Using Factor Scores in Discriminant Analysis

#1
I am doing research on whether corporations that engage in earnings fraud can be identified up to three years before the fraud occurs. My end result will be a discriminant analysis that can be used to look at the likelihood of a company engaging in earnings fraud.
I have three years of financial data for 29 matched pairs. I wanted to decrease the number of variables I have, so I did a factor analysis. I came up with three variables: DEBT, INCOME and CASHFLOW.
Now, I want to do discriminant analysis using these variables. My question is how to do that because each variable has several components. For example, CASHFLOW contains operating cash flows, investing cash flows, and financing cash flows. Do I use an average of these in my discriminant analysis?
Everything I can find says that factors scores can be used in later analyses, but they don't say exactly how. Help!
 

Miner

TS Contributor
#2
Your software should have some provision for storing the factor scores. Once these have been stored they become the data, you can use each factor as a variable in the discriminant analysis.
 
#3
Thanks for your quick reply. I should clarify my issue; I think it's less related to SPSS than it is to statistical procedure. Here's what happens:
1. I do discriminant analysis with a ten financial values as independent variables and TYPE as my grouping variable (engaged in earnings fraud or not).
2. I want to decrease the number of variables used, so I run a factor analysis, which identifies three variables.
3. Now I want to use those three variables to redo the discriminant analysis. How do I do that? What are my independent variables in the discriminant analysis?
 

Miner

TS Contributor
#4
Run your factor analysis in SPSS and store your factor scores. These become your data. There will be factor scores for each of the three factors. The three factors become three IVs.

Run your discriminant analysis on the factor scores for the three factors.