Factor analysis- varimax and correlations?

TheNibbler

A colleague has been working on a factor analysis to validate a questionnaire in a new patient group. We are both fairly inexperienced in this type of analysis and are trying to figure it out ourselves as we go.

In line with previous work she has used Varimax rotation (using SPSS). However it seems that previous papers looking at this questionnaire (and other papers of a similar type) report correlations between factors even when they state that they have used Varimax rotation. As I understood it, Varimax rotation doesn't allow for correlation between factors and indeed the correlations produced between our factors are 0. Is there any logical way to both use Varimax rotation and produce correlations between factors? I’m guessing I’m missing something obvious?!

Ostrom

You might be missing a step. You can report correlations between factors using summed scales once the factors have been identified, and it is common to report inter-item correlations before reduction. Rotation techniques are used to clarify the factor structure - the outcome is basically the same, you just (sometimes) get clearer separation. Is it possible you read a table of correlations between factors or factor scores in the paper?

TheNibbler

Thanks-that sounds logical. I'll have another look at it today and maybe come back if I'm still confused!