I have a scale with 7 subscales. I wish to remove 2 subscales (before I run my CFA) since they are not of interest in my research, and they sort of overlap with another factor in my full-fledge SEM model. Can I remove them on such bases?
hi
becasue you wish is not enough reason to remove the 2 subscales.
but if the literature shows overlapp or low relevance to the main concept then yes, and you can run the CFA to find out the indeces if it support the new structure of the scale.
Thanks Prof Muayyad for your answer.
I have ran CFA on the new structure, and it seems that the indices showed moderate fit, similar to the indices that I got when running the CFA with all 7 subscales.
Since literature does not show an overlap (the other factor is my proposed mediator that has not exactly been proposed before) or low relevance of the subscales to the latent factor of interest, then I have no basis to remove the subscales, is that right?
Your advice is very much appreciated.
in every study, choosing specific subscales to fit the purpose, the research quesiton or the hyposthses is controlled by the researcher. so you dont have to keep the 7 factors in the scale. you can select the other 5 factors which fit your study only.
in any instruments, we must make sure that the scales and sub-scales used are valid (meaning it measures what we want to measure). therefore I say yes, I do that at times.
This could happen Especially when you are adapting instruments invented in other cultures where your culture is by nature different from the culture where the instrument was originally invented (for ezample your culture could be more collective by nature, and the culture of the original instrument was more individualist).
(for ezample: emotional intelligence in the USA vs emotional intelligence in Malaysia)
Or it could also happen when the construct you want to measure is in slightly different context from the context of the original instrment
(for example job satisfaction in a public services, vs job satisfaction in a profit-oriented companies)
I say yes, when that is needed. your focus should be VALIDITY - it measures what you want to measure.
Thanks Prof Muayyad, that helps!
Thanks Mai! I agree. Appreciate your answer.
- Badges
- Science topic
- Similar topics
- Psychology
- Emotional Psychology