I conducted exploratory factor analysis (EFA) on 69 variables and sample of 346 (1:5 variable to sample ratio). I used Principal Axis Factoring (as it is the most used extraction method for common factor analysis) with Promax rotation (initially I used Direct Oblimin, but some of my factors had negative correlations with my variables, hence I used Promax). I set values for factor loadings above 0.3.

The problem that I am facing is that the number of factors on the basis of Eigenvalue (>1) is 18 which is too high and explains only total 49% variance. My scree plot shows 2 (or 3 at most) factors. For 2 factors, my total variance is 26% and for 3 factors 32 %. Even when I delete items on the basis of communalities (>0.3) and repeat the procedure for both 2 or 3 factors, my total variance still remains low (<50%). Can you suggest what could possibly be done to increase my total variance for minimum number of factors as possible?


1 Answer 1


Collect new data?

You either have variables that are poor indicators of the constructs you are interested in, or you have variables that are measuring too many constructs.

These are your results. You don't like them. Your results are determined by your data, not by what you hoped to find. But that's OK. We usually don't get the results that we wanted or liked. (If we did, we wouldn't need to collect data and analyze it, and our lives would be much simpler.)


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.