Hello everyone, I’m now working on my thesis about work-family/life interface.
I got some statistical issues that Ive nearly heard it before, so I’m here to raise the question and hope someone could give the suggestion.
I’ve got and 2 mediators which mediate X-Y by dual routes. Besides, there are 2 moderators which moderator1(w) possibly would moderate x-m1 & x-m2 at first stage, the others(z) would moderate m1-y & m2-y. Forming the model is called moderated mediation.
First,
moderator1(w) is a moderator which is assessing by 2 concepts, and I looked up the literature by Kreiner(2006) suggested that it could use latent congruence model by SEM approach. And others recommended response surface regression by polynomial regression.
By reading articles, I have no idea how to use these methods to fit in my moderated mediation model. Does anyone have met such questions?
Secondly,
the x somehow could have the curvilinear effect on Y which totally mediated by m1 & m2. Is it even possible put x quadratic into latent congruence model or response surface regression into analysis the moderated mediation model?
Typically, you would fit such a model within the framework of path analysis/structural equation modeling (SEM), which can handle mediators (indirect effects), moderators (interaction effects), and non-linear (e.g., quadratic) terms. Most books/chapters/papers on SEM will include a discussion of path analysis.
There's an article which is written by Su et al., (2019), and they mentioned few advantages of using SEM dealing with congruence issues compared to polynomial regression.
The above suggestions are reasonable. However, testing latent congruence has some problems including multi-correlinearity issues. refer to the following article.
Edwards, J. R. (2009). Latent variable modeling in congruence research: Current problems and future directions. Organizational Research Methods, 12, 34-62.
Meanwhile, to test the indirect effect of the congruence model, refer to the compound coefficient approach which conveys more information than the block variable approach (Edwards, n.d.). The terms that represent the shape along the F = L or F= −L line for the indirect effect are tested by the confidence intervals (CIs) with the Monte Carlo simulation (20,000 replications, Preacher et al., 2010).
Edwards, J. R. (nd). Mediated Polynomial Regression. Available at: http://public.kenan–flagler.unc.edu/faculty/edwardsj/
MediatedPolynomialRegression.htm.
Then, you can test the moderated mediation effects by adopting/modifying the typical moderated mediation tests.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Regression