Data Collection

Collection of primary data, Secondary data, Measures

of central tendency-mean, mode, median, Measures of
dispersion- Range, Mean deviation, Standard deviation,
Measures of asymmetry (skewness)
 Data Collection

• Data - Piece of information collected about the

phenomenon under study for analysis.
• Data can take different forms like
• Facts or statistics
• Opinions
• Perceptions and
• Feelings
Why do we collect data?
• We collect data in order to obtain information about a
particular subject,
• to keep records,
• to make decisions about important issues and
• to pass on information.
• Data collection is therefore an important aspect of any
type of research study.
• Inaccurate data collection can impact the results of a
study and ultimately leads to invalid results.
Quantitative vs. Qualitative
 Quantitative data

• Quantitative
• It is used to quantify variables of any form by
generating numerical data or data that can be
transformed into usable statistics.
• It uses measurable data to formulate facts, to
uncover patterns in research and generalize results
from a large sample population.
 Qualitative data
• Qualitative
• It is normally used to gain an understanding of
reasons, opinions, processes, and perceptions. It
uses a relatively smaller sample and is non statistical.
• A mixed method using quantitative and qualitative
methods is very common, especially with case studies,
and can be helpful in the triangulation of data required
to achieve validity.
Data Collection
 Data Collection

• The research question, objective and strategy defines

whether you need qualitative or quantitative data.
• Aim is to gathering and summarising the data -
Transform data into meaningful information in order to
identify and measure variables.
• An ideal data collection procedure should be clear,
unbiased, reliable and valid.
 Data Collection

The 5 W’s of data collection are:

1. What data is to be collected?
2. From whom data is to be collected?
3. Who will collect data?
4. From where the data will be collected?
5. When is the data collected?
Methods of Data Collection

• Begins after a research problem has been defined and

research design/ plan chalked out.
• While deciding about the method of data collection
there are two types of data viz.,
1. primary data and
2. secondary data
Methods of Data Collection
Primary Data Collection
Secondary Data Collection
Types of Primary data

• Demographics/Socio economics - it includes gender,

marital status, age, income level and so on
• Psychographics/Lifestyle - it includes hobbies, interests,
opinions on certain subjects like political or religious
opinions and so on
• Personal values - it includes altruistic or more positive,
more negative and so on
Types of Primary data
• Awareness/Knowledge - It includes awareness about
the economy, awareness about the product category in
question, awareness about the characteristics of the
product and so on.
• Intentions - how do you intend to use your product?
• Attitudes/Motivations/Benefits - what kind of benefits
do you perceive from the product? - how are you
motivated to purchase a particular product
Types of Primary data

• Actual behavior - It can be of different types.

• It can be your past purchase behavior which is
already what you have done or
• It can include also behavior which is supposed to be
done in the future something like a purchase
intention.
Primary vs. Secondary data
Characteristic Primary data Secondary Data
Secondary data refer to
Primary data are those
those data that have
Definition that are collected for
already been collected
the first time.
by some other person.
These are original
These are not original
because these are
because someone else
Originality collected by the
has collected these for
investigator for the
his own purpose.
first time.

These are in the form These are in the

Nature of Data
of raw materials. finished form.
Primary vs. Secondary data
Characteristic Primary data Secondary Data
These are more reliable and These are less reliable and
suitable for the enquiry less suitable as someone
Reliability and Suitability because these are collected else has collected the data
for a particular purpose. which may not perfectly
match our purpose.

Collecting primary data is Secondary data requires

Time and Money quite expensive both in the less time and money; hence
terms of time and money. it is economical.

No particular precaution or Both precaution and editing

editing is required while are essential as secondary
Precaution and Editing using the primary data as data were collected by
these were collected with a someone else for his own
definite purpose. purpose.
Collection of Primary data
There are several methods of collecting primary data.
Important ones are:
(i) observation method,
(ii) interview method,
(iii) through focus groups
(iv) through questionnaires,
(v) through schedules, and
(vi) other methods
Collection of Primary data
Other methods of data collection, particularly used by big
business houses in modern times. which include

(i) warranty cards;

(ii) distributor audits; (vi) through projective
(iii) pantry audits; techniques;
(iv) consumer panels; (vii) depth interviews, and
(v) using mechanical (viii) content analysis.
devices;
 Observation Method
• It is a technique of collecting all the data or acquiring
information through occurrences that can be observed
through senses, with or without mechanical devices.
• Use to :
1. understand an ongoing process or situation.
2. gather data on individual behaviours or interactions
between people.
3. know about physical setting.
4. where other methods are not possible.
 Why observe users?

• There's really three reasons probably more but the

three major one's are that
• some actions are really difficult to articulate.
• some descriptions may conflict with the actual
behavior
• participants will not know which details of their
actions are relevant to you
How to observe participants?

• The common steps followed are

1. Prepare your notes
2. Schedule and arrange access
3. Conduct the observation and
4. Analyze
 Prepare your notes
• What do you hope to capture in your notes?
• Are there other types of evidence you want to capture?
• What granularity is relevant to you?
• How will these notes be qualified or quantified?
Schedule and Arrange access

• Consider setting, presentation, etc,

• Follow up to conform
 Conduct Observation

• Provide clear instructions for the participant, if

necessary
• Set up any necessary equipment (e.g., recording, notes)
• Take notes using your preprepared structure
• Review notes after observation, write field notes if
appropriate
Analyze
Observation Method: Advantages

1. Collect data where and when an event or activity is

occurring.
2. Does not rely on people’s willingness or ability to
provide information.
3. Validity of data is more.
 Observation Method:
Disadvantages

1. Susceptible to observer’s biases.

2. Hawthorne Effect - It refers to a phenomenon in which
participants alter their behavior as a result of being
part of an experiment or study.
3. Expensive & Time consuming.
4. Do not provide data about the rationale of people’s
activities.
Observation Method: Tools

• Qualitative Observation
• Log (Field Diary)
• Anecdotes
• Video recording
• Quantitative Observation
• Checklist
• Rating scale
 Log (Field Diary)

• It is a daily record of events and conversations in the

field.
• A log is a historical listing of how researchers have
spent their time and can be used for planning purposes,
for example
• for keeping track of responses, and
• for reviewing what work has already been completed.
Log (Field Diary)
 Anecdotes

• It is something which is closely related to the value or

idea - It focus on behavior of particular interest.
• Anecdotes typically selects specific kinds of events or
behaviors for observation before hand.
• The observer objectively & accurately records the
information.
 Video Recording

• Video-tapes are mechanical devices used to capture

complex behaviors that might elude notice by on-spot
observers.
 Video Recording
Advantages:
1.Captures even the fine units of behavior such as micro-
momentary facial expressions.
2.Offers possibility of checking the accuracy of coders or
the recording skills of participant observers.
3.Useful as training guide.
4.Easier to conceal camera than human behavior.
 Video Recording

Dis-advantages:
1.Technical problems with lightening, camera lens etc.
2.Camera angle adopted could present a lop-sided view of
an event or situation.
3.Participants may be more self-conscious in front of
camera.
 Checklist
• Checklists includes several items on a topic and require same
response format of all items.
• It consists of a list of items with a place to check or to mark
“Yes or No”.
• A checklist enables the observer to record present or absence
of trait.
• It consists of a listing of steps, activities, and behaviours which
the observer records when an incident occurs.
• Useful for evaluating skills, behaviour, conditions, personality,
manifestations etc.
 Checklist…..

Advantages:
1.Allow inter-individual comparisons.
2.Simple method to record observation.
3.Useful in evaluating learning activities.
4.Useful in containing the attention of the observer.
5.Decreases the chances of error in observation.
 Checklist…..
Dis-Advantages
1. Does not indicate quality of performance, so usefulness
is limited.
2. Only a limited content of overall clinical performance
can be evaluated.
3. Only the presence or absence of an attribute, behavior
or performance parameter may be assessed.
4. Degree of accuracy cannot be assessed.
 Rating Scales
• Resemble check lists but used when finer discriminations
is required and indicate the degree to which a trait is
present.
• Rating scales provide systematic procedures for
obtaining, recording and reporting the observer’s
judgement.
• By a rating is meant the judgement of one person by
another.
 Rating Scales: Types

1. Graphic Rating Scale

2. Descriptive Rating Scale
3. Numerical Rating Scale
 Graphic Rating Scale
In this scale , performance is printed horizontally at
various points from lowest to highest. It includes
numerical points on the scale. It is anchored by two
extremes presented to respondents for evaluation of a
concept or object.
Descriptive Rating Scale
This scale does not use numbers but divide the assessment
into series of verbal phrases to indicate the level of
performance.

Personnel Level of Performance

Very Moderately
Active Passive
Active Active
A
B
C
Numerical Rating Scale
It divides evaluation criteria in to a fixed number of
points, but defines numbers at the extremes only.
1 - Very Poor
2 - Poor
3 - Moderate
4 - Good
5 - Very Good
 Rating Scale
Advantage
1. Easy to make and administer.
2. Easy to score.
3. Less time – consuming.
4. Can be used for large population.
5. Can be used to evaluate performance, skills and product
outcomes.
 Rating Scale

Disadvantage
1. Difficult or dangerous to fix-up rating about many
aspects of an individual.
2. Chances of subjective evaluation, thus scales may
become unscientific and unreliable.
 Interview
• Involves verbal communication between the researcher
and the subject during which information is provided to
the researcher.
• Interview is a conversation carried art with the definite
purpose of obtaining certain information by means of
spoken word.
• Most common method in qualitative and descriptive
studies.
 Interview: Benefits
• Provides in–depth and detailed information
• Data from illiterate subjects
• Higher Response
• Clarify mis-understanding
• Ask questions at several levels
• Helps to gather supplementary information
• Use of special devices
• Accuracy can be checked
• Flexible and adaptable
 Types of Interview

1. Unstr ucture d inte r view (n o n directive or n o n

standardised)
2. Semi structured interview (or) focused interview
3. Completely structured interview
Types of Interview
Types of Interview
Types of Interview
Unstructured Interview
Unstructured Interview
• Characterised by a flexibility of approach to questioning.
• Interviewer freely asks the questions according to his
own wish when required or omit the certain questions.
• Subjects are encouraged to talk freely and only very
few questions asked to direct the trend of the interview.
• Researcher is having greater freedom to record or omit
the responses.
• It may be also called as ‘Grand Tour’ technique.
Unstructured Interview

Advantages
1. Information obtained in such a casual manner enhances
reliability and credibility of data.
2. Explorative and qualitative studies
3. Less prone to interviewer biases.
4. “Probs” questions can be assuring additional information
to clarify.
Unstructured Interview
Disadvantages
1. Interviewer requires great deal of knowledge and skill
in order to analyse the data.
2. Information cannot be compared.
3. Analysis will be difficult.
4. Data interpretation based on researcher’s perception and
subjective feelings.
5. Time wasting.
Semi-structured Interview
• Used when researcher have a list of topics or broad questions
that must be addressed in an interview
• Interviewer's make a topic guide/interview guide containing
broad list of topics to be covered in an interview.
• Topic guide- A set of questions or list of topics
• Participants are assured to talk freely about the topic mention
in topic guide.
• Researcher will ask questions to different ways to different
participants.
• Includes both closed ended and open ended questions.
Semi- structured Interview

Advantages
1. Less prone to interviewer's bias.
2. More information can be explored from the respondent’s.
3. Needed data is collected.
4. Guides the interview.
Semi- structured Interview

Disadvantages
1. Some of the information may not be revealed.
2. Need to prepare a topic guide.
Structured Interview
 Structured Interview
• It involves asking the same questions, in the same order,
and in same manner of all respondents in a study.
• It commonly have fixed type, and closed ended
questions.
• It also known as standardized interview.
• Interviewers are not permitted to change even specific
wording also.
• It increases the reliability and credibility of data.
 Structured Interview

Advantages
1. Data of two interviews are easily comparable.
2. Recording, coding and analysis of data is easy.
3. Avoids irrelevant purposeless conversation.
 Structured Interview

Disadvantages
1. In-depth information may not be possible.
2. Exploration of data is limited.
3. It may not cover all the possible responses or
respondent views.
Focus–Group Interviews
Focus–Group setting
Focus–Group Interviews
• In this interviews, homogenous group of 5-10 people whose
opinions and experiences are solicited simultaneously.
• The interviewer/ researcher guide the discussion according
to written set of questions or topics.
• It is a planned discussion.
• Duration of the interview ranges from 1.5-2 hours.
• All the verbal & non-verbal information is recorded.
• Ample opportunity is given to respondents to express their
views.
Focus – Group Interviews

Advantages
1. Efficient and can generate a lot of information.
2. Stimulates new ideas and creative concepts
3. Involves many participants at one time.
4. Participants may feel comfortable to answer in a group
with similar interests.
Focus – Group Interviews

Disadvantages
1. Chances of client or researcher’s bias.
2. May be difficult to moderate by one person.
3. Data difficult to code, analyze and interpret.
4. Focus group may not be representative of entire
population.
Joint Interviews
 Joint Interviews
• They are conducted simultaneously to understand the
phenomenon involving two or more parties. People involved
in the interview are intimately related.
• Merits:
• Helpful in observing dynamics between two key actors.
• De-merits:
• Only supplements information.
• May be un-comfortable to participants as some things
cannot be discussed in front of other people.
 Life Histories
• They are narrative self-disclosures about individual life
experiences.
• Researchers ask respondents to provide, often in
chronologic sequence, a narration of their ideas and
experiences, either orally or writing.
• Life histories are usually done in the ethnographic
studies.
• Example: A study involving experiences of women who had
simultaneously experienced abuse and physical disability.
 Critical Incidents
• It is a method of gathering information about people’s
behaviors by examining specific incidents relating to the
behavior under investigation.
• The word ‘critical’ means that the incident must have had a
positive or negative impact on some outcome.
• Example: A study involving the outcomes of stress management
program. The 5-week program taught ‘mind-body-spiritual’
technique of silently repeating a mantra with spiritual meaning.
3-months later, critical incident interviews with 55 participants
yielded 147 incidents involving application of the technique.
Photo- Elicitation Interviews
• It involves an interview stimulated and guided by photographic images.
• Photographs of the participant’s world are taken up by the researcher
themselves or by the participants and become a stimulus for
discussion.
• Example: In an attempt to explore meaning and experience of hope
among young people living in Australia, participants can be given a
disposable camera to take photos showing hope for them and then
questioned during interviews.
• Participants need to be continually reassured that their taken-for
granted explanations of the photos are providing new and detailed
information.
 Questioning
• This method allows the researcher to gather information
by asking the questions orally (interview) or by means of
a formal, written document (questionnaire).
• Questionnaire - is a structured instrument consisting of
a series of questions prepared by researcher on a paper
and that a subject is asked to complete either through
pencil or through a computer and is used to gather data
for phenomenon under study.
 Questioning
• The instrument is called a SAQ(Self-Assessment Questionnaire)
when respondents complete the instrument themselves, usually
in a paper & pencil format.
• SAQ also known as ‘survey’ .
• Methods of Questionnaire Administration:
1. Direct Administration-Researcher will distribute the
questionnaire and respondent answer items by writing or
checking against correct response.
2. Post or e-mail including all electronic means. (Mailed
questionnaire)
Questionnaire: Types of Questions

Open-ended questions
• Provide opportunity to the respondents to express their
opinions and answers in their own way.
• No predetermined set of responses.
• Provide true, insightful and unexpected suggestions.
Questionnaire: Types of Questions

Closed-ended questions
• Closed–ended questions or fixed-alternative questions-
Response alternatives are pre-specifie d by the
researcher.
• Facilitate easy statistical analysis.
• Can be asked to different groups at different intervals.
Closed-ended questions
1. Dichotomous Questions
2. Multiple-choice Questions
3. Cafeteria Questions
4. Rank-order Questions
5. Contingency Questions
6. Rating Questions
7. Likert Questions
8. Bi-polar Questions
9. Matrix Questions
Dichotomous Questions

• Require respondent to make choices between two

responses such as
• yes/no
• male/female
Multiple-choice Questions

• Require respondents to make choice between two- or

more than two response alternatives.
 Cafeteria Question
• Requires respondents to select a best response that
most closely corresponds to their views.
• E.g. What do you think about hormone-replacement
therapy (HRT) ?
a. It is dangerous, should be avoided.
b. One should be cautious while using it.
c. I am uncertain about my views.
d. It is beneficial, should be promoted.
Rank-Order Questions
• Requires respondents to rank their responses from most
favorable to least favorable.
• E.g. What according to you is most important for your
life?
• Money ( )
• Education ( )
• Family ( )
• Health ( )
Contingency Questions

• A next question is asked if a respondent gives a

particular response to previous question only.
• E.g. Are you stressed?
• No
• Yes (If yes what are the reasons………… )
 Rating Questions
• Requires respondents to judge something along an
ordered dimension.
• Respondent is asked to rate a particular issue on a scale
that ranges from poor to good. They may provide a
number of choices.
Question 1 2 3 4

How do you rank a quality of

Good Fair Poor Very Poor
education in India?
 Likert Questions
• Helps to know how strongly the respondent agrees with
a particular statement thus helping to assess the
feelings of the respondents towards certain issue.

Question 1 2 3 4 5

This community is
Strongly Strongly
a good place to Agree Uncertain Disagree
Agree Disagree
raise children.
 Matrix Questions
• Include multiple questions and identical response
categories, placed one under the other, forming a matrix.
• E.g. Please let me know your weekly schedule of the
following:
Mon Tues Wed Thurs Fri Sat Sun
Gym √ √ √ √ √
Aerobics √
Eating Out √ √
Drink √
Scales: Types
 Likert Scale

• A Likert scale consists of several declarative items that

express a view point on a topic.
• Respondents are asked to indicate the degree to which
they agree or disagree with the the statement.
• Usually contains five degrees (but 3 - 7 may used).
• Developed using item analysis approach.
Likert Scale
 Likert Scale: Uses

• Measure the attitudes, values, and feelings of the

people about specific concepts such as situations, people,
places, objects, programs, practices, policies and so on.
• Quantified measurement of qualitative attributes.
• Assess opinions of the respondents about particular
abstract concept.
 Likert Scale
Advantages
1. Easy to construct
2. More reliable and valid tool to measure psychosocial
variables
3. Easy to administer
4. Le ss tim e co nsu m in g du r in g co nstr uctio n an d
administration
 Likert Scale
Disadvantages
1. Respondents may feel forced to answer pre-planned
items
2. Feelings of the respondents may not be fully assessed.
3. Difficulty in justifying the selection of number of
categories & numerical assignment to these categories.
4. Casual approach on part of respondent may mislead
Semantic Differential Scale
• Semantic Differential Scale consists of two opposite
adjectives with a 7-point scale between them.
• Respondent is asked to rate a given concept by selecting
one point on the scale that best describes his or her
point of view.
• The adjectives commonly used such as
• effective/ineffective,
• good/bad, or
• important/unimportant.
Semantic Differential Scale
PRACTITIONERS

Competent 7 6 5 4 3 2 1 Incompetent

Worthless 7 6 5 4 3 2 1 Valuable

Pleasant 7 6 5 4 3 2 1 Unpleasant
Semantic Differential Scale: Uses

1. Patient Satisfaction Survey

2. Customer Satisfaction Survey
3. Employee Survey
4. Marketing Survey
5. Operational Research
6. Personality Measurements
7. Clinical psychology
Semantic Differential Scale

Advantages
1. Easy to construct
2. Highly flexible.
3. Useful in evaluating several concepts such as person,
place, situation, abstract idea, controversial issue etc
Semantic Differential Scale

Disadvantages
1. Difficult to select relevant concepts appropriate for a
given study.
2. Time consuming to find bipolar adjectives.
 Visual Analogue Scale
• It is used to measure the intensity of certain sensations and
feelings such as pain, discomfort, anxiety, alertness, severity of
clinical symptoms, functional ability, and attitude towards
environmental conditions.
• It is a 100 mm horizontal or vertical line with a statement at
either end representing one extreme of the dimension being
measured.
• It requires subjects to respond for particular phenomenon
measured, which is later measured by using a ruler from left
end.
Visual Analogue Scale
 Visual Analogue Scale

Advantages
1. Reliable and Valid tool to measure the intensity of
certain sensations and feelings.
2. Rating of highly subjective phenomenon is possible by
this scale.
3. Most useful in studying changes in the phenomenon.
 Visual Analogue Scale

Disadvantages
1. Cannot be used to compare results across group of
individuals at same time.
 Schedules
• It is very much like the collection of data through
questionnaire, with little difference which lies in the fact
that schedules are being filled in by the enumerators
who are specially appointed for the purpose.
• These enumerators along with schedules, go to
respondents, put to them the questions from the
proforma in the order the questions are listed and
record the replies in the space meant for the same in
the proforma.
 Differences
Questionnaire Schedules
Filled out by the researcher - interpret questions
Sent through mail
when necessary.

Spend money only in preparing the questionnaire Money has to be spent in appointing enumerators

Non-response is generally very low - able to get

Non-response is usually high
answers to all questions

Identity of respondent is not known. Identity of respondent is known.

Questionnaire method is likely to be very slow The information is collected well in time

Only when respondents are literate and cooperative When the respondents happen to be illiterate.

The success lies more on the quality of the Success depends upon the honesty and competence
questionnaire itself of enumerators
 Warranty cards

Warranty cards are usually postal sized cards which are

used by dealers of consumer durables to collect
information regarding their products.
Distributor or store audits

• Distributor or store audits are performed by distributors

as well as manufactures through their salesmen at
regular intervals.
• To estimate market size, market share, seasonal
purchasing pattern and so on.
• Data are recorded on inventories on hand either by
observation or copying from store records.
 Pantry audits
• It is a technique used to estimate consumption of the
basket of goods at the consumer level
• Investigator collects an inventory of types, quantities
and prices of commodities consumed.
• Objective in a pantry audit is to find out what types of
consumers buy certain products and certain brands, the
assumption being that the contents of the pantry
accurately portray consumer’s preferences.
 Consumer panels
• An extension of the pantry audit approach on a regular
basis
• A consumer panel is essentially a sample of consumers
who are interviewed repeatedly over a period of time.
• Consumer panels are of two types viz.,
1. Transitory consumer panel - effect of a particular
phenomenon in before and after basis
2. Continuing consumer panel - indefinite period with a
view to collect data on a particular aspect
Use of mechanical devices
• The use of mechanical devices has been widely made to
collect information by way of indirect means.
• Eye camera - record the focus of eyes of a respondent
• Pupilometric camera - record dilation of the pupil
• Psychogal vanometer - the extent of body excitement
• Motion picture camera - to record movement of a buyer
• Audiometer - to find out the type of programmes liked
• are the principal devices so far developed and commonly
used by modern big business house
 Projective techniques

Projective techniques for the collection of data have been

develope d by psychologists to use projections of
respondents for inferring about underlying motives, urges,
or intentions which are such that the respondent either
resists to reveal them or is unable to figure out himself.
 Depth interviews

• Designed to discover underlying motives and desires

• Depth interviews require great skill on the part of the
interviewer and at the same time involve considerable
time.
 Content-analysis

• Content-analysis consists of analysing the contents of

documentary materials such as books, magazines,
newspapers and the contents of all other verbal
materials which can be either spoken or printed
• The analysis of content is a central activity whenever
one is concerned with the study of the nature of the
verbal materials.
Collection of Secondary data
Secondary data may either be published data or unpublished data. Usually published
data are available in:
• Various publications of the central, state are local governments
• Various publications of foreign governments or of international bodies and their
subsidiary organisations;
• technical and trade journals
• books, magazines and newspapers
• reports and publications of various associations connected with business and industry,
banks, stock exchanges, etc.;
• reports prepared by research scholars, universities, economists, etc. in different fields;
and
• public records and statistics, historical documents, and other sources of published
information.
Characteristics of Secondary Data

• Before using secondary data, must see that they possess

following characteristics:
1. Reliability of data
2. Suitability of data
3. Adequacy of data
 Selection of Method

• Researcher must judiciously select the method/methods

for his own study, keeping in view the following factors:
1. Nature, scope and object of enquiry
2. Availability of funds
3. Timefactor
4. Precision required
 Nature, scope and object of
enquiry

• Most important factor affecting the choice of a method.

• The method selected should be such that it suits the
type of enquiry that is to be conducted by the
researcher.
• This factor is also important in deciding whether the
data already available (secondary data) are to be used
or the data not yet available (primary data) are to be
collected.
 Availability of funds

• Availability of funds for the research project determines

to a large extent the method to be used for the
collection of data.
• Finance, in fact, is a big constraint in practice and the
researcher has to act within this limitation.
 Timefactor

• Availability of time has also to be taken into account in

deciding a particular method of data collection.
• Some methods take relatively more time, whereas with
others the data can be collected in a comparatively
shorter duration.
 Precision required

Precision required is yet another important factor to be

considered at the time of selecting the method of
collection of data.
 Statistics
• We have all come across Statistics in our everyday lives.

• Many comparisons that we routinely make involve

numbers - we may not be conscious of it, but we have
been using Statistics for a long long time.

• At a high level, statistics is a discipline that is dedicated

to drawing actionable insights from available *data.

• If the data is indeed not available, then it will have to

be generated carefully via a survey, or any other
mechanism.
*The information we gather with experiments and surveys is collectively called data.
Statistics
 Statistics
There are two broad flavours of Statistics:
 Statistics
1. The goal of Descriptive Statistics is to describe the
data - we stu dy metho ds of summarising the
information we have collected for an analysis.
2. Inferential Statistics focuses on drawing inferences
(conclusions) on the properties of a population by
analysing the data of a sample of the population
• Such analysis hinges on the assu mptions of
randomness and there is a degree of uncertainty in
the inferences.
Statistics
 Statistical method

• It help us to investigate questions in an objective

manner, that involves four components:
1. formulate a statistical question,
2. collect data,
3. analyze data, and
4. interpret results.
 Types of Data
• Life would be uninteresting if everyone looked the
same, ate the same food, and had the same thoughts.
Fortunately, variability is everywhere, and statistical
methods provide ways to measure and understand it.
• For some characteristics in a study, we often see
variation among the subjects.
• Other characteristics may vary both by subject and
across time.
 Types of Data

• Data is categorised into two types

1. Numerical(Quantitative) data - consist of numbers
representing counts or measurements.
2. Categorical(Qualitative) data - consist of names or
labels that belongs to one of a set of distinct
categories.
 Quantitative Data

Examples include:
• The ages (in years) of survey respondents
• The numbers of eggs that hens lay
• The amounts of milk from cows
• The weights of shipments received
• The numbers of cans of a drink
 Quantitative Data
Quantitative data can be further described by distinguishing
between discrete and continuous types.
1. Discrete data result when the number of possible values
is either a finite number or a “countable” number. (That is,
the number of possible values is 0 or 1 or 2, and so on.)
2. Continuous data result from infinitely many possible
values that correspond to some continuous scale that
covers a range of values without gaps, interruptions, or
jumps.
 Quantitative Data
Examples include:
• The ages (in years) of survey respondents - Discrete
• The numbers of eggs that hens lay - Discrete
• The amounts of milk from cows - Continuous
• The weights of shipments received - Continuous
• The numbers of cans of a drink - Discrete
• The volume or weight of Drink - Continuous
 Categorical data

Examples include:
• Geographical locations of retail outlets
• Rankings of subordinates at yearly reviews
• The political party affiliations
• The numbers that are sewn on the shirts of Players
 Data matrix

C h aracte r i st ic s o f
Something or someone
something or someone
Eg. Football player data
Eg. Football teams data
Case: CONDITION
Variables and constants
Football League
Football League data
Cases - Row wise
Variables - Column wise
Observations
Data matrix
 Data matrix
• U s u a l ly w e d o n o t p r e s e nt
complete data matrix to other
people because it is often huge.
• It doesn't give a clear overview of
the statistical information contained
within the data matrix.
• While presenting the information
make use of summaries of data in
the forms of tables and graphs
(or) by measuring Central tendency
and Dispersion
Measurement of data
Level of Measurement
• In applying statistics to real problems, the level of
measurement of the data helps us decide which
procedure to use.
• Another common way of classifying data is to use four
levels of measurement:
1.Nominal,
2.Ordinal,
3.Interval, and
4.Ratio.
 Nominal Level
• The lowest level of data measurement is the nominal level.

• It is characterised by data that consist of names, labels,

or categories only. The data cannot be arranged in an
ordering scheme (such as low to high).

• Survey responses • Employee ID

• The political party • Sex
• Security numbers • Religion
• telephone numbers • Pincode
Nominal Level
 Ordinal Level
• Ordinal-level data measurement is higher than the
nominal level.
• Data at this level if they can be arranged in some order,
but differences (obtained by subtraction) between data
values either cannot be determined or are meaningless.
• Nominal and Ordinal data are non metric data and are
sometimes referred to as qualitative data.
• Course Grades
• Likert-type scales
• Ranks
 Interval Level
• The interval level of measurement is like the ordinal level,
with the additional property that the difference between
any two data values is meaningful. However, data at this
level do not have a natural zero starting point.

• Temperatures
• Years
• the change in stock price.
• % change in employment,
• the % return on a stock
Interval Level
 Ratio Level
• Ratio-level data measurement is the highest level of
data measurement
• The ratio level of measurement is the interval level with
the additional property that there is also a natural zero
starting point (where zero indicates that none of the
quantity is present). For values at this level, differences
and ratios are both meaningful.
• Distances • Height • Time
• Prices • Weight • Volume
Ratio Level
Levels of Measurement
There is a natural zero starting
Ratio: Example: Distances
point and ratios are meaningful.

Differences are meaningful, but Example: Body

Interval: there is no natural zero starting temperatures in degrees
point and ratios are meaningless. Fahrenheit or Celsius
Categories are ordere d, but Example: Ranks of
Ordinal: differences can’t be found or are colleges in News and
meaningless. World Report

Categories only. Data cannot be

Nominal: Example: Eye colors
arranged in an ordering scheme.
Levels of Measurement
Comparison b/w levels

The concentric squares denote

that each higher level of
data can be analyzed by any
of the techniques used on
lower levels of data but, in
addition, can be used in other
statistical techniques.
Comparison b/w levels
• Nominal data are the most limited data in terms of the
types of statistical analysis that can be used with
them.
• With ratio data, a statistician can make ratio comparisons
and appropriately do any analysis that can be
performed on nominal, ordinal and interval data.
• Some statistical techniques require ratio data and
cannot be used to analyze other levels of data.
 Statistical Techniques
Statistical techniques can be separated into two categories:
1. Parametric statistics
• Parametric statistics require that data be interval or
ratio.
2. Nonparametric statistics.
• If the data are nominal or ordinal, nonparametric
statistics must be used. Nonparametric statistics can
also be used to analyze interval or ratio data.
Univariate Statistical Tests
Frequency Distributions
• When working with large data sets, it is often helpful to
organise and summarize the data by constructing a table
called a frequency distribution.
• Definition: A frequency distribution (or frequency table)
shows how a data set is partitioned among all of several
categories (or classes) by listing all of the categories
along with the number of data values in each of the
categories.
Frequency Table: Hair color
Frequency Table: Hair color
Categorical data
Quantitative data
Quantitative - Ordinal
• Quant. data doesn’t give a good
overview because it barely tells you
more than the original data matrix.

• T h e va r i a b le w e i g ht w a s a
quantitative data, sometimes it’s
possible turned into an ordinal
variable

• With ordinal data the advantage is

that yo u get a much bette r
overview but lose some information.
Ordinal Level
Ordinal - Quantitative

• It is very easy to recode

quantitative data into ordinal
ones. However, the other way
around is impossible.
• One cannot recode ordinal
variables into quantitative
ones.
Frequency Table
• Use data matrix as the source of
all statistical analyses.
• To present fin dings to other
people make use of summaries of
data.
• One very good way to summarize
is by making frequency tables.
• If necessary quantitative variables
can be recoded into ordinal ones.
 Stem-and-Leaf Plots
• A stem-and-leaf plot is constructed by separating the
digits for each number of the data into two groups, a
stem and a leaf.
• The leftmost digits are the stem and consist of the
higher valued digits.
• The rightmost digits are the leaves and contain the
lower values.
• For example, if 34 is one of the numbers, the stem is 3
and the leaf is 4.
 Stem-and-Leaf Plots
• For numbers with more than two digits, division of stem
and leaf is a matter of researcher preference.
• This technique is simple and provides a unique view of
the data.
• One advantage of such a distribution is readily see
the scores are in the upper or lower end of each
bracket and the spread of the scores.
• A second advantage of stem-and-leaf plots is that the
values of the original raw data are retained
Stem-and-Leaf Plots
Stem-and-Leaf Plots
 Charts & Graphs

The two primar y graphical

displays for summarising a
categorical variable are the pie
chart and the bar graph.
Graphs for Categorical data
 Pie Chart

• A pie chart is a circular depiction of data where the

area of the whole pie represents 100% of the data and
slices of the pie represent a percentage breakdown of
the sub levels(relative magnitudes) .
• They are widely used in business, particularly to depict
such things as budget categories, market share, and
time/resource allocations.
 Pie Chart

• The use of pie charts is minimized in the sciences and

technology because pie charts can lead to less accurate
judgments than are possible with other types of graphs.
• Generally, it is more difficult for the viewer to interpret
the relative size of angles in a pie chart than to judge
the length of rectangles in a bar chart.
Pie Chart
 Bar Graph
• A bar graph or chart contains two or more categories
along one axis and a series of bars, one for each
category, along the other axis.
• The length of the bar represents the magnitude of the
measure (amount, frequency, money, percentage, etc.) for
each category.
• The bar graph is qualitative because the categories are
non-numerical, and it may be either horizontal or
vertical.
 Bar Graph

• A bar graph generally is constructed from the same

type of data that is used to produce a pie chart.
• However, an advantage of using a bar graph over a pie
chart for a given set of data is that for categories that
are close in value, it is considered easier to see the
difference in the bars of bar graph than discriminating
between pie slices.
Bar Graph
Pie Vs. Bar Chart
Pie Vs. Bar Chart
Graphs for Quantitative data

• L et ’s ex p lo re h o w to
su m marize quantitative
variables graphically and
visualise their distribution.
• There are three types of
displays—the dot plot,
stem-and-leaf plot, and
histogram
 Dot Plots
• A relatively simple statistical chart that is generally
used to display continuous, quantitative data is the dot
plot.
• In a dot plot, each data value is plotted along the
horizontal axis and is represented on the chart by a dot.
• If multiple data points have the same values, the dots
will stack up vertically.
• It may not be possible to display a large number of of
the data values along the horizontal axis.
Dot Plots
Dot Plots
 Dot Plots

Dot plots can be especially useful for observing the overall

shape of the distribution of data points along with
identifying data values or intervals for which there are
groupings and gaps in the data.
Dot Plots
 Histograms
• A histogram is a graph that uses bars to portray the
frequencies or the relative frequencies of the possible
outcomes for a quantitative variable(in given class
intervals).
• If the class intervals used along the horizontal axis are
• equal, then the height of the bars represent the
frequency of values in a given class interval.
• unequal, then the areas of the bars (rectangles) can
be used for relative comparisons of class frequencies.
Histogram
Histogram
 Histograms
• With a dot plot or a stem-and-leaf plot, it’s easy to
reconstruct the original data set because the plot shows
the individual observations. This becomes unwieldy for
large data sets.
• In that case, a histogram is a more versatile way to
graph the data and picture the distribution. It uses bars
to display and summarize frequencies of different
outcomes.
Histogram
Shape of a Distribution

• These distributions very often, but not all histograms

have same shape.
• The skewed histogram is not symmetric, because the one
side of the distribution stretches out further than the
other.
• A histogram can also be skewed to the left or to the
right.
Shape of a Distribution
Example: Skewed Distribution
Measure of Central Tendency
Measures of central tendency

• A measure that is used to describe a set of data (or) it

answers, What is a typical value for the observations?
• Measures of central tendency can yield such information
as the average, the middle, and the most frequently
occurred.
• Measures of central tendency do not focus on the span
of the data set or how far values are from the middle
numbers.
 Mode
• The mode is the most frequently occurring value in a
set of data.
• Organising the data into an ordered array helps to
locate the mode.
• In the case of a tie for the most frequently occurring
value, two modes are listed - Bimodal.
• Data sets with more than two modes are referred to
as multimodal.
 Mode

• The mode is an appropriate

measure of central tendency
for nominal-level and ordinal
level data.
 Mode - Example
Find Mode for the following Box plots.
Mode
 Median
• The median is the middle value in an ordered array of
numbers.
• The level of data measurement must be at least ordinal
for a median to be meaningful.
• The median is unaffected by the magnitude of extreme
values. For this reason, the median is often the best
measure of location to use in the analysis of variables
such as house costs, income, and age.
Median - odd
Median - Even
 Mean

• The mean is the sum of the observations divided by the

number of observations. It is interpreted as the balance
point of the distribution.
• Population Mean

• Sample Mean
Mean - Example
Properties of the Mean

• The mean is the balance point of the data: If we were

to place identical weights on a line representing where
the observations occur, then the line would balance by
placing a fulcrum at the mean.
• Usually, the mean is not equal to any value that was
observed in the sample.
Properties of the Mean
Properties of the Mean
• The mean can be highly influenced by an outlier, which is
an unusually small or unusually large observation.
Properties of the Mean
• An outlier is an observation that falls well above or
well below the overall bulk of the data.
Properties of the Mean
• For a skewed distribution, the mean is pulled in the
direction of the longer tail, relative to the median.
Measure of central Tendency

• If your variable is
• Categorical - use the mode,
• Quantitative - median or
the mean.
Mean Vs. Median
• If your data has
• Influential outliers - Median
• Highly skewed - Mean.
Measure of Dispersion

• Measures of central tendency yield information about

the center or middle part of a data set. However,
measures of variability, to describe the spread or the
dispersion of a set of data.
• Using measures of variability in conjunction with
measures of central tendency makes possible a more
complete numerical description of the data.
Measure of Dispersion
Measure of Dispersion
 Range

• The range is the difference between the largest and

the smallest observations.
• An advantage of the range is its ease of computation,
but its application is limited.
• A disadvantage of the range is that, because it is
computed with the values that are on the extremes of
the data, it is affected by extreme values
Range
 Interquartile Range
• Leaves extreme values
• Divides distribution into four parts
Interquartile Range
 Interquartile Range

• It might still be useful to look for possible outliers in

study.
• As a rule of thumb, observations can be qualified as
outliers if they lie more than
• 1.5 IQR below the first quartile,
• 1.5 IQR above the third quartile.
Interquartile Range
Graph - Dispersion
Box plot - Example
Range and IQR
Variance, SD and MAD

• These measures takes into account all values of Data

matrix.
• These measures are not meaningful unless the data are
at least interval- level data.
• The variance and standard deviation are widely used in
statistics
Mean Absolute Deviation
• The mean absolute deviation (MAD) is the average of
the absolute values of the deviations around the mean
for a set of numbers.

• Because it is computed by using absolute values, it is less

useful in statistics than other measures of dispersion.
But it is used occasionally as a measure of error.
 Variance
• Mathematicians developed an alternative mechanism for
overcoming the zero-sum property of deviations from
the mean.
• The variance is the average of the squared deviations
about the arithmetic mean for a set of numbers.
Population Variance Sample Variance
Variance: Example
Variance: Example
 Variance

• Larger the variance, more the values spread out around

the mean
• The disadvantage of variance is metric of the variance is
the metric of the variable under analysis squared
• to solve this problem the straight solution is to take
square root of the variance, is called standard
deviation.
 Standard Deviation
• The standard deviation is a popular measure of variability.
• The standard deviation is the square root of the
variance, It is the average distance of an observation
from the mean
• Standard Deviation of
Population Sample
 Measures of shape

• Measures of shape are tools that can be used to

describe the shape of a distribution of data.
• There are two measures of shape
1. skewness and
2. kurtosis
 Skewness
• A distribution of data in which the right half is a mirror
image of the left half is said to be symmetrical.
• Skewness is when a distribution is asymmetrical or lacks
symmetry.
• The skewed portion is the long, thin part of the curve.
• Many researchers use skewed distribution to denote
that the data are sparse at one end of the distribution
and piled up at the other end.
Coefficient of Skewness
• This coefficient compares the mean and median in light of
the magnitude of the standard deviation.
 Kurtosis
• Kurtosis describes the amount of peakedness of a
distribution.
• Distributions that are high and thin are referred to
as leptokurtic distributions.
• Distributions that are flat and spread out are
referred to as platykurtic distributions.
• Between these two types are distributions that are
more “normal” in shape, referred to as mesokurtic
distributions.
Kurtosis
Inferential Statistics:
Parametric tests
Important parametric tests- z, t, F, Chi-Square,
ANOVA significance
Inferential Statistics
Inferential Statistics
Inferential Statistics
 Inferential Statistics

• In the inferential statistics process,

• a researcher selects a random sample from the
population,
• computes a statistic on the sample, and
• reaches conclusions about the population parameter
from the statistic.
Sample Collection
 Parametric tests
The important parametric tests are:
1. z-test
2. t-test
3. χ2-test and
4. F-test
• All these tests are based on the assumption of
normality i.e., the source of data is considered to
be normally distributed.
 Parametric tests
• In some cases the population may not be
normally distributed, yet the tests will be
applicable on account of the fact that we mostly
deal with samples and the sampling distributions
closely approach normal distributions.
• These tests are applied on data measured on
quantitative scale of measurement (Interval and
Ratio level data)
 Assumptions
• The variable being quantified must be in higher scales of
measurement
• The data should be continuous in nature to qualify for
parametric tests
• The data should be independent
• The data should be collected from a population that
follows normal distribution of the variable being quantified
• The samples drawn from the population must follow
homogeneity of variance
 Non Parametric tests
• Non- parametric tests imply that the tests are not based
on the assumptions that the data is drawn from a
probability distribution defined through parameters such
as mean, proportion and standard deviation.

• The test is not about the population parameter such as

mean and standard deviation.

• However, in these distribution-free tests there is a certain

amount of ignoring of information, they are weak and less
efficient than the relevant standard parametric tests.
Univariate Statistical Tests
 Random Variable
• A random variable is a variable (typically
represented by X) that has a single numerical
value, determined by chance, for each outcome of
a procedure.
or
• A random variable is a numerical measurement of
the outcome of a random phenomenon.
Random Variable
Random Variable
Probability distribution
A probability distribution is a description that gives
the probability for each value of the random
variable. It is often expressed in the format of a
graph, table, or formula.
Probability Distribution
Normal Distribution
How to Calculate Probabilities?

• The total area under normal curve is 1.

• Area between any two values of R.V under
normal curve gives probability
• But finding areas under complicated bell shaped
curve every time is difficult - Concept of
Standard Normal Distribution is used.
Standard Normal Distribution

It is a normal probability distribution with mean as 0

and standard deviation is 1. The total area under its
density curve is equal to 1.
How to Calculate Probabilities?

• To calculate probability probability, our first step is

to convert the obtained Normal distribution to the
Standard Normal distribution.
• Find out Z Score.
Find Probability using table
Normal Distribution: Example

The Graduate Management Aptitude Test (GMAT), is widely

used by graduate schools of business in the United States as
an entrance requirement.
The scores of GMAT are normally distributed. Then probabilities
of achieving scores over various ranges can be determined.
In a recent year, the mean GMAT score was 494 and the
standard deviation was about 100. What is the probability that
a randomly selected score from this administration of the
GMAT is between 600 and the mean?
Normal Distribution: Example
Normal Distribution: Example
P(494 <= x <= 600/mean = 494 and sd = 100) = ?
To find probability first convert numbers(x) into z-
scores
Normal Distribution: Example
Normal Distribution: Example
Normal Distribution: Example
• For the same GMAT examination, what is the probability
of randomly drawing a score that is 550 or less?
P (x <= 550/mean = 494 and sd = 100) = ?
z score = (550-494)/100 = 0.56
Normal Distribution: Example
Normal Distribution: Example
Normal Distribution: Example

• What is the probability of obtaining a score

greater than 700 on a GMAT test?
P (x > 700/ mean = 494 and sd = 100) = ?
Z score = (700-494)/100 = 2.06
Normal Distribution: Example
Normal Distribution: Example
• What is the probability of getting a score between 350 and 450
on the same GMAT exam?
P(350 < x < 450/mean = 494 and sd = 100) = ?
z = (350-494)/100 = -1.44
z = (450-494)/100 = -0.44
Normal Distribution: Example
Normal Distribution: Example

• What is the probability of randomly obtaining a

score between 300 and 600 on the GMAT exam?
P(300 < x < 600/mean = 494 and sd = 100) = ?
Normal Distribution: Example

z score = (300-494)/100 = -1.94

z score = (600-494)/100 = 1.06
Normal Distribution: Example
 Sampling Distribution

The sampling distribution of a statistic (such as a

sample m ean or sample pro portio n) is the
distribution of all values of the statistic when all
possible samples of the same size n are taken from
the same population.
Sampling Distribution
Sampling Distribution Mean
The sampling distribution of the mean is the
distribution of sample means, with all samples having
the same sample size n taken from the same
population.
Example: Mean
Sampling Distribution: Variance

• The sampling distribution of the variance is the

distribution of sample variances, with all samples
having the same sample size n taken from the
same population.
Sampling Distribution: Proportion

The sampling distribution of the proportion is the

distribution of sample proportions, with all samples
having the same sample size n taken from the same
population.
Central Limit Theorem
• A theorem which describes the behavior of the
Sampling Distribution
• Statement: If samples of size n are drawn
randomly from a population that has a mean of µ
and a standard deviation of 𝛔, the sample means,
ẍ, are approximately normally distributed for
sufficiently large sample sizes (n>=30) regardless
of the shape of the population distribution.
Central Limit Theorem
 Central Limit Theorem
Conclusions:
• The distribution of sample means ẍ will, as the
sample size increases, approach a nor mal
distribution.
• The mean of all sample means is the population
mean µ.
• The standard deviation of all sample means is
𝛔ẍ = 𝛔/√n.
Central Limit Theorem
Sampling Distribution: Example

• Suppose that during any hour in a large

department store, the average number of
shoppers is 448, with a standard deviation of 21
shoppers.
• What is the probability that a random sample
of 49 different shopping hours will yield a
sample mean between 441 and 446 shoppers?
Sampling Distribution: Example

The problem is to determine P(441 <= x <= 446).

Sampling Distribution: Example
Sampling Distribution: Example

The probability of a
value being between z
= -2.33 and -0.67 is
0.2415.
 Statistical Inference
There are two types of statistical inference methods
—estimation of population parameters and testing
hypotheses about the parameter values.
Taxonomy of Inferential Techniques
Hypothesis testing
Parametric tests
 Estimation
• We can estimate the value of a population parameter in
two ways:
1. Point estimate - a single number that is our best
guess for the population parameter (or) by means of
2. Interval estimate - a range of values within which
we expect the parameter to fall.
• The probability that the interval contains the population
value is what we call the confidence level.
Point Vs. Interval Estimate
 Point estimate

• A point estimate is a single value (or point) used

to approximate a population parameter.
• The sample proportion ṕ is the best point
estimate of the population proportion p.
• For a population mean 𝛍, the sample mean ẍ is
a point estimate of.
Properties of Point Estimators

• A good estimator of a parameter has two

desirable properties
• Property 1: A good estimator has a sampling
distribution that is centered at the parameter it
tries to estimate. An estimator with this property
is said to be unbiased.
Properties of Point Estimators
Properties of Point Estimators

Property 2: A good estimator has a small standard

deviation compared to other estimators. This tells us
the estimator tends to fall closer than other
estimates to the parameter.
Properties of Point Estimators
 Point estimate

• If other random samples are taken from the

population, the point estimates derived from
those samples are likely to vary,
• Because of variation estimating with an
interval estimate is often preferable to using
a point estimate.
 Interval Estimation
• An interval estimate adds precision by giving an
interval of numbers around the point estimate - a point
estimate with a margin of error.
• A interval estimate is a range (or an interval) of values
used to estimate the true value of a population
parameter.
• Interval estimates contain the parameter with a certain
degree of confidence, they are referre d to as
confidence intervals.
Confidence level
• We are 95% confident that the
interval from LL to UL actually
does contain the true value of
the population proportion p.
• From figure: 19 out of 20 are
likely to contain the population
mean if a 95% confidence
interval is used
 Confidence level

• The confidence level is the probability (1 - 𝞪)

that the confidence interval actually does contain
the population parameter, assuming that the
estimation process is repeated a large number of
times.
• Most commonly used Confidence level is 0.95
Confidence level
 Critical Values
• It is the number on the borderline separating
sample statistics that are likely to occur from
those that are unlikely to occur.
• The number z𝞪/2 is a critical value that is a z
score with the property that it separates an area
of 𝞪/2 in the right tail of the standard normal
distribution.
Critical Values
Formula to find CI
 Hypothesis testing
• It is significance test to analyze evidence in
favour of the astrologers’ claim (Null Hypothesis).
• In statistics, a hypothesis is a statement about a
population, usually claiming that a population
parameter takes a particular numerical value or
falls in a certain range of values.
• If your data contradicts this claim then you
conclude that the claim made must be wrong.
Hypothesis testing
Steps of a Significance Test

• A significance test has five steps

• Step 1: Assumptions
• Step 2: Hypotheses
• Step 3: Test Statistic
• Step 4: P-Value
• Step 5: Conclusion
 Assumptions
• Each significance test makes certain assumptions
or has certain conditions under which it applies.
• First, specify the variable and parameter.
• The assumptions commonly pertain to
• the method of data production (randomization),
• the sample size, and
• the shape of the population distribution.
 Check assumptions
• These are the two assumptions researcher needs
to ensure before applying test
• The data should be collected from a population
that follows normal distribution of the
variable being quantified
• The samples drawn from the population must
follow homogeneity of variance
Quantile-Quantile Plot
• The quantile-quantile (q-q) plot is a graphical
technique for determining if two data sets come
from populations with a common distribution
(Gaussian, Uniform, Exponential or even Pareto
Distribution, etc.)
• It’s very useful to know whether the distribution
is normal or not so as to apply various statistical
measures on the data.
Q-Q Plot
Q-Q Plot
Q-Q Plot
Q-Q Plot
Q-Q Plot
Q-Q Plot
 Shapiro Wilk test
• The Shapiro Wilk test is the most powerful test when
testing for a normal distribution.
• The Shapiro Wilk will return the test statistic called W
and the P-Value.
• If the P-Value of the Shapiro Wilk Test is larger than
0.05, we assume a normal distribution
• In small samples these tests are underpowered to
detect quite major deviations from normality which can
be easily detected through graphical methods.
 Boxplot
• A quick way to get an idea about the variability
within each group is to use a boxplot.
• The variability within each group is represented
by the vertical size of each box; i.e., the
interquartile range (IQR).
• The boxplot shows that the variability is roughly
equal for each group.
Boxplot
 Levene’s Test
• Levene’s test is an equal variance test. It can be used to
check if our data sets fulfill the homogeneity of variance
assumption before we perform the parametric tests.
• The null hypothesis formed in the Levene’s test is that
the groups we are comparing have equal variance
• The test statistic of Levene’ test, W, its significance is
tested against the F-distribution using k -1 and N -k as
degrees of freedom and the chosen level of significance
α (usually 0.05 or 0.01).
Example: Scuba-diving
 Hypotheses
• Each significance test has two hypotheses about a
population parameter:
1. The null hypothesis is a statement that the parameter
takes a particular value.
• usually represents no effect - Nothing extraordinary
is going on
• We test this by assuming (or pretend) it is true and
reach a conclusion to either reject it or fail to
reject it.
 Hypotheses

2. The alternative hypothesis states that the

parameter falls in some alternative range of
values.
• represents an effect of some type.
• the symbolic form of the alternative hypothesis
must use one of these symbols: <, >, ≠.
Hypothesis
Example: Hypothesis
Example: Hypothesis
 Types of Hypothesis
Example
Researcher made an assumption on
Proportion of people attending sessions(p)
• Null Hypothesis
• H0: p = 0.5
• Alternative Hypothesis
• Two tailed test: Ha or H1: p ≠ 0.5
• Left tailed test: Ha or H1: p < 0.5
• Right tailed test: Ha or H1: p > 0.5
Hypothesis: Scuba-diving
 Types of Hypothesis
Example
Customer requires diameter of a cylindrical
bar is exactly 50mm(µ)
• Null Hypothesis
• H0: µ = 50mm
• Alternative Hypothesis
• Two tailed test: Ha or H1: µ ≠ 50mm
• Left tailed test: Ha or H1: µ < 50mm
• Right tailed test: Ha or H1: µ > 50mm
 Test Statistic
• The test statistic measures the distance between
the point estimate of the parameter and its null
hypothesis value.
• It is usually by the number of standard errors
between them - TEST STATISTIC
• Example: Mean of sample drawn = 49mm
z statistic = (49 - 50)/(sd) = ….
Test Statistic
Test Statistic
Test Statistic
Test Statistic
Test Statistic: Scuba-diving
Test Statistic: Scuba-diving
Test Statistic: Scuba-diving
Test Statistic: Scuba-diving
 P-Value/Critical Value

• To interpret a test statistic value, we use a

probability summary of the evidence against the
null hypothesis, H0.
• The Probability values/Statistic values of different
test statistics are obtained from corresponding
test statistic tables
 P-Value/Critical Value
• The P-value is the
probability that the test
statistic equals the observed
value or a value even more
extreme.
• It may not be of practical
concern, even though the
te st may be hig hly
signficant.
P-Value/Critical Value
 Significance level?
• When testing a null hypothesis, we arrive at a
conclusion of rejecting it or failing to reject it.
Such conclusions are sometimes correct and
sometimes wrong (even if we do everything
correctly). - Reason: Random Sampling
• It summarizes the two different types of errors
that can be made, along with the two different
types of correct decisions.
 Significance level?
Type I error is RTN: Reject True Null (hypothesis), whereas a
type II error is FRFN: Fail to Reject a False Null (hypothesis).
RouTiNe FoR FuN

•
Example: Courtroom
Example: Courtroom
Significance level-Decision
Significance level-Decision
• Type I error is equal to the significance level or the
P(Type I error/null hypothesis is true) is equals alpha.
• It seems to be tempting to just decrease the
significance level.
• If you decrease the p(Type I error) leads to increase
the p(Type II error) and vice versa.
• The power of a test is the probability of rejecting
the null hypothesis, given that it is false, or 1-p(Type
II error)
Significance level-Decision
P-Value: Scuba-diving
P-Value/Critical Value
P-Value/Critical Value
P-Value/Critical Value
P-Value: Scuba-diving
 Conclusion
• The significance level (denoted by 𝛂) is the probability that the
test statistic will fall in the critical region (more extreme or
occurs intentionally) when the null hypothesis is actually true.
• If the test statistic falls in the critical region, we reject the
null hypothesis
• We can define the confidence level for a confidence interval to
be the probability 1 - 𝛂.

• Common choices for 𝛂 are 0.05, 0.01, and 0.10, with 0.05 being
most common.
 Conclusion
• For a two tailed test
Conclusion: Scuba-diving
 Conclusion
• Fail to reject Null Hypothesis (ACCEPT)
• If P value > 𝛂/2 Value(Two tailed)

> 𝛂 Value(Left tailed)

> 𝛂 Value(Right tailed)

• Z/t/χ2/F value<z𝛂/2 value(Two tailed)

< z𝛂value(Left tailed)

< z𝛂 value(Right tailed)

• Otherwise, Reject the Null hypothesis

If Hypothesis is two tailed
P value of two tailed test
Conclusion of two tailed test
If significance level is 0.01
Conclusion
 Conclusion

Whether the calculated P value/Critical value

o ccu r s by chance (n ear to e stimate) o r
intentionally (extreme)?
How to fail to reject(accept) or reject a null
hypothesis(H0)?
• This is answered by the significance level (𝛂)
Example 2: Scuba-diving
Hypothesis: Scuba-diving
Hypothesis: Scuba-diving
Test statistic: Scuba-diving
Test statistic: Scuba-diving
Test statistic: Scuba-diving
P Value: Scuba-diving
Conclusion: Scuba-diving
If Hypothesis is two tailed
If Significance level is 0.01
Conclusion: Two tailed
Hypothesis testing: More detailed
Example 3: Scuba-diving
Hypothesis: Scuba-diving
Assumptions: Scuba-diving
Test statistic: Scuba-diving
Conclusion: Scuba-diving
 Z test: Testing a Claim About a
Mean: 𝛔 Known

• One of the most basic hypothesis tests is a test

about a population mean.
• A researcher might be interested in testing to
determine whether an established or accepted
mean value for an industry is still true or
• A researcher in testing a hypothesized mean
value for a new theory or product.
 Z test

• There are conditions that must exist for this test

to be meaningful.
• T h e p o p u lat i o n m u st fo llo w a n o r m a l
distribution
• The sample size is 30 or greater
• The population standard deviation is known
 z Test: Example
• People have died in boat accidents because an obsolete
estimate of the mean weight of men was used. Using the
weights of the simple random sample of men from Data.
• From data we obtain these sample statistics: n = 40 and x
= 172.55 lb.
• Research from several sources suggests that the
population standard deviation is given 26 lb. Use these
results to test the claim that men have a mean weight
greater than 166.3 lb.
 z Test: Example
Step 1: Assumptions
The three basic assumptions of a test about a mean
are as follows:
• The variable is quantitative and population mean is
claimed as greater than 166.3lb
• The data production employed randomization.
• The population distribution is approximately normal.
This assumption is most crucial when n is small
 z Test: Example
Step 2: hypotheses
• The null hypothesis in a test about a population mean has
the form
H0: µ = µ0 = 166.3lb
• where µ0 denotes a particular value for the population
mean. The right-sided alternative hypothesis
H1: µ > µ0 = µ > 166.3lb (Original Claim)
• Other possible alternative hypotheses,
H1: µ < µ0 = µ < 166.3lb and H1: µ ≠ µ0 = µ ≠ 166.3lb
 z Test: Example
Step 3: Test Statistic
The test statistic is the distance between the sample
mean x and the null hypothesis value µ0 as
measured by the number of standard errors
between them. This is measured by
z Test: Example
 z Test: Example
Step 4: P-Value/ Critical Value
This is a right-tailed test, so the P-value is the area
to the right of z = 1.52, which is 0.0643(1-0.9357).
 z Test: Example
Step 4: P-Value/ Critical Value
This is a right-tailed test, so the P-value is the area
to the right of z = 1.52, which is 0.0643(1-0.9357).
 z Test: Example
Step 5: Conclusion
• Because the P-value of 0.0643 is greater than the
significance level of 𝛂 = 0.05 (Assume), we fail to
reject the null hypothesis.
or
• Because the Z statistic value 1.52 is less than the
Critical value 1.645 (Z𝛂), we fail to reject the null
hypothesis.
 z Test: Example
Step 5: Conclusion
• The P-value of 0.0643 tells us that if men have a mean
weight given by µ0 = 166.3 lb, there is a good chance
(0.0643) of getting a sample mean of 172.55 lb.
• A sample mean such as 172.55 lb could easily occur
by chance.
• There is not sufficient evidence to support a conclusion
that the population mean is greater than 166.3 lb
 Testing the Mean with a Finite
Population
• If the hypothesis test for the population mean is
being conducted with a known finite population.
• The population information can be incorporated
into the hypothesis-testing formula.
• Doing so can increase the potential for rejecting
the null hypothesis.
z test: Known Population
• For all 600 sole proprietor CPAs working across the
United States, it is found that the average net income
is $74,914.
• An accounting researcher wants to test this figure by
taking a random sample of 112 sole proprietor
accountants and average net income is $78,695.
• Assume the population standard deviation of net
incomes for sole proprietor CPAs is $14,530.
z test: Known Population
Step 1: Assumptions
The three basic assumptions of a test about a mean
are as follows:
• The variable is quantitative and population mean is
claimed as equal to $74,914.
• The data production employed randomization.
• The population distribution is approximately normal.
This assumption is most crucial when n is small
z test: Known Population
Step 2: hypotheses
• Null Hypothesis - Population mean is claimed as
equal to $74,914.
H0: µ = µ0 = $74,914
• Alternative Hypothesis - Population mean is not
equal to $74,914.
H1: µ ≠ µ 0
z test: Known Population

Step 3: Test Statistic

z test: Known Population

Step 4: P-
Value/ Critical
Value
P-value =
1-0.9987 =
0.0013
z test: Known Population
Step 5: Conclusion
• Assume significance level as 0.05
0.0013 < 0.05
• Critical value with 0.05 significance level is 1.645
3.05 > 1.645
• Reject Null hypothesis - Mean is changes from its
claimed value
 z test: Summary
• To investigate the significance of the difference
between
• an assumed population mean μ/proportion and a
sample mean Ẍ/proportion. (variances known)
• t h e m e a n s /p r o p o r t i o n s /co u nt s o f t w o
populations. (variances known and equal/ not
equal)
 t test: Testing a Claim About a
Mean: 𝛔 Not known

• Very often when a researcher is gathering data

to test hypotheses about a single population
mean, the value of the population standard
deviation is unknown
• Then the researcher uses sample standard
deviation as an estimate of it.
• In such cases, the z test cannot be used.
 Testing a Claim About a Mean:
𝛔 Not known

• There are conditions that must exist for this test

to be meaningful.
• The population must follow a normal distribution
• The sample size is 30 or less
• The population standard deviation is unknown
(using sample standard deviation instead)
• The degrees of freedom (df) is n-1
 t test: Example
• We noted that people have died in boat accidents
because an obsolete estimate of the mean weight
of men was used.
• We obtain these sample statistics: n = 40 and x =
172.55 lb, s = 26.33 lb.
• Use these results to test the claim that men have
a mean weight greater than 166.3 lb.
 t test: Example
Step 1: Assumptions
The three basic assumptions of a test about a mean
are as follows:
• The variable is quantitative and population mean is
claimed as greater than 166.3lb
• The data production employed randomization.
• The population distribution is approximately normal.
This assumption is most crucial when n is small
 t test: Example
Step 2: hypotheses
• The null hypothesis in a test about a population mean has
the form
H0: µ = µ0 = 166.3lb
• where µ0 denotes a particular value for the population
mean. The right-sided alternative hypothesis
H1: µ > µ0 = µ > 166.3lb (Original Claim)
• Other possible alternative hypotheses,
H1: µ < µ0 = µ < 166.3lb and H1: µ ≠ µ0 = µ ≠ 166.3lb
 t test: Example

Step 3: Test Statistic

df = n-1 = 40-1 = 39
 t test: Example
Step 4: Critical Value
• Using this test statistic of t = 1.501, we now
proceed to find the critical value from t-
distribution Table. With df = 39.
• Assume significance level of 0.05 in one tail to
find the critical value.
tcritical = 1.685.
 t test: Example
Step 5: Conclusion
• Because the test statistic of , t < t𝛂 (1.501 < 1.685)

• Fail to reject the null hypothesis.

 t test: Example

Step 5: Conclusion - Interpretation

• Because we fail to reject the null hypothesis, we
conclude that there is not sufficient evidence to
support a conclusion that the population mean is
greater than 166.3 lb,
 t test: Summary
• To investigate the significance of the difference
between
• an assumed population mean μ and a sample
mean x̄. (variance unknown)
• the means of two populations. (variances
unknown and equal/unequal)
• the means of two populations. (No assumption is
made about the population variances)
Paired Comparison Tests
1. Two-Mean, Equal Variance t-Test - In a two-
mean, equal variance, t-test, the test is between 2
sample means ( Xbar1 versus Xbar2 ) when σ1
and σ2 are unknown, but considered equal.
2. Two-Mean, Unequal Variance t-Test - In a two-
mean, unequal variance, t-test, the test is between
2 sample means ( Xbar1 versus Xbar2 ) when σ1
and σ2 are unknown, but are not considered equal.
 Testing a Claim About a Standard
Deviation or Variance
• At times a researcher needs to test hypotheses about a
population variance.
• In some areas like statistical quality control, to make
conclusions validating claims on variance or standard
deviations are required rather than mean in those cases χ2
test is used.

• The χ2 test of this section is not robust against a departure

from normality, meaning that the test does not work well if
the population has a distribution that is far from normal.
 χ2 test
• The Chi-Square statistic is used in instances where
comparing the population variance becomes more
important than comparing population means. The two
most popular cases to apply chi-square is
1. Comparing variances when the variance of the
population is known, and
2. Comparing observed and expected frequencies of
test outcomes when there is no defined population
variance.
 Properties of the Chi-Square
Distribution
1. All values of χ2 are nonnegative, and the distribution is
not symmetric
2. There is a different χ2 distribution for each number of
degrees of freedom
 χ2 test: Example
• A manufacturing firm has been working diligently to
implement a just-in-time inventory system for its
production line.
• Ideally, the tubes would arrive just as the operator
needs them.
• However, because of the supplier and the variables
involved in getting the tubes to the line, most of the
time there will be some buildup of tube inventory.
 χ2 test: Example
• The company expects that, on the average, about 20
pneumatic tubes will be at the station. However, the
production superintendent does not want the variance
of this inventory to be greater than 4.
• On a given day, the number of pneumatic tubes piled
up at the workstation is determined eight different
times and the following number of tubes are recorded.
23 17 20 29 21 14 19 24
 χ2 test

Step 1: Assumptions
• The variable is quantitative and variance of tubes
are restricted to less than 4.
• Randomization, such as random sampling or a
randomized experiment
• The number of tubes is normally distributed.
 χ2 test
Step 2: hypotheses
• The null hypothesis is that the variance is
acceptable with no problems—the variance is equal
to (or less than) 4.
H 0: 𝛔2 = 4

• The alternative hypothesis is that the variance is

greater than 4.
H1: 𝛔2 > 4
 χ2 test
Step 3: Test Statistic
The observed chi-square value is calculated with s2
value is 20.9821 (from given data)

df = n - 1 = 8 - 1 = 7
 χ2 test
Step 4: P-Value/ Critical Value
• By assuming significance level as 0.05, Critical
value is
χ20.05,7 = 14.0671
χ2 test
 χ2 test
Step 5: Conclusion
• Because this observed chi-square value, χ2= 36.72,
is greater than the critical chi-square table value,
χ20.05,7 = 14.0671.

• The decision is to reject the null hypothesis.

• On the basis of this sample of eight data
measurements, the po pulation variance of
inventory at this workstation is greater than 4.
 χ2 test: Summary
• To investigate the difference between
• a sample variance s2 and an assumed population
variance σ 2.
0

• a population variance σ2and an assumed value σ 2.

0

• observed data arranged in K classes, and the

theoretically expected frequencies in the K classes.
• between K counts. (Poisson distribution)
 F-Test: Testing Hypotheses About
Two Population Variances

• Sometimes researchers are interested in studying

the difference in two population variance rather
than a mean or proportion
• It is most often used when comparing statistical
models that have been fitted to a data set to find
which best fits the population.
 F-test: Applications
• Suppose a manufacturing plant made two batches of an item,
produced items on two different machines, two different shifts.
To determine whether there is more variability in one than
another.
• Variance is sometimes used as a measure of the risk of a stock
in the stock market. The greater the variance, the greater the
risk.
• Apply to cases of comparing the precision of two measuring
devices, or the relative stability of two manufacturing processes.
 F test
• In testing hypotheses about two population variances, the
sample variances are used. This ratio of two sample
variances formulates what is called an F value.

where s12 is the larger

• An assumption underlying the F distribution is that the

populations from which the samples are drawn are normally
distributed. It is extremely sensitive to violations of the
assumption.
Properties of the F distribution

• The F distribution is not symmetric.

• Values of the F distribution cannot be negative.
• The exact shape of the F distribution depends on
the two different degrees of freedom.
• Statistic used if F, calculated by
 F-test: Example
Suppose a machine produces metal sheets that are specified to
be 22 millimeters thick. Because of the machine, the operator,
the raw material, the manufacturing environment, and other
factors, there is variability in the thickness. Two machines
produce these sheets. Operators are concerned about the
consistency of the two machines. To test consistency, they
randomly sample 10 sheets produced by machine 1 and 12
sheets produced by machine 2. The thickness measurements of
sheets from each machine are given in the table.
 F-test: Example
Whether the variance
from each sample comes
from the same population
va r i a n c e ( p o p u lat i o n
variances are equal) or
from different population
va r ia n ce s ( p o p u lat i o n
variances are not equal)?
 F-test: Example
Step 1: Assumptions
The three basic assumptions are as follows:
• The variable is quantitative and assume that two
machines are equally consistent
• The data production employed randomization.
• Assume sheet thickness is normally distributed in
the population.
 F-test: Example
Step 2: hypotheses
• Null hypothesis is variance of both machines are
same
H0: 𝛔12 = 𝛔22

• Alternate hypothesis is variance of both machines

are not same
H1: 𝛔12 ≠ 𝛔22
 F-test: Example
Step 3: Test Statistic
• The appropriate statistical test is
F = s12/s22 = (0.11378/0.02023) = 5.624
• Because n1 = 10 and n2 = 12, the degrees of freedom
numerator for the upper-tail critical value is
v1 = n - 1 = 10 - 1 = 9
• The degrees of freedom denominator for the upper-tail
critical value is
v2 = n -1 = 12-1 = 11.
 F-test: Example
Step 4: P-Value/ Critical Value
• Because we are conducting a two-tailed test, 𝝰/2
= .025.
F statistic, F0.025,9,11 = 3.59
• The lower-tail critical value can be calculated
from the upper- tail value by using formula
F0.975,11,9 = (1/F0.025,9,11) = 0.28
F-test: Example
F-test: Example
 F-test: Example

Step 5: Conclusion
• The observed F value is 5.62, which is greater
than the upper-tail critical value of 3.59.
• Thus, the decision is to reject the null
hypotheses. The population variances are not
equal.
 F-test: Summary
• To investigate the significance of the difference
between
• two population variances. (variance ratio test)
• the overall mean of K sub-populations and an
assumed value μ0 for the population mean.
• two counted results (Poisson distribution)
• To test the null hypothesis that K samples are from K
populations with the same mean. (analysis of variance)
ANOVA
More than two groups?
• We could compare the means against each other
using a “two sample t-test” which tests the
statistical differences between the means of two
groups.
More than two groups?
• This increases the chance of a Type 1 error.
• For example: setting the alpha value to 5% and
comparing 5 different groups gives us 10 pairs of
tests.
Not having type I error = (1-0.05)10 = 60%
• Uses the F-statistic to compare the variances of
the populations.
 ANOVA Assumptions
• All samples must be randomly selected
• The populations must be normally distributed
• The samples must be independent from one
another
• Each population must have the same variance
• N-Way means there are N factors that describes
the cause of the variation in the data
 ANOVA
• Null hypothesis in ANOVA
H0: All means are equal.
• Alternative hypothesis is
H1: All means are not equal
 ANOVA
• The ANOVA will test whether the means of all the
populations are equal for different levels of one factor.
• The “F-ratio” is used to do this.
 ANOVA
• “F-ratio” is the ratio: between groups variance / within
groups variance
• The F-ratio gets larger as the distribution overlap gets
smaller  a larger F indicates a difference in the group
means.
ANOVA Table
 ANOVA: Example
A process can be run at three temperatures for
conditioning, say 200o, 220o and 240oC. Using the below
data of four batches run randomly, determine whether the
temperature significantly affects the moisture content (%)
at a significance level of 0.05.
ANOVA Table
Test Statistic
 Conclusion
• The test statistic is = 8.85/0.37=23.92
• From the F-table, we can see that critical value of
F=4.26.
• Since this is one rejection region right tail test
and the test statistic exceeds critical value, so the
null hypothesis is rejected.
• At 0.05 level of significance, given data indicates
that temperature is impacting moisture content
One-way versus Two-way

• 1-way ANOVA - testing only one factor

• Example: the effect of hours spent studying on
exam grade
• 2-way ANOVA - testing two factors
• Example: the effect of hours spent studying and
the amount of sleep that the students received
the night before the test on their exam grade
One-way versus Two-way
One-way versus Two-way

• 1-way ANOVA - testing only one factor

• Example: the effect of hours spent studying on
exam grade
• 2-way ANOVA - testing two factors
• Example: the effect of hours spent studying and
the amount of sleep that the students received
the night before the test on their exam grade
One-way ANOVA
Thank You