Dear Researchers,
I am analyzing a research model of brands in the consumer behavior perspective in PLS-SEM by using SMART PLS 3 software. I am facing some problems as following:
Research model
IV (4 reflective constructs) => Mediator => DV with 3 constructs (1 moderator with higher order 3 reflective constructs moderate between IV & med. And med. & DV.
1. In the research model, IV has 4 reflective constructs as a higher-order construct, but due to the validity and reliability issue, I have removed 1 construct. Now after analysis, IV has 3 constructs. Is it ok and acceptable?
2. Moderator analysis between IV & mediator, results are path coefficient -0.072 T value (2.447) P-value (0.015). As mentioned path coefficient value is negative but results shown significant. I am not understanding this myth. I need your guidance about this logic and results interpretation.
3. I directly performed moderation analysis with the two-stage method. Can you please guide how to perform moderation analysis with higher-order constructs?
Looking forward to your prompt and positive response
Thank you
Asif Ali Safeer
When the moderator is negative he tends to be a false positive so his model is questionable should be done in parts would be more recommended
Hi. the effect size need to checked out. sometimes the moderator effect size is too small. Thanks.
Hi,
See this link, it is useful for your issue
https://www.smartpls.com/
Best Regards
Hi,
1. Validity and reliability issue indicate that second order scale is not well compatible with your data. Consider using first order construct if conceptually possible or single order construct scale exist in literature.
2. Negative moderation in not a problem if conceptual justification exist. However , in your case moderation value is though significant but it is quite low. Its showing very weak moderation effect.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS