The factor analysis that I've created is a rotated PCA.
When you run the factor analysis you have the option to save the factor scores. SPSS will create new variables at the end of the data set (one for each factor) that contain each factor score. What I'm trying to do is to be able to create these scores manually.
My assumption was that I could have SPSS export the "Factor Score Coefficient Matrix". From this matrix, I could get the coefficient for each variable in the factor solution and then create the equations to calculate each factor score. My reasons for doing this is that I need to create these factor scores on another data set using SQL.
When I follow this process (using the factor score coefficient matrix) I cannot match the scores that are created automatically by SPSS. Not sure what I'm missing, although I wonder if the rotation has something to do with it?
Main Topics
Browse All Topics





by: richdiesalPosted on 2009-01-13 at 11:28:37ID: 23366245
Your question is a little vague... a factor analysis will produce different output depending on what kind of factor analysis it is - principle components, maximum likelihood, etc.
Can you clarify "score the solution"? Are you trying to produce a fit index? Or are you just trying to look at unrotated/rotated factor loadings and trying to produce a solution?