The original scale has 4 subscales, and I will use it as one of my essential outcome measures. I need to use 2 subscales only to assess specific changes.
Dear Tasneem Alatwi,
Concerning your question, I invite you to read "Franke, G. H. (1997). "The whole is more than the sum of its parts": The effects of grouping and randomizing items on the reliability and validity of questionnaires. European Journal of Psychological Assessment, 13(2), 67–74. https://doi.org/10.1027/1015-5759.13.2.67"
Findings seriously call into question the admissibility of subscale extraction for self-report inventories.
It's acceptable, but the scholarly justification (your study questions and objectives, the lack of alternative suitable scales, preference of the sound psychometric properties of the present sub-scale to others, and so on) must be clearly stated. It's been done in a lot of studies.
I agree with Narendra Nath Samantaray . As long as the subscales are valid for what you want to measure and also sufficiently reliable, this can be fine. Notice that the reliabilities of the subscales tend to be lower because they consist of fewer items than the overall score. You should make sure that the reliabilities of the subscales are sufficiently high for your purposes and/or use a statistical technique that allows you to correct for measurement error (unreliability) in your analyses of the subscale scores such as confirmatory factor analysis or structural equation modeling.
- Badges
- Science topic
- Similar topics
- Geoscience
- Asia
- India