Factor Analysis Transformations: Thesis Help Needed Please!

#1
Hi

I'm really hoping that someone will be able to help me?

I am currently in the process of running a factor analysis. The individual items were quite badly skewed/kurtotic, so I have transformed them and run the exploratory FA.

I ran the FA both before and after transforming the variables, and the transformations had virtually no effect on the factor weightings.

However, for reasons I can't seem to get my head around, the factor weightings have changed direction (i.e. if there was originally a positive weighting on one factor it has now changed to a negative weighting).

I would understand why this had happened if the effect was only happening on variables that had been negatively skewed and therefore reflected (so that the signs would change from positive to negative) but this is not the case. The variables that have changed direction appear to be totally random.

My problem is that I don't know what to report in the main document.

The weightings themselves haven't changed, so I'm tempted to say that I did the transformations and it didn't affect the factor structure, therefore the raw data weightings will be reported rather than the transformed FA results. But I don't know if this is acceptable.

I don't know if I've explained the problem clearly enough so I apologise if it's not clear. I am really struggling with this and working to a tight time pressure so if anyone could offer me any help at all I'd really be very grateful???

Thanks in advance!

Kelly.
 

JohnM

TS Contributor
#2
"I ran the FA both before and after transforming the variables, and the transformations had virtually no effect on the factor weightings."

Based on this statement, and the fact FA is an exploratory method, I think you can just use the before-transformation loadings....