Hi all,
Need your suggestion on my research.
I am working in CFA which is derived after EFA with 2 factors having 7 and 8 items, but issue is in CFA, AVE is coming less than 0.5 hence I am forced to delete the items with less than 0.4 standardized beta. But now only 4 items remain in the each factor {and all model fit indices , AVE and CR is in acceptable range }against 7 and 8 which where derived from EFA.
Is this acceptable to delete so many item after EFA?
Note : EFA and CFA is done on different data.
@Uday For appropriate model identification in covariance based CFA, there should be a minimum of three observed variables under each latent variable which is met in your case. Now when it comes to the query whether it's appropriate to delete so observed items from the model or not. It depends on whether the observed variables left under each latent variable are able to capture the construct as per it's operational definition or not. If they do the justice then it's fine and you can go ahead with the model. Wishing you best with your work.
Uday,
1) how is the model fit in terms of chi-square, degrees of freedom and what's the N?
2) More background on the question wordings and supposed meaning of the latent factors would be helpful
3) How are the loadings?
IMHO, things such as a AVE are side-criteria used to cover behind some more fundamental issues.
Beyond that, as Imran noted, you don't need much indicators (although more are better AS LONG AS the structure is valid), however, deleting indicators provides the chance to mask misfit. Hence, the model should be further tested, for instance by incorporating further variables which should be related to the factors IF they are what the suppose to be. Often a re-specified model from the first round blows again up when incorporating further variables.
Best,
Holger
You need a minimum of 3 items for each factor but the question is not only statistic now: are your 3 items being able to capture the construct your instrument is supposed to measure? If yes, you can consider reduce the number of items. If not, you may inform about how much is affected the contruct whit that results (trying to understaind why some items loadings are not so good in your specific sample). You can review, differences in sample in EFA and CFA, item-content overlapping, etc. In all cases you are not confirming the goodness of your model configuration with your new sample. Best
- Badges
- Science topic
- Similar topics
- Engineering
- Modeling