Hierarchical model - interpretation of FOC contributions

[lottevdheuv]

Hi!

I'm using the repeated indicators approach to estimate four hierarchical models with the same exogenous constructs but different endogenous constructs. The exogenous constructs include one reflective-formative construct (i.e. place attachment) and one formative-formative construct (Ii.e. place meaning). There is also one reflective exogenous construct (i.e. centrality-to-lifestyle). The endogenous constructs are intentions for four different types of behavior. I chose to estimate a separate model for each behavior type rather than create one big model with all behavior types as endogenous constructs, because my sample size is small (N=119). However, I did run the big model out of curiosity.

What I found difficult to interpret is that the contributions/paths of the FOC constructs (i.e. types of place meanings) to the SOC (i.e. overall place meaning) were of different sizes for each of the four models (e.g. in one model meanings related to natural beauty had the highest contribution, while in another model meanings related to social safety scored the highest). All contributions in all models were significant. How can this happen, and how do I interpret the differences in these contributions between models? Is there a way to assess whether these contributions are statistically different?

Another interesting result was that, while the path "centrality -> place meaning" was positive and significant in all of the separate models, this relationship was insignificant (and negative) in the big model.

Thanks in advance for any guidance you can provide regarding these questions! It would also be helpful to hear your thoughts on whether I should keep the four separate models or rather report on the big model only.

Best wishes,
Lotte