How to identify redundant items from Modification Indices (MI) during the model fitting of AMOS? Is it recommended value more than 15 or more than 10 recommended?
Modification indices (MIs) could be used to identify potential cross-loadings and/or correlated error terms (indicating shared specific variance between indicators/items). The question, however, would be why you would want to eliminate "redundant" items. Typically, redundancy is a good thing for indicators that pertain to the same factor because redundancy means reliability of measurement in that context. Also, I would not stick to arbitrary numerical cutoff values because those are usually meaningless/not helpful.
Hey Thameera
The magic number is actually 3.84 - this is the critical value of the chi-square for 1 degree of freedom with p = .05. The thinking is that if you add a parameter to your model, this will 'cost' you 1 df, so you need some benefit - the benefit should be that the model chi-square decreases by at least 3.84 so that the model is better, in terms of the chi-square at p
- Badges
- Science topic
- Similar topics
- Mathematics