Hi! Anything to throw light to the following question would be appreciated: I'm conducting SPSS macro PROCESS, Model 8 (Hayes, 2018) to test my moderation model. PROCESS allows to test a single dependent variable at a time. My dependent variable contains two higher order factors and each factor has four subdimensions. I am confused if I can combine the items of each dimension ( treat them as a single factor) and create a single factor to test my hypotheses in PROCESS? If yes, is there any criteria on which I need to base it?
Dear Gulandom, when in presence of more than one dependent variable the method of choice is canonical correlation analysis (https://www.statisticssolutions.com/canonical-correlation/)
where you can check immediately the relative importance of the different factors of the dependent Y in relation to the regressors...
Considering the factorial information you have, why are you running moderation and not an SEM? I guess, you could combine the items on each dimension and run it as a single factor. However, you have higher-order factors that may account for some of the variance/effect. With SEM, you can find out how much each factor loads on your variable.
Thank you so much for all the answers! I'd like to be clear on a couple one things: my independent and moderating variables have single factors while my mediating and dependent variables each have subdimensions. The point is my research focuses on the factors from a general perspective rather emphasizing the the effects of each subdimension. Thus, I decided to combine the items. I'm new in this area so I want to make sure if I'm in the right path.
And dear Kenny Borgas , I recently started working on SEM, I'm not quite familiar with it yet. Would it be correct if I run a confirmatory factorial analysis on SEM to check the reliability of the items and them move on with PROCESS to check my hypotheses?
No. CFA is used to test/confirm a theoretical framework that accesses whether your proposed model (factor structure) can be replicated. In other words, it tests the fit between your observed data and your theorized model that specifies the hypothesized causal relations between your latent factors and their observed indicator variables.
- Badges
- Science topic
- Similar topics
- Counseling Psychology
- Confusion