I have a scale with 67 items for which i have run EFA. i am confused which values of factors loading to take for reporting Component Matrix or Rotated Component Matrix?
Hi Amna Farzand Ali ,
Simple answer is: if your measurement tool's factor structure unidimensional, you cant use Rotated Component Matrix because there are no factors. You can use Component Matrix. If factor structure ise multidimensional and if you used one of rotation techniques, yes you must report Rotated Component Matrix.
I suggest you pay attention to two issues. The first is to use the correct factor exraction technique (principal component analysis is not factor analysis!), Second, you should choose the correct rotation technique.
Good luck with your research
Hello Amna,
Hakan Koğar is correct on all points. To his response I would add this: If you have multiple factors and use an oblique rotation (allowing for factors to correlate), then the factor pattern and factor loading/structure matrices are different, and each has importance. With orthogonal rotation, the factor pattern and factor loading/structure matrices are one and the same, and you'll only need to report it.
Good luck with your work.
La estructura dimensional de la Escala de Actitudes hacia la Estadística de Auzmendi, 1992 en su aplicación a estudiantes de 2 y3semestre de Psicología que cursen actualmente la materia de análisis de datos de la universidad Piloto de Colombia. Según los datos obtenidos con una muestra de 109 participantes de ambos sexos que fueron seleccionados por censo, se concluye que no son estimables las estructuras dimensionales propuestas por Auzmendi. Al mismo tiempo, se realizó una solución factorial basada en cuatro dimensiones y quince ítems, con capacidad para explicar el 63,29% de la varianza del instrumento y con una fiabilidad alfa de Cronbach igual a 0,846. Palabras clave: análisis factorial, escala, estadística, medición de actitud, validación
Thanks for your response Hakan Koğar my scale is multidimensional.
As far as you have recommended to use correct factor extraction technique other than PCA.
My question is why most of the researchers are using it if isn't a factor extraction technique? as i have applied PCA with varimax the results are quite different than other as promax. why it so? please guide.
My sample size is 405 participants with 67 items. I divided my scale in 8 major categorizes but when i apply EFA it gives quite different results almost 17 factors. where is the problem.
Hi again Amna Farzand Ali ,
Using PCA at FA is simply misusage of an statistical analysis. In this case, I think it is important that PCA takes place by default in the factor analysis tab in SPSS.
As David Morse mentioned, varimax and promax are 2 different rotation technique. If the correlation between latent variables is high, you should use a oblique rotation technique such as promax.This is all about factor correlations.
If you think that the measurement tool has a factor structure of 8 factors and 67 items as you mentioned, then it may be better to use a confirmatory statistic. Like CFA. But for this you need the theoretical basis of the factor structure and / or previous research findings.
Good luck
- Badges
- Science topic
- Similar topics
- Mathematics
- Algebra
- Matrix