values new variables from factor analysis for subsequent regression analysis

azad

New Member
#1
What values of factors should be used for further analysis like regression?
i am using a 5 point Likert scale in a questionnaire. i did factor analysis and found four independent variables (IV1, IV2, IV3, IV4) loading on factor1. Subsequently i want to do a regression analysis between factor1 (a new variable) and a dependant variable (DV1). The values of all IVs and DV1 are between 1 and 5. They are typed in columns for analysis. Now which values will i use for factor1 for regression analysis? One idea is to average the values of IVs where i get the averages like 4.3, 2.1, 3.8 etc. in a column. I can do regression analysis between this column and the values of DV1, which are stored in another column. However SPSS itself creates new factors (as new variables) adjacent to the columns of data during factor analysis. Can i do regression analysis between this column and DV1 ? The values of the factor1 given by SPSS is like 0.037, 0.988, 0.2633, -0.5838 etc. SPSS help file says that the factors can be used as new variables for further analysis. Could i use the above column values or is there any other procedure pending?
 

JohnM

TS Contributor
#4
A good reference for this is "Multivariate Data Analysis, 5th Edition" ISBN: 0-13-894858-5, authors: Hair, Anderson, Tatham, Black.

The chapter on Factor Analysis directly addresses your question, providing the reader with several possible approaches and the pros and cons of each.