For my study, it was developed a questionnaire to explore clinician’s therapeutic skills. It has 25 items. I run an EFA with Varimax rotation using SPSS and it seems loading for 4 factors. However, when doing a parallel analysis using JASP it seems loading for 2 factors. I am more prone for the 2 factors structure. What are your thoughts?
N = 194
Chronbac’s Alpha= .947
McDonald omega: 0.948
Mean=98.67; SD 14.597
Kaiser-Meyer-Olkin= .939
Barlett. Chi-square test= 2954.375
No item correlate greater then 0.8
I already eliminated factors that load highly for more than 2 factors, and eliminated items that load for less than .32 (Tabachnick & Fidell, 2014)
SPSS EFA Varimax rotation. Four factors:
Loading: 1st factor 45.28% of variability; 2nd factor 7.46%; 3rd factor 5.71%; 4th factor 4.01%
JASP Parallel Analysis Varimax rotation. Two factors:
Loading: 1st factor 8.08%; 2nd Factor 4.09%
Dear Sandro Voi ,
EFA is quite a complex technique as there are many things that you need to balance. It is quite usual to run your EFA repeatedly and to need to remove about half the items in your questionnaire for your solution to stabilise. It is also quite possible to end up with different results with different approaches. I would recommend starting with an orthogonal rotation and only attempting an oblique rotation if it fails to identify factors satisfactorily. The choice of orthogonal rotation depends upon whether you plan to analyse the data you as well as to validate your structure. Please see my study guide for more information.
If you have two more factors recommended promax and parallel analysis, look at the covariance matrix presented by JASP, if you have a pre-determined structural model you would recommend CFA.
- Badges
- Science topic