Dear Kushneel,

To answer your questions:

1. Yes. Factor analysis treats the data as ordinal. It doesn't matter if the items have different meanings or numbers of values. However, in order to make the factor item scores comparable, I think it is good idea to convert to a scale with the same limits.

2. Yes. Save the regression scores and interpret the construct according to the item scores on its corresponding factor.

Dear Kushneel,

To answer your questions:

1. Yes. Factor analysis treats the data as ordinal. It doesn't matter if the items have different meanings or numbers of values. However, in order to make the factor item scores comparable, I think it is good idea to convert to a scale with the same limits.

2. Yes. Save the regression scores and interpret the construct according to the item scores on its corresponding factor.

Thanks Peter for the quick heads up. As for (2), lets say, if I regress my factor score as dependent variable and have log income as x-variable which gives me a coefficient of positive 0.45, How do I interpret this?

I agree with Peter. The answer to the follow-up question: Start by graphing the relationship, and interpreting the factor scores. You might need to rerun the factor analysis and use a rotation method to make interpretation of the factor scores easier.