I want to test the moderating role of a categorical variable on two hypothesized relationships. There are just these two relationships in the model. The sample size is 323. The sample distribution for the two categories in the moderating variable is 172 and 151 respectively. The invariance analysis results show that loading & weight of the construct measurement model do not differ significantly within the model. Therefore, MGA could not be performed.
I have two queries:
1. If I just report the result of invariance analysis in my paper, will it be acceptable in a journal.
2. What alternative tests can I apply to test the moderation?
Thanks for your support
Hello Najmul,
the invariance of the factor loading is a mere prerequisite for comparing issues on the latent level. Your invariance test shows that this is fulfilled. That's good.
The invariance of the structural effect reflects your theoretical moderator effect. Your invariance test shows that the groups don't differ. That's science.
It may be that reviewers will be picky of the non-sign. difference as journals like significance in the same way as little children like toys and ice cream. You will know.
What is more critical however, that your overall model only consists of two effects, and not actions to control for potential confounders had been undertaken. THIS is what I as a reviewer would taks to reject the paper. Perhaps you have indeed measured reasonable controls. If yes include them.
Take a look into this thread for further information
https://www.researchgate.net/post/What_is_the_role_of_control_variables_in_SEM#view=557d9d275cd9e3a0a58b45b6
Best whishes
--Holger
@Najmul Invariance insignificance at measurement level doesn't forbid you to apply MGA at structural level. Seeing ay your sample distribution I would encourage you to go for MGA and check invariance at structural level.
- Badges
- Science topic
- Similar topics
- Social Science
- Media
- Social Media