Multivariate Statistical Analysis: Professor Dr. Muhammad Mohsin Butt Department of Marketing School of Business Studies

CHAPTER 24:
MULTIVARIATE Professor Dr. Muhammad

Mohsin Butt
STATISTICAL Department of Marketing

School of Business Studies
ANALYSIS
TEXTBOOK: BUSINESS RESEARCH METHODS, 8TH
EDITION, ZIKMUND, BABIN, CARR AND GRIFFIN
LEARNING OUTCOMES
CHAPTER 22
1. Understand what multivariate statistical analysis involve and
its types
2. Interpret results from multiple regression.
3. Interpret results from multivariate analysis of variance.
4. Interpret basic exploratory factor analysis.
WHAT IS MULTIVARIATE DATA
ANALYSIS
1. Research that involve three or more variables, or with the underlying
dimensions among multiple variables will be dealt through multivariate
analysis.
2. Variate is a mathematical way in which a set of variables can be represented
with a single equation.
3. For example, we learn computing variate by combining the score of each item
related to a single variable, but we did without counting for the relative
weightage of each indicator.
4. In other term you might say how much variance is explained by an indicator and
how much was unexplained.
CLASSIFYING MULTIVARIATE
TECHNIQUES
• Two basic groups of multivariate techniques are dependence method and
interdependent methods.
• The dependent technique looks at explain one or more dependent variables.
Multivariate regression and MANOVA are common techniques.
• The interdependence technique is to look at variables as groups and no distinction is
made between dependent or independent variables. Factor analysis, cluster analysis
and multidimensional scaling are some of the common techniques.
MULTIPLE REGRESSION
ANALYSIS
Multiple regression is analysis of association in which the effect of two or more than two
independent variables on a single dependent variable is examined simultaneously.
Dummy variables are used to code the different levels of dichotomous variables.
In simple regression beta coefficient was the slop of X on Y, and because we fitted only one
straight line, we had only one beta coefficient.
In multiple regression we are fitting as many straight lines as many independent variables
in our model and thus we have a unique beta coefficient for each independent variable.
In multiple regression R2 explained the combined effect of all the independent variables.
In other words, the sum of individual R2 of each independent variable will be slightly
higher /equal to R2 by the combined model.
The statistics is used to predict the overall model significance by comparing the variance
explained by the predictors (sum of square for regression)and the sum of square of error.
INTERPRETING MULTIPLE
REGRESSION MODEL
1. Examine the model f test, if the model is not significant, no need to further
looking into anything, it means that your independent variables are not
predicting anything. May the relationship between your variables is not linear.
2. Look at the standardized and unstandardized beta coefficients for each
independent variable.
3. Look at R2 value to see how much variance in dependent variable is explained by
the independent variables. It is explanatory information, not prediction.
4. The predictive power of regression model comes from beta coefficients of its
independent variables and the error term associated with it.
5. Always opt for multicollinearity diagnosis
ANOVA (N-WAY) AND
MANOVA
This technique looks at multiple independent variables and their impact of multiple
dependent variables.
The interpretation of N-way or MANOVA is similar to multiple regression.
We first look at F statistics of overall model fit. A significant p value indicate the
model fits with our data.
Look at individual f test for each independent variable.
For categorical independent variables look for the mean value.
For each continuous variable look for beta values.
The multivariate test are followed by univariate test (such as t-test, or one way
ANOVA to look at the effect of each variable.
EXPLORATORY FACTOR ANALYSIS,
CFA AND FULL STRUCTURE MODEL
Let's, learn how and when to use EFA, CFA and Full SEM.
1. EFA
2. CFA (used to establish validity and reliability of a scale).
3. Full structural Model (running regression analysis but this time the model
include and measure error term).

Multivariate Statistical Analysis: Professor Dr. Muhammad Mohsin Butt Department of Marketing School of Business Studies

Uploaded by

Copyright:

Available Formats

Multivariate Statistical Analysis: Professor Dr. Muhammad Mohsin Butt Department of Marketing School of Business Studies

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Multivariate Statistical Analysis: Professor Dr. Muhammad Mohsin Butt Department of Marketing School of Business Studies

Uploaded by

Copyright:

Available Formats

CHAPTER 24:

MULTIVARIATE Professor Dr. Muhammad

STATISTICAL Department of Marketing

You might also like