Go back to DSc8450 home

Conceptual Issues in The Decision Process For Multivariate Analysis

Stage 1: Define the Research Problem, Objectives and MultivariateTechniques to be Used

Is the research exploratory or confirmatory?
First define the problem conceptually or theoretically.
Identify the dimensions or constructs of interest before specific variables.
Strive for model parsimony in selecting model constructs.

Stage 2: Develop an Analysis Plan

Distinguish between independent and dependent variables.
Define the types of relationships involved (linear vs. nonlinear).
Select specific variables as measures of the model constructs.

Stage 3: Evaluate the Assumptions of the Multivariate Technique

Ensure that the correct type of variable (metric or nonmetric) is used.
Understand the implications of each assumption.
Evaluate the need for meeting each assumption.

Stage 4: Estimate the Multivariate Model and Assess Overall Model Fit

Is the estimation controlled by the researcher or performed by the multivariate method?
What is the meaning of overall model fit?
Are there conceptual bases for model respecification if needed?

Stage 5: Interpret the Variate

What is the practical significance of the results?
Should the focus be on the variate or individual variables?
How do significant variables relate to the research objective?

Stage 6: Validate the Model

What limits in generalizability are due to data collection or sampling processes?
How do you interpret different results that may arise between estimation and validation samples?
How does generalizability relate to the research question?

Go back to DSc8450 home