Sometimes, We have to eleminate items having poor score or loading to make the model fit. But if a latent variable have only 1 or 2 items after eleminating items, is it possible to publish the paper in renowned journals?
Harun,
whether your research gets published depends on a lot of different things, so I would not use that as my primary criterion.
Rather, I would try to understand WHY there is misfit in the model (e.g. by looking at the residual covariance matrix), and then discuss if there are theoretically plausible explanations for what is going on. Eliminating items is always an option, but only after understanding what went "wrong".
Heiko
Harun,
MISFIT can arise from various sources starting from the 'literature review' itself. If the variables, items, scales, population, sampling procedure etc are wrongly identified, it can also lead to problems. There can be human and methodological errors contributing to it. You may consider checking your data for possible errors and biasness.
You might also be aware that there are publications that recommends minimum three items per latent variable, so deleting items might not be a very good idea unless you have strong theoretical justification. Next, you can also check for 'modification indices' suggested by software such as AMOS, LISREL etc.. However, you must have a clear understanding and reasoning before accepting any of the modification indices in your model.
About publication, it depends on the journal you are trying to publish. They have there own criteria for acceptance. Good Luck with your research
- Badges
- Science topic
- Similar topics
- Linguistics
- Composition Studies
- Publisher