Multivariate Analysis

Multivariate Analysis
The term “multivariate statistics” is
appropriately used to include all statistics
where there are more than two variables
simultaneously analysed.
Multivariate analysis (MVA) techniques
allow more than two variables to be
analysed at once
Multiple regression is not typically included
under this heading, but can be thought of as
a multivariate analysis
….is concerned with data that consists of
sets of measurements on a number of
individuals or objects. The data include
simultaneous measurements on many
variables.
Multivariate thinking
The essence of multivariate thinking is to
expose the inherent structure and
meaning revealed within these sets of
variables through application and
interpretation of various statistical
methods.
  Example

Hypertension depends on age, gender,

cholesterol level, food habit, smoking etc.
 Objectives of MA

Data reduction or structural

simplification: without sacrificing
valuable information, it makes
interpretation easier.
Sorting and grouping
Investigation of the dependence
among variables
Prediction
Hypothesis construction and testing
 Differences between Univariate
and Multivariate Analysis
Univariate means that assuming the
response variable is influenced by only
one other factor.
Example: SAT scores are influenced by
GPA. You may be looking to test the
hypothesis that as the GPA rises, the SAT
score rises. Here you are assuming that
the GPA is the only factor influencing the
score of SAT.
Multivariate means assuming the
response variable is influenced by multiple
factors (and even combinations of factors).
Example: you might assume that SAT
scores are influenced by GPA and sex
(male or female). This type of analysis
also leads to possibilities of cross terms,
meaning there is some effect of being
male and having a certain GPA on the
SAT scores vs being female and having a
certain GPA.
  Application of MA
1. Data reduction or simplification
Using data on several variables related to
cancer patient responses to radiotherapy:
number of symptoms such as sore throat
or nausea, amount of activity (on a scale
of 1-5), amount of sleep (on a 1-5 scale),
amount of food consumed (on a scale of
1-3) etc, a simple measure of patient
response to radiotherapy can be
constructed. (See Exercise 1.15.)
Track records from many nations can be
used to develop an index of performance
for both male and female athletes.

2. Sorting and grouping

Data on several variables related to
computer use can be employed to create
clusters of categories of computer jobs
that allow a better determination of
existing (or planned) computer utilization.
Measurements of several physiological
variables can be used to develop a
screening procedure that discriminates
alcoholics from non-alcoholics.
3. Investigation of the dependence
among variables
Data on several variables can be used to
identify factors that were responsible for
client success in hiring external
consultants.
Measurements of variables related to
innovation, on the one hand, and variables
related to the business environment and
business organization, on the other hand,
can be used to discover why some firms are
product innovators and some firms are not.
4. Prediction
• The associations between test scores, and
several high school performance variables,
and several college performance variables
can be used to develop predictors of
success in college.
cDNA microarray experiments (gene
expression data) are increasingly used to
study the molecular variations among
cancer tumors. A reliable classification of
tumors is essential for successful
diagnosis and treatment of cancer.
5. Hypotheses testing
Several pollution-related variables can be
measured to determine whether levels for a
large metropolitan area were roughly
constant throughout the week, or whether
there was a noticeable difference between
weekdays and weekends. (See Exercise
1.6.)
Experimental data on several variables can
be used to see whether the nature of the
instructions makes any difference in
perceived risks, as quantified by test scores.
 The Organization of Data
 Variable1 Variable2 ... Variablek ... Variable p
Item1: x11 x12 ... x1k ... x1p
Item2: x21 x22 ... x2k ... x2p
.
.
.
Item j : xj1 xj2 ... xjk ... xjp
.
.
.
Itemn: xn1 xn2 ... xnk ... xnp
Or we can display these data as a
rectangular array, called X, of n rows and
p columns:

 x11 x12 ... x1k ... x1p 

 
 x21 x22 ... x2k ... x2 p 
 . 
 
 . 
 . 
X  
 x j1 x j2 ... x jk ... x jp 
 
 . 
 . 
 
 . 
x xn 2 ... xnk ... xnp 
 n1

The array X, then, contains the data consisting of all the observations on all of
the variables.
 Descriptive Statistics
Sample Mean
Sample Variance

The square root of the sample variance, is

known as the sample standard deviation.
Sample Covariance
Sample correlation coefficient
 Properties of Sample correlation
coefficient
Graphical Presentation
Scatter Plot
Dot Plot
Scatter Plot
A scatter plot (also called
a scatterplot, scatter graph, scatter
chart, scattergram, or scatter diagram) is
a type of plot or mathematical
diagram using Cartesian coordinates to
display values for typically two variables for
a set of data.
 Purpose: To identify the type of
relationship (if any) between two
quantitative variables