144 ´ 5812)

r=
(6 ´ 4680 - (144) 2 (6 ´ 5631618 - (5812) 2

r = - 0.9490

Coefficient of correlation r is -0.949.

 Correlation 333

Example 21.3.
Following data were obtained for plotting calibration curve of drug. Determine correlation
coefficient by both methods .

Data:
Concentration (mg/l) 5 10 15 20 25 30
Absorbance 0.11 0.2 0.31 0.4 0.51 0.62
1. Definitional formula
Solution:
1. Construct the following table
x y (x-`X) (y-`Y) (x-`X) (y-`Y) (x-`X)2 (y-`Y)2
5 0.11 -12.5 -0.248 3.1 156.25 0.0616
10 0.20 -7.5 -0.158 1.18 56.25 0.025
15 0.31 -2.5 -0.048 0.12 6.25 0.0023
20 0.40 2.5 0.042 0.10 6.25 0.0017
25 0.51 7.5 0.152 1.14 56.25 0.0023
30 0.62 12.5 0.262 3.27 156.25 0.0684
105 2.15 00 00 8.925 437.5 0.1822
Sx Sy S (x-`X) S (y-`Y) S (x-`X) (y-`Y) S (x-`X) S (y-`Y)
2 2

2. Calculate the mean for both x variable (`X) and y variable (`Y).
`X = 105/6 = 17.5
`Y = 2.15/6 = 0.358
S (x-`X) (y-`Y) = 8.925
S (x-`X) = 437.5
2

S (y-`Y) = 0.1822
2

3. Calculate, correlation coefficient, r by using definitional formula

å(x - X) (y - Y)
r=
å(x - X) 2 (y - Y) 2

8.925
r= = 0 .999
437.5 ´ 0 .1822

Correlation coefficient is 0.999.

 334 Textbook of Computer Applications and Biostatistics

Computational Formula
Solution:
1. Develop the table as shown below
2 2
x y x y xy
5 0.11 25 0.0121 0.55
10 0.2 100 0.04 2
15 0.31 225 0.0961 4.65
20 0.4 400 0.16 8
25 0.51 625 0.2601 12.75
30 0.62 900 0.3844 18.6

Sx Sy Sx Sy Sxy
2 2

105 2.15 2275 0.9527 46.55

2. Calculate, correlation coefficient, r by using following formula
n å xy - å x å y
r=
n å x - ( å x) 2
2
n å y 2 - ( å y) 2

(6 ´ 4 6 . 55 ) - (105 ´ 2 . 15 )
r=
(6 ´ 2 275 - (105 ) 2 (6 ´ 0 . 9527 - ( 2 . 15 ) 2

r = 0.999
Coefficient of correlation r is 0.999

Estimation of Correlation coefficient by using Excel

Example 21.1
Samples of drug products are stored in their original containers under normal conditions and
sampled periodically to analyse the content of the medication. Determine correlation coefficient.
Data:
Time (months) 6 12 18 24 36 48
Content (mg) 995 984 973 960 952 948

Solution:
Step 1:
Open new file in MS-Excel as Book 1.
Enter the data into an Excel datasheet (Sheet 1).
Worksheet will appear as follows:
 Correlation 335

Figure 21.1 Data entry

Step 2:
In MS-Excel, select Tools menu from Menu bar. Then, it will display pull down menus.
From pull down menus, select DataAnalysis option. Instantly, DataAnalysis dialog box will appear.
Step 3:
Select Correlation option fromAnalysis Tools and then click on Ok.

Figure 21.2 Window of Data Analysis

Step 4: In the following Dialog box, enter the input range that corresponds to the data columns
($A$1:$B$7). Click on Labels and then click on OK.

Figure 21.3 Window of Correlation

 336 Textbook of Computer Applications and Biostatistics

The results will appear in a new worksheet, as shown here:

Time (X) Content Remaining (Y)
Time (X) 1
Content Remaining (Y) -0.949074491 1

So, correlation coefficient is -0.949.

Summary
Correlation
The relationship between two metric continuous variables is called correlation.
Types of correlation
+ 1.0 perfect positive correlation
0.0 no correlation
- 1.0 perfect negative correlation
Significance of correlation
< 0.2 Slight correlation
0.20 - 0.40 Low correlation
0.40-0.70 Moderate correlation
0.70-0.90 High correlation
> 0.9 Very high correlation
Computation of correlation coefficient
Using definitional formula Using computational formula
å(x - X) (y - Y) n å xy - å x å y
r= r=
å(x - X) (y - Y)
2 2 n å x - (å x) 2 n å y 2 - (å y)2
2

Multiple Choice Questions

1. The relationship between two metric continuous variables is called as ________.
a.Association b. Correlation
c. Regression d. Relation
2. Correlation coefficient for parametric data can be calculated by using _________.
a. Spearman rank b. Pearson
c. a and b d. None of above
3. The correlation coefficient provides:
a. measure of the extent to which changes in one variable cause changes in another variable.
b. a measure of the strength of the linear association between two categorical variables.
c. a measure of the strength of the association between two categorical variables.
d. a measure of the strength of the linear association between two quantitative variables.
 Correlation 337

4. Correlation involves
a. Independent variable b. Response variable
c. a and b d. None of above
5. A statistical test used to determine whether a correlation coefficient is statistically significant is
called the ___________.
a. One-way analysis of variance b. t-test for independent samples
c. Chi-square test for contingency tables d. t-test for correlation coefficients
6. In this the variable are directly proportional to each other.
a. perfect positive correlation b. low correlation
c. moderate correlation d. no correlation
7. The correlation coefficient value of 0.70-0.90 means ______.
a. low correlation b. high correlation
c. moderate correlation d. very high correlation
8. If one variable increases, other variable decreases but the fall is directly proportional to each other.
This is _______ correlation.
a. perfect positive b. moderate
c. perfect negative d. no
9. For non parametric data, correlation is measured by _________.
a. Spearman rank b. Pearson
c. a and b d. none of above
10. The value of correlation coefficient < 0.2 suggests_________.
a. low correlation b. moderate correlation
c. slight correlation d. high correlation

Exercise
1. Following area under curves were observed after clinical trials of different formulations of same
drug. Calculate correlation coefficient between dose andAUC
Dosage 100 300 600 900 1200
AUC 1.07 5.82 15.85 25.18 33.12

2. Following data were obtained after diffusion experiment of drug. Calculate correlation coefficient.
Time (h) 1 2 3 4 5 6 7
Amount in donor compartment (mg) 248.7 246.6 244.9 243.4 242.1 241.9 240.1

3. Data given are the values for age (X, in years) and systolic blood pressure (Y, in mm/Hg) for 15
women. Calculate correlation coefficient between age and systolic BP.
X 42 46 42 71 80 74 70 80 85 72 64 81 41 61 75
Y 130 115 148 100 156 162 151 156 162 158 155 160 125 150 165
 338 Textbook of Computer Applications and Biostatistics

4. The following are the height (cm) and the weight ( kilogram) of 10 men. Calculate correlation
coefficient between height and weight.
Height 162 168 174 176 180 180 182 184 186 186
Weight 65 65 84 63 75 76 82 65 80 81

5. Following data were obtained during laboratory experiment of muscular contraction of a rabbit
intestine. The height of the curve was considered as the response to the drug. Calculate correlation
coefficient between dose and response.
Dose of drug (mcg) 0.3 0.4 0.6 0.8 0.9 1.2
Response (mm) 54 59 60 65 70 75

6. In a study on the elimination of a drug in man, the following data were recorded. Calculate
correlation coefficient between time and drug concentration.
Time (h) 0.5 1 2 3 4 5
Drug Concentration (g/ml) 0.39 0.34 0.27 0.2 0.16 0.1

7. An experiment was conducted to study the effect on sleeping time of increasing the dosage of a
barbiturate. Three readings were made at each of three dose levels. Calculate correlation coefficient
between dose and sleeping time.
Dosage (mM/kg) 3 3 3 10 10 10 15 15 15
Sleeping Time (Hrs) 4 6 5 9 8 7 13 11 9

8. Calculate correlation coefficient for following data of variables.

x 43 62 52 41 53 51 59 46
y 50 56 53 46 48 57 55 42

9. Find if any correlation exists between two variables x and y

x 10 09 11 12 13 09 11 12 10 11 12 14
y 12 14 11 14 11 10 10 14 16 15 12 13

10. Find the value of Karl Pearson’s coefficient of correlation

x 12 9 8 10 11 13 7
y 14 8 6 9 11 12 3

Answers:
Multiple Choice Questions
1. b 2. b 3. d 4. b 5. d 6. a 7. b 8. c 9. a 10. c
 Correlation 339

Exercise
1. r = 0.998 2. r = -0.984
3. r = 0.564 4. r = 0.514
5. r = 0.981 6. r = -0.993
7. r = 0.900 8. r= 0.680
9. r = -0.01 10. r = 0.948
 Chapter 22
LINEAR REGRESSION

Learning objectives
When we have finished this chapter, we should be able to
1. Differentiate between correlation and regression.
2. Understand meaning of regression.
3. Determine regression coefficient (b).

Linear Regression
Linear regression is a statistical method to evaluate how one or more independent variables
influence outcomes for one continuous dependent variable through a linear relationship. For
example, we can predict the weight of children (up to 10 yrs), based on their age. Here age is known as
predictor and also known as independent variable. That which is predicted (weight) is referred to as
dependent variable. By the use of a regression equation, we can predict scores on the dependent
variable from those of independent variable. This is done by finding another constant called
‘Regression Coefficient’.
Let us understand difference between correlation and regression as both describe the
strength of the relationship between two or more continuous variables.
Correlation Linear Regression
1. Relationship between two variables is 1. Relationship between two variables is
established but response for dependent established but response for dependent
variable (y) cannot be predicted based on variable (y) can be measured based on
independent variable (x). independent variable (x).

2. Correlation involves only dependent or 2. Regression involves at least one

response variables. independent variable which is under
researchers control.

3. Correlation simply describes the strength 3. Regression in addition provides a method

and direction of the relationship. for describing the nature of relationship
between two or more continuous variable.

4. Correlation coefficient gives the idea of 4. Regression coefficient is useful for pre-
relationship between two or more continuous dicting the value of y from the value of
variable. x, or vice versa.

340
 Linear Regression 341

Meaning of Regression
Regression is based on the relationship or association between two (or more) variables. In
this analysis we have known variables which are used to predict the other variable. The known
variable (or variables) is called the independent variable (or variables). The variable we are trying to
predict is the dependent variable. In regression, we have only one dependent variable in our
regression equation. However, we can use more than one independent variable.
Regression equation can be expressed as,
Y = a + bX ...1
Where,
Y = Dependent variable
X = Independent variable
a = Y intercept
b = Slope of line
The above equation is linear in X and Y. Also graphically it represents a straight line. So
straight line is called as regression line.
The regression analysis confined to study of only one independent variable is called the
simple regression. If we are interested in studying relationship between a dependent variable with
more than one independent variable then it is called as multiple regression.

Regression coefficient (b)

Regression coefficient is a measure of the change in one dependent character (Y) with one
unit change in the independent character (X).

Calculation of Regression coefficient by least square method

Formula for calculating regression coefficient is

N å xy - ( å x) ( å y)
b= ...2
N å x 2 - ( å x) 2
Where
åxy = Summation of product of x and y.
åy = Summation of y
åx = Summation of x
åx = Summation of x
2 2

(åx) = Square of summation of x

2

N = Number of observations
b = Regression coefficient (Slope of line)
 342 Textbook of Computer Applications and Biostatistics

Steps for calculating regression coefficient

1. Develop a table as shown below
x y x2 y2 xy
2 2
x1 y1 x 1 y 1 x1y1
2 2
x2 y2 x 2 y 2 x2y2
2 2
x3 y3 x 3 y 3 x3y3
2 2
x3 y3 x 3 y 3 x4y4
2 2
xn yn x n y n xnyn
åx = åy = åx = åy = åxy =
2 2

2. Write the corresponding values of x and y in column 1 and 2 respectively and calculate their sums
(åx and åy )
3. Write squared values of x and y in column 3 and calculate its sum.
4. Write the squared values of y in column 4 and calculate its sum.
5. Write the product of values of x and y in column 5 and calculate its sum .
6. Using formula of regression coefficient find the value of ‘b’.
N å xy - (å x) (å y)
b=
N å x 2 - (å x) 2

Where
åxy = Summation of product of x and y.
åy = Summation of y
åx = Summation of x
åx = Summation of x
2 2

(åx) = Square of summation of x

2

N = Number of observations
b = Regression coefficient (Slope of line)
7. Now the y intercept (a) can be calculated by using formula
a = ( åy - b åx)/N ...3
8. Now, put the values of a and b in regression equation y = a + bx and find the value of y
corresponding value of x.

Example 22.1
Samples of drug products are stored in their original containers under normal conditions and
sampled periodically to analyse the content of the medication. Determine slope and intercept by least
square method.
 Linear Regression 343

Data:
Time (months) 6 12 18 24 36 48
Content (mg) 995 984 973 960 952 948

Solution:
1. Develop a table as shown below
x y x2 y2 xy
6 995 36 990025 5970
12 984 144 968256 11808
18 973 324 946729 17514
24 960 576 921600 23040
36 952 1296 906304 34272
48 948 2304 898704 45504
144 5812 4680 5631618 138108
åx åy åx åy åxy
2 2

2. Using formula of regression coefficient find the value of ‘b’.

N å xy - ( å x) ( å y) ( 6 ´ 138108) - (144 ´ 5812 )

b= =
N å x 2 - ( å x) 2 ( 6 ´ 4680 ) - (144 ) 2

828648 - 836928 - 8280

b = = = - 1 .1 2 7
28080 - 20736 7344

3. Now the y intercept can be calculated by using formula

( å y - b å x) (5812 - (-1.127 ´144)) 5812 + 162 × 28

a= = = = 995.72
N 6 6
4. Now, y = a+ bx. Once the values of a and b are known, then the value of y corresponding to x can be
calculated. For example, if value of y for corresponding value of x =18 is to be determined, then
y = a + bx
y = 995.72 - 1.127(18)
= 995.72-20.286
= 975.4
This can be used to cross check whether values of a and b are correct.

Example 22.2
The values for the birthweight (X, in kgs) and the increase in weight between 70 and 100
days of life, expressed as a percentage of the birth weight (Y) for 9 infants. Perform linear regression
 344 Textbook of Computer Applications and Biostatistics

analysis and determine slope and intercept. Report graphical display of data.
Data:
X 112 111 107 119 80 81 84 106 94
Y 63 66 72 52 118 120 114 72 91

Solution:
1. Develop a table as shown below
x y x2 y2 xy
112 63 12544 3969 7056
111 66 12321 4356 7326
107 72 11449 5184 7704
119 52 14161 2704 6188
80 118 6400 13924 9440
81 120 6561 14400 9720
84 114 7056 12996 9576
106 72 11236 5184 7632
94 91 8836 8281 8554
894 768 90564 70998 73196
åx åy åx2 åy2 åxy

2. Using formula of regression coefficient find the value of ‘b’.

N å x y - ( å x ) ( å y ) ( 9 ´ 7 3 1 9 6 ) - (8 9 4 ´ 7 6 8 )
b= =
N å x 2 - (å x)2 ( 9 ´ 9 0 5 6 4 ) - (8 9 4 ) 2

658764 - 686592 - 27828

b= = = - 1 .7 6
815076 - 799236 15840

3. Now the y intercept can be calculated by using formula

( å y - b å x) (768 - (-1.76 ´ 894)) 768 + 1573 × 44

a= = = = 259.84
N 9 9
4. Now, y = a+ bx. Once the values of a and b are known, then the value of y corresponding to x can be
calculated.
y = a + bx
y = 259.84 - 1.76 (x)
 Linear Regression 345

5. Graphical display of data

140
120

Increase in weight
100
80
60
40
20
0
60 80 100 120 140
Birth weight
Estimation of Linear Regression using Excel

Example 22.1
Samples of drug products are stored in their original containers under normal conditions and
sampled periodically to analyse the content of the medication. Determine slope and intercept by least
squar method. Determine content after 20 months.
Data:
Time (months) 6 12 18 24 36 48
Content (mg) 995 984 973 960 952 948
Excel Solution
Step 1:
Open new file in MS-Excel as Book 1. Enter the data into an Excel datasheet (Sheet 1).
Worksheet will appear as follows:

Figure 22.1 Data entry

 346 Textbook of Computer Applications and Biostatistics

Step 2:
In MS-Excel, select Tools menu from Menu bar. Then, it displays pull down menus. From
pull down menus, select DataAnalysis option. Instantly, DataAnalysis dialog box will appear.
Step 3:
Select Regression option fromAnalysis Tools and then click on Ok.

Figure 22.2 Window of Data analysis

Step 4: In the following Dialog box, enter the input range that corresponds to the data columns.
For Input Y Range box, type: $B$1:$B$7
For Input X Range box, type:$A$1:$A$7
Also you can select ranges
Click on Labels and
Then click on OK.

Figure 22.3 Window of Regression

The results will appear in a new worksheet, as shown here:
 Linear Regression 347

SUMMARY OUTPUT

Regression Statistics
Multiple R 0.9490745
R Square 0.9007424
Adjusted R Square 0.875928
Standard Error 6.5469646
Observations 6

ANOVA
df SS MS F Significance F
Regression 1 1555.882 1555.882 36.29918 0.003824076
Residual 4 171.451 42.86275
Total 5 1727.333

Standard Upper Lower Upper

Coefficients t Stat P-value Lower 95%
Error 95% 95.0% 95.0%
Intercept 995.72549 5.226327 190.5211 4.55E-09 981.21488 1010.236 981.2149 1010.236
Time (X) -1.127451 0.187133 -6.02488 0.003824 -1.647014168 -0.607888 -1.647014 -0.607888
In above results, b = -1.127; a = 995.72

Regression equation
y = a + bx
y = 995.72 -1.127x

Summary
Regression
Regression is based on the relationship or association between two (or more) variables.
Regression equation, Y = a + bX
Regression coefficient (b)
Regression coefficient is a measure of the change in one dependent character (Y) with one
unit change in the independent character (X)

N å xy - (å x) (å y)
b=
N å x 2 - (å x) 2

a = ( åy - b åx)/N

Multiple choice questions

1. Given that we have collected pairs of observations on two variables X and Y , we would consider
fitting a straight line with X as an explanatory variable if:
a. the change in Y is an additive constant.
b. the change in Y is a constant for each unit change in X
c. the change in Y is a fixed percent of Y
d. the change in Y is exponential
 348 Textbook of Computer Applications and Biostatistics

2. The least squares regression line is the line which:

a. is determined by use of a function of the distance between the observed Y ’s and the predicted Y ’s.
b. has the smallest sum of the squared residuals of any line through the data values.
c. for which the sum of the residuals about the line is zero.
d. has all of the above properties
3. Regression equation can be expressed as,
a. Y = a + bX b. X = a + bY
c. Y = abX d. X = abY
4. Regression is based on
a. the relationship or association between two (or more) variables.
b. the comparison between two (or more) variables
c. a & b d. none of above
5. _________is the set of procedures used to explain or predict the values of a dependent variable
based on the values of one or more independent variables.
a. Regression analysis b. Regression coefficient
c. Regression equation d. Regression line
6. _________ involves at least one independent variable which is under researchers control.
a. Correlation b.Association
c. Regression d. None of above
7. _________ is useful for predicting the value of y from the value of x .
a. Correlation coefficient b. Regression coefficient
c. Pearson coefficient d. Association coefficient
8. The regression analysis confined to study of only one independent variable is called ________.
a. simple regression b. multiple regression
c. simple correlation d. multiple correlation
9. The known variable which is used to predict other variable is _______.
a. independent variable b. dependent variable
c. regression variable d. response variable
10. _________ involves only dependent or response variable.
a. Simple regression b. Multiple regression
c.Association d. Correlation

Exercise
o
1. In an experiment investigating the breakdown of aspirin in a pharmaceutical product stored at 25 C
is given below. Perform linear regression analysis.
Time 18 35 51 68 85 101 118 136 166
Aspirin remaining 603.6 601.1 597.9 594.3 591.4 587.3 581.8 580 578.4
 Linear Regression 349

Time 197 229 260 292 326 355

Aspirin remaining 570.5 561.4 557.4 549.2 546 541
2. Following area under curves were observed after clinical trials of different formulations of same
drug. Calculate slope and intercept.
Dosage 100 300 600 900 1200
AUC 1.07 5.82 15.85 25.18 33.12
3. Following data were obtained after diffusion experiment of drug. Determine slope and present data
graphically.
Time (h) 1 2 3 4 5 6 7
Amount in donor compartment (mg) 248.7 246.6 244.9 243.4 242.1 241.9 240.1
4. Following data were obtained during laboratory experiment of muscular contraction of a rabbit
intestine. The height of the curve was considered as the response to the drug. Perform linear
regression analysis.
Dose of drug (mcg) 0.3 0.4 0.6 0.8 0.9 1.2
Response (mm) 54 59 60 65 70 75
6. Data give the values for age (x, in years) and systolic blood pressure (y, in mm/Hg) for 15 women.
Perform linear regression analysis.
X 42 46 42 71 80 74 70 80 85 72 64 81 41 61 75
Y 130 115 148 100 156 162 151 156 162 158 155 160 125 150 165
7. An experiment was conducted to study the effect on sleeping time of increasing the dosage of a
certain barbiturate. Three readings were made at each of three dose levels. Perform linear regression
analysis.
Dosage (mM/kg) 3 3 3 10 10 10 15 15 15
Sleeping Time (Hrs) 4 6 5 9 8 7 13 11 9
8. In a study on the elimination of a drug in man, the following data were recorded. Perform linear
regression analysis.
Time (h) 0.5 1 2 3 4 5
Drug Concentration (g/ml) 0.39 0.34 0.27 0.2 0.16 0.1
9. Perform regression analysis of following data
Age (years) 66 38 56 72 42 45 55 47 36 63
Blood pressure 145 124 147 160 125 124 150 128 118 149
(mm of Hg)
10. Obtain the regression equation of sale of medicine and advertising expenses.
Sales of Medicines 190 240 250 300 310 335 300
Advertising expenses 5 10 15 20 20 30 30
 350 Textbook of Computer Applications and Biostatistics

Answers:
Multiple Choice Questions
1. b 2. b 3. a 4. c 5. a 6. c 7. b 8. a 9. a 10. d

Exercise
1. Intercept 606.98, Slope -0.19.
2. Intercept -2.3, Slope 0.03.
3. Intercept 249.38 , Slope -1.36 .
250
249
Amount diffused (mg) 248
247
246
245
244
243
242
241
240
239
0 2 4 6 8
4. Intercept 47.96 , Slope 22.68. Time (h)
5. Intercept 0.0013, Slope 0.02 .
0.7
0.6
Absorbance

0.5
0.4
0.3
0.2
0.1
0
0 5 10 15 20 25 30 35
Concentration
6. Intercept 99.96 , Slope 0.7.
7. Intercept 3.37 , Slope 0.495.
8. Intercept 0.406 , Slope -0.063.
9. Intercept 79.03, Slope 1.114
10. y = a + bx y = -28.62 + 0.171x
 Linear Regression 351

Important Points & Formulaes At A Glance

Presentation of data
Presentation of data Nominal Ordinal Metric Continuous Metric Discrete
Histogram, frequency
Pie chart, polygon, Box & Whisker, Bar chart, Line plot,
Graphical Bar chart Stem & Leaf plot
Bar chart Point plot
Ogive, Scatter plot

Measures of Location & Dispersion

Data Ungrouped data Grouped data Metric discrete data

Most frequently æ f1 - f 0 ö
Mode Mode = l1 + çç ÷÷ ´ i
è 2f1 - f 0 - f 2 ø
occurring score

Odd number
= Size or value of
æ n +1ö
ç ÷ th observation
è 2 ø
(n/2) - c.f. Median
Median Even number Median = l1 + ´i = value of (n/2)th observation
f
= Size or value of

n æn ö
th + ç +1÷th
2 è 2 ø observation
2

Mean Mean (X) =

åX
Mean (X) = A +
åf di i
´i Mean (X) = åf X i i
N N N

P
Percentile Pth Percentile = (N + 1)th value
100

Range Range = Lowest value to Highest value

IQR Interquartile range = Q1 to Q3

(å X ) 2 ( å fd ) 2 (å f i X i ) 2
SD å X2 - å fd 2 - å f i X i2 -
s= N s= N ´i s= N
N -1 N -1 N -1

SD
Coefficient of variation (CV) = Relative Standard Deviation = CV x 100
mean

351
 352 Textbook of Computer Applications and Biostatistics

Probability
Number of outcome that favour the event (m)
Probability [P E ] =
Total number of outcomes (N)
Probability of composite outcomes: P(Ei or Ej)= P(Ei) + P(Ej)
Probability of complementary event: p(`E ) = 1 - p(E)
Probability of intersect: p(Aand B) = p(AÇB)=p(A) x p(B)
Probability of conjoint: p(Aor B)= p(AÈB) =p (A) + p (B) - p(AÇB)
Conditional probability: p(A) given B = p(A|B) = p(AÇB) /p(B)
Probability of binomial distribution:
æn ö æn ö n! n!
P(X) = çç ÷÷ p X q n - X çç ÷÷ = P(X) = pXq n -X
èXø è X ø X!(n - X)! X!(n - X)!
Poisson distribution:

e- m m X mX
P(X) = = m
X! e X!
Estimation of Confidence Interval
Standard error of Mean
SD
SEM =
N
At 95% confidence, Population mean = Sample mean± 1.96 x SE
Confidence interval
z s
p % = X ± 1 - a /2
N
Key stages in hypothesis testing
1. State the null hypothesis 2. State the alternative hypothesis
3. Select the level of significance 4. Select number of tails
5. Test the statistics 6. Compare table and observed value
7. Decision

Decision rule
1.Determine p value 2. Compare it with the critical value, usually 0.05.
3. If the obtained p value is less than critical value, reject null hypothesis; otherwise accept it.

Types of error
Type I error is the probability of rejecting a true null hypothesis
Type II error is the probability of accepting a false H0.
 Important Points & Formulaes At A Glance 353

Choice of statistical test

z-test t-test

X - m0 X - m0
Single sample z= t (N-1)df =
s/ N S/ N
X1 - X 2 (X1 - X 2 ) S2p S2p
z= t= SE (X1 - X2 ) = +
SE (X1 - X 2 ) SE (X1 - X 2 ) N1 N2
Two independent
samples ( N 1 - 1) s 12 + ( N 2 - 1) s 22
s 12 s 22 S 2p =
SE(X1 - X 2 ) = + N1 + N 2 - 2
N1 N 2

Two paired (å X) 2
x-O x-O å X2 -
samples t= = SD = N
SE SD/ N N -1

ANOVA
(n 1 - 1 )s 12 + (n 2 - 1 )s 22 + (n 3 - 1 )s 32 . . . (n k - 1 )s 2k
M SW =
N
(n1 X1 ) + (n 2 X 2 ) + (n 3 X 3 ) . . . (n k X k )
XG =
Definitional N
formula n 1 ( X1 - X G ) 2 + n 2 ( X 2 - X G ) 2 + n 3 ( X 3 - X G ) 2 . . . n k ( X k - X G ) 2
MSB =
K -1
MSB
F=
MSW

(å X) 2
SST = å X 2 -
Computational N
formula (å X1 )2 (å X2 )2 (å X3 )2 (å XT )2
SSB = + + -
n1 n2 n3 N

Correlation Coefficient Regression Coefficient

Definitional å(x - X) (y - Y) Regression equation, Y = a + bX

formula r=
å(x - X) 2 (y - Y) 2 N å xy - (å x) (å y)
b=
N å x 2 - (å x) 2
Computational n å xy - å x å y
r= a = ( åy - b åx)/N
formula n å x - (å x) 2 n å y 2 - (å y)2
2
Appendix I
Normal-Curve Areas

z .00 .01 .02 .03 .04 .05 .06 .07 .08 .09
0.0 .0000 .0040 .0080 .0120 .0160 .0199 .0239 .0279 .0319 .0359
0.1 .0398 .0438 .0478 .0517 .0557 .0596 .0636 .0675 .0714 .0753
0.2 .0793 .0832 .0871 .0910 .0948 .0987 .1026 .1064 .1103 .1141
0.3 .1179 .1217 .1255 .1293 .1331 .1368 .1406 .1443 .1480 .1517
0.4 .1554 .1591 .1628 .1664 .1700 .1736 .1772 .1808 .1844 .1879
0.5 .1915 .1950 .1985 .2019 .2054 .2088 .2123 .2157 .2190 .2224

0.6 .2257 .2291 .2324 .2357 .2389 .2422 .2454 .2486 .2517 .2549
0.7 .2580 .2611 .2642 .2673 .2704 .2734 .2764 .2794 .2823 .2852
0.8 .2881 .2910 .2939 .2967 .2995 .3023 .3051 .3078 .3106 .3133
0.9 .3159 .3186 .3212 .3238 .3264 .3289 .3315 .3340 .3365 .3389
1.0 .3413 .3438 .3461 .3485 .3508 .3531 .3554 .3577 .3599 .3621

1.1 .3643 .3665 .3686 .3708 .3729 .3749 .3770 .3790 .3810 .3830
1.2 .3849 .3869 .3888 .3907 .3925 .3944 .3962 .3980 .3997 .4015
1.3 .4032 .4049 .4066 .4082 .4099 .4115 .4131 .4147 .4162 .4177
1.4 .4192 .4207 .4222 .4236 .4251 .4265 .4279 .4292 .4306 .4319
1.5 .4332 .4345 .4357 .4370 .4382 .4394 .4406 .4418 .4429 .4441

1.6 .4452 .4463 .4474 .4484 .4495 .4505 .4515 .4525 .4535 .4545
1.7 .4554 .4564 .4573 .4582 .4591 .4599 .4608 .4616 .4625 .4633
1.8 .4641 .4649 .4656 .4664 .4671 .4678 .4686 .4693 .4699 .4706
1.9 .4713 .4719 .4726 .4732 .4738 .4744 .4750 .4756 .4761 .4767
2.0 .4772 .4778 .4783 .4788 .4793 .4798 .4803 .4808 .4812 .4817

2.1 .4821 .4826 .4830 .4834 .4838 .4842 .4846 .4850 .4854 .4857
2.2 .4861 .4864 .4868 .4871 .4875 .4878 .4881 .4884 .4887 .4890
2.3 .4893 .4896 .4898 .4901 .4904 .4906 .4909 .4911 .4913 .4916
2.4 .4918 .4920 .4922 .4925 .4927 .4929 .4931 .4932 .4934 .4936
2.5 .4938 .4940 .4941 .4943 .4945 .4946 .4948 .4949 .4951 .4952

2.6 .4953 .4955 .4956 .4957 .4959 .4960 .4961 .4962 .4963 .4964
2.7 .4965 .4966 .4967 .4968 .4969 .4970 .4971 .4972 .4973 .4974
2.8 .4974 .4975 .4976 .4977 .4977 .4978 .4979 .4979 .4980 .4981
2.9 .4981 .4982 .4982 .4983 .4984 .4984 .4985 .4985 .4986 .4986
3.0 .4987 .4987 .4987 .4988 .4988 .4989 .4989 .4989 .4990 .4990

Also, for z = 4.0, 5.0 and 6.0, the areas are 0.49997, 0.4999997, and 0.499999999.

354
Appendix 2
Critical values of the t distribution

Two-tailed test One-tailed test

df a = 0.10 a = 0.05 a = 0.01 a = 0.10 a = 0.05 a = 0.01
1 6.314 12.706 63.657 3.078 6.314 31.821
2 2.920 4.303 9.925 1.886 2.920 6.965
3 2.353 3.182 5.841 1.638 2.353 4.541
4 2.132 2.776 4.604 1.533 2.132 3.747
5 2.015 2.571 4.032 1.476 2.015 3.365
6 1.943 2.447 3.707 1.440 1.943 3.143
7 1.895 2.365 3.499 1.415 1.895 2.998
8 1.860 2.306 3.355 1.397 1.860 2.896
9 1.833 2.262 3.250 1.383 1.833 2.821
10 1.812 2.228 3.169 1.372 1.812 2.764
11 1.796 2.201 3.106 1.363 1.796 2.718
12 1.782 2.179 3.055 1.356 1.782 2.681
13 1.771 2.160 3.012 1.350 1.771 2.650
14 1.761 2.145 2.977 1.345 1.761 2.624
15 1.753 2.131 2.947 1.341 1.753 2.602
16 1.746 2.120 2.921 1.337 1.746 2.583
17 1.740 2.110 2.898 1.333 1.740 2.567
18 1.734 2.101 2.878 1.330 1.734 2.552
19 1.729 2.093 2.861 1.328 1.729 2.539
20 1.725 2.086 2.845 1.325 1.725 2.528
21 1.721 2.080 2.831 1.323 1.721 2.518
22 1.717 2.074 2.819 1.321 1.717 2.508
23 1.714 2.069 2.807 1.319 1.714 2.500
24 1.711 2.064 2.797 1.318 1.711 2.492
25 1.708 2.060 2.787 1.316 1.708 2.485
26 1.706 2.056 2.779 1.315 1.706 2.479
27 1.703 2.052 2.771 1.314 1.703 2.473
28 1.701 2.048 2.763 1.313 1.701 2.467
29 1.699 2.045 2.756 1.311 1.699 2.462
30 1.697 2.042 2.750 1.310 1.697 2.457
40 1.684 2.021 2.704 1.303 1.684 2.423
60 1.671 2.000 2.660 1.296 1.671 2.390
120 1.658 1.980 2.617 1.289 1.658 2.358
?? 1.645 1.960 2.576 1.282 1.645 2.326
a denotes the level of significance and df the number of degrees of freedom.

355
 Appendix 3
Values of F 0.05 Degrees of freedom for numerator
1 2 3 4 5 6 7 8 9 10 12 15 20 24 30 40 60 120 ??
1 161 200 216 225 230 234 237 239 241 242 244 246 248 249 250 251 252 253 254
2 18.5 19.0 19.2 19.2 19.3 19.3 19.4 19.4 19.4 19.4 19.4 19.4 19.4 19.5 19.5 19.5 19.5 19.5 19.5
3 10.10 9.55 9.28 9.12 9.01 8.94 8.89 8.85 8.81 8.79 8.74 8.70 8.66 8.64 8.62 8.59 8.57 8.55 8.53
4 7.71 6.94 6.59 6.39 6.26 6.16 6.09 6.04 6.00 5.96 5.91 5.86 5.80 5.77 5.75 5.72 5.69 5.66 5.63
5 6.61 5.79 5.41 5.19 5.05 4.95 4.88 4.82 4.77 4.74 4.68 4.62 4.56 4.53 4.50 4.46 4.43 4.40 4.37
6 5.99 5.14 4.76 4.53 4.39 4.28 4.21 4.15 4.10 4.06 4.00 3.94 3.87 3.84 3.81 3.77 3.74 3.70 3.67
7 5.59 4.74 4.35 4.12 3.97 3.87 3.79 3.73 3.68 3.64 3.57 3.51 3.44 3.41 3.38 3.34 3.30 3.27 3.23
8 5.32 4.46 4.07 3.84 3.69 3.58 3.50 3.44 3.39 3.35 3.28 3.22 3.15 3.12 3.08 3.04 3.01 2.97 2.93
Critical Values of F Distribution

9 5.12 4.26 3.86 3.63 3.48 3.37 3.29 3.23 3.18 3.14 3.07 3.01 2.94 2.90 2.86 2.83 2.79 2.75 2.71
Degrees of freedom for denominator

10 4.96 4.10 3.71 3.48 3.33 3.22 3.14 3.07 3.02 2.98 2.91 2.85 2.77 2.74 2.70 2.66 2.62 2.58 2.54
11 4.84 3.98 3.59 3.36 3.20 3.09 3.01 2.95 2.90 2.85 2.79 2.72 2.65 2.61 2.57 2.53 2.49 2.45 2.40
12 4.75 3.89 3.49 3.26 3.11 3.00 2.91 2.85 2.80 2.75 2.69 2.62 2.54 2.51 2.47 2.43 2.38 2.34 2.30
13 4.67 3.81 3.41 3.18 3.03 2.92 2.83 2.77 2.71 2.67 2.60 2.53 2.46 2.42 2.38 2.34 2.30 2.25 2.21
14 4.60 3.74 3.34 3.11 2.96 2.85 2.76 2.70 2.65 2.60 2.53 2.46 2.39 2.35 2.31 2.27 2.22 2.18 2.13
15 4.54 3.68 3.29 3.06 2.90 2.79 2.71 2.64 2.59 2.54 2.48 2.40 2.33 2.29 2.25 2.20 2.16 2.11 2.07
16 4.49 3.63 3.24 3.01 2.85 2.74 2.66 2.59 2.54 2.49 2.42 2.35 2.28 2.24 2.19 2.15 2.11 2.06 2.01
17 4.45 3.59 3.20 2.96 2.81 2.70 2.61 2.55 2.49 2.45 2.38 2.31 2.23 2.29 2.15 2.10 2.06 2.01 1.96
18 4.41 3.55 3.16 2.93 2.77 2.66 2.58 2.51 2.46 2.41 2.34 2.27 2.19 2.15 2.11 2.06 2.02 1.97 1.92
19 4.38 3.52 3.13 2.90 2.74 2.63 2.54 2.48 2.42 2.38 2.31 2.23 2.16 2.11 2.07 2.03 1.98 1.93 1.88
20 4.35 3.49 3.10 2.87 2.71 2.60 2.51 2.45 2.39 2.35 2.28 2.20 2.12 2.08 2.04 1.99 1.95 1.90 1.84
21 4.32 3.47 3.07 2.84 2.68 2.57 2.49 2.42 2.37 2.32 2.25 2.18 2.10 2.05 2.01 1.96 1.92 1.87 1.81
22 4.30 3.44 3.05 2.82 2.66 2.55 2.46 2.40 2.34 2.30 2.23 2.15 2.07 2.03 1.98 1.94 1.89 1.84 1.78
23 4.28 3.42 3.03 2.80 2.64 2.53 2.44 2.37 2.32 2.27 2.20 2.13 2.05 2.01 1.96 1.91 1.86 1.81 1.76
24 4.26 3.40 3.01 2.78 2.62 2.51 2.42 2.36 2.30 2.25 2.18 2.11 2.03 1.98 1.94 1.89 1.84 1.79 1.73
25 4.24 3.39 2.99 2.76 2.60 2.49 2.40 2.34 2.28 2.24 2.16 2.09 2.01 1.96 1.92 1.87 1.82 1.77 1.71
30 4.17 3.32 2.92 2.69 2.53 2.42 2.33 2.27 2.21 2.16 2.09 2.01 1.93 1.89 1.84 1.76 1.74 1.68 1.62

356
40 4.08 3.23 2.84 2.61 2.45 2.34 2.25 2.18 2.12 2.08 2.00 1.92 1.84 1.79 1.74 1.69 1.64 1.58 1.51
60 4.00 3.15 2.76 2.53 2.37 2.25 2.17 2.10 2.04 1.99 1.92 1.84 1.75 1.70 1.65 1.59 1.53 1.47 1.39
120 3.92 3.07 2.68 2.45 2.29 2.18 2.09 2.02 1.96 1.91 1.83 1.75 1.66 1.61 1.55 1.50 1.43 1.35 1.25
?? 3.84 3.00 2.60 2.37 2.21 2.10 2.01 1.94 1.88 1.83 1.75 1.67 1.57 1.52 1.46 1.39 1.32 1.22 1.00
 Appendix 3
Degrees of freedom for numerator
Values of F 0.01

357
1 2 3 4 5 6 7 8 9 10 12 15 20 24 30 40 60 120 ??
1 4052 5000 5403 5625 5764 5859 5928 5982 6023 6056 6106 6157 6209 6235 6261 6287 6313 6339 6366
2 98.5 99.0 99.2 99.2 99.3 99.3 99.4 99.4 99.4 99.4 99.4 99.4 99.4 99.5 99.5 99.5 99.5 99.5 99.5
3 34.1 30.8 29.5 28.7 28.2 27.9 27.7 27.5 27.3 27.2 27.1 26.9 26.7 26.6 26.5 26.4 26.3 26.2 26.1
4 21.2 18.0 16.7 16.0 15.5 15.2 15.0 14.8 14.7 14.5 14.4 14.2 14.0 13.9 13.8 13.7 13.7 13.6 13.5
5 16.3 13.3 12.1 11.4 11.0 10.7 10.5 10.3 10.2 10.1 9.89 9.72 9.55 9.47 9.38 9.29 9.20 9.11 9.02
6 13.7 10.9 9.78 9.15 8.75 8.47 8.26 8.10 7.98 7.87 7.72 7.56 7.40 7.31 7.23 7.14 7.05 6.97 6.88
7 12.2 9.55 8.45 7.85 7.46 7.19 6.99 6.84 6.72 6.62 6.47 6.31 6.16 6.07 5.99 5.91 5.82 5.74 5.65
8 11.3 8.65 7.59 7.01 6.63 6.37 6.18 6.03 5.91 5.81 5.67 5.52 5.36 5.28 5.20 5.12 5.03 4.95 4.86
Critical Values of F Distribution

9 10.6 8.02 6.99 6.42 6.06 5.80 5.61 5.47 5.35 5.26 5.11 4.96 4.81 4.73 4.65 4.57 4.48 4.40 4.31
Degrees of freedom for denominator

10 10.0 7.56 6.55 5.99 5.64 5.39 5.20 5.06 4.94 4.85 4.71 4.56 4.41 4.33 4.25 4.17 4.08 4.00 3.91
11 9.65 7.21 6.22 5.67 5.32 5.07 4.89 4.74 4.63 4.54 4.40 4.25 4.10 4.02 3.94 3.86 3.78 3.69 3.60
12 9.33 6.93 5.95 5.41 5.06 4.82 4.64 4.50 4.39 4.30 4.16 4.01 3.86 3.78 3.70 3.62 3.54 3.45 3.36
13 9.07 6.70 5.74 5.21 4.86 4.62 4.44 4.30 4.19 4.10 3.96 3.82 3.66 3.59 3.51 3.43 3.34 3.25 3.17
14 8.86 6.51 5.56 5.04 4.70 4.46 4.28 4.14 4.03 3.94 3.80 3.66 3.51 3.43 3.35 3.27 3.18 3.09 3.00
15 8.68 6.36 5.42 4.89 4.56 4.32 4.14 4.00 3.89 3.80 3.67 3.52 3.37 3.29 3.21 3.13 3.05 2.96 2.87
16 8.53 6.23 5.29 4.77 4.44 4.20 4.03 3.89 3.78 3.69 3.55 3.41 3.26 3.18 3.10 3.02 2.93 2.84 2.75
17 8.40 6.11 5.19 4.67 4.34 4.10 3.93 3.79 3.68 3.59 3.46 3.31 3.16 3.08 3.00 2.92 2.83 2.75 2.65
18 8.29 6.01 5.09 4.58 4.25 4.01 3.84 3.71 3.60 3.51 3.37 3.23 3.08 3.00 2.92 2.84 2.75 2.66 2.57
19 8.19 5.93 5.01 4.50 4.17 3.94 3.77 3.63 3.52 3.43 3.30 3.15 3.00 2.92 2.84 2.76 2.67 2.58 2.49
20 8.10 5.85 4.94 4.43 4.10 3.87 3.70 3.56 3.46 3.37 3.23 3.09 2.94 2.86 2.78 2.69 2.61 2.52 2.42
21 8.02 5.78 4.87 4.37 4.04 3.81 3.64 3.51 3.40 3.31 3.17 3.03 2.88 2.80 2.72 2.64 2.55 2.46 2.36
22 7.95 5.72 4.82 4.31 3.99 3.76 3.59 3.45 3.35 3.26 3.12 2.98 2.83 2.75 2.67 2.58 2.50 2.40 2.31
23 7.88 5.66 4.76 4.26 3.94 3.71 3.54 3.41 3.30 3.21 3.07 2.93 2.78 2.70 2.62 2.54 2.45 2.35 2.26
24 7.82 5.61 4.72 4.22 3.90 3.67 3.50 3.36 3.26 3.17 3.03 2.89 2.74 2.66 2.58 2.49 2.40 2.31 2.21
25 7.77 5.57 4.68 4.18 3.86 3.63 3.46 3.32 3.22 3.13 2.99 2.85 2.70 2.62 2.53 2.45 2.36 2.27 2.17
30 7.56 5.39 4.51 4.02 3.70 3.47 3.30 3.17 3.07 2.98 2.84 2.70 2.55 2.47 2.39 2.30 2.21 2.11 2.01
40 7.31 5.18 4.31 3.83 3.51 3.29 3.12 2.99 2.89 2.80 2.66 2.52 2.37 2.29 2.20 2.11 2.02 1.92 1.80
60 7.08 4.98 4.13 3.65 3.34 3.12 2.95 2.82 2.72 2.63 2.50 2.35 2.20 2.12 2.03 1.94 1.84 1.73 1.60
120 6.85 4.79 3.95 3.48 3.17 2.96 2.79 2.66 2.56 2.47 2.34 2.19 2.03 1.95 1.86 1.76 1.66 1.53 1.38
?? 6.63 4.61 3.78 3.32 3.02 2.80 2.64 2.51 2.41 2.32 2.18 2.04 1.88 1.79 1.70 1.59 1.47 1.32 1.00
 BIBLIOGRAPHY

Belle GV, Fisher LD, Heagerty PJ, Lumley T. Biostatistics A methodology for the health sciences. John
Wiley & Sons, Inc.America, 2004.
Bowers D. Medical Statistics from Scratch An Introduction for Health Professionals. John Wiley &
Sons, Ltd., England, 2008.
David J. Pharmaceutical Statistics. Pharmaceutical Press, London, 2002.
De Muth JE. Basic statistics and Pharmaceutical applications. Marcel Dekker Inc, New York, 1999.
Freund JE. Modern elementary statistics. Prentice Hall International Inc, London, 1984.
Gupta V. Statiatical analysis with Excel. VJ Books Inc, Canada, 2002.
Kothari CR. Research methodology methods and techniques. New Age International Pvt Ltd., New
Delhi, 2004.
Le CT. Introductory statistics. John Wiley & Sons, Inc.,America, 2003.
Mahajan BK. Methods in Biostatistics. Jaypee, New Delhi, 2010.
Mithal P, Goel R. Computer fundamentals. Paragon International Publishers, New Delhi, 2007.
Nagpal DP. Computer fundamentals. S. Chand & Company Ltd, New Delhi, 2008.
Oka MM. Computer Fundamentals. 6th Edition. Everest Publishing House, Pune, 2001.
O’Leary TJ, O’Leary LI. Computer Essentials. Tata McGraw-Hill, New Delhi, 2002.
Po WAL. Statistics for pharmacists. Blackwell Publishing Ltd, India, 2006.
Rajaraman V. Fundamentals of Computers. 4th Edition. Prentice-Hall of IndiaPvt Ltd., New Delhi,
2004.
Ram B. Computer fundamentals architecture and organization. New Age International Publishers,
Delhi, 2007.
Rao BT. Methods of Biostatistics. Paras Medical Publishers, Hyderabad, 2010.
Row P. Essential statistics for the pharmaceutical sciences. John Wiley & Sons, Ltd., England, 2007.
Thakur PS. Manchanda R, Nand P. Computer in Pharmacy. 2nd Edition. Birla Publications Pvt Ltd.,
New Delhi, 2002.
Shah YI, Paradkar AR, Dhayagude MG. Introduction to biostatistics and computer applications. Nirali
Prakashan, Pune, 2007.
Sinha PK, Sinha PP. Computer Fundamentals. 4th Edition. BPB Publication, New Delhi, 2007.
Tipnis HP, BajajA. Clinical Pharmacy. Career Publications, Nasik, 2009.

358

View publication stats