|
|
 |
| Combination of linear and logistic paths |
 
|
| Message/Author |
|
|
|
|
| We are estimating a mediation model with a latent independent variable, a latent mediator and a dichotomous dependent variable (latent – latent – dichotomous). We are wondering, if it is sensible to estimate the indirect effect, i.e. the combination of linear and logistic links. Mplus allows us to use MODEL INDIRECT as well as MODEL CONSTRAINT – however, we are puzzled by the interpretation of the coefficient and the standard error. Many thanks! |
|
|
|
|
Yes, a sensible estimation of the indirect effect can be done. See the paper on our website's Mediation page (left margin's special topics) which shows that you should get counterfactually-defined effects which are suitable for a binary outcome:
Muthén, B. & Asparouhov T. (2015). Causal effects in mediation modeling: An introduction with applications to latent variables. Structural Equation Modeling: A Multidisciplinary Journal, 22(1), 12-23. DOI:10.1080/10705511.2014.935843 Click here to download the paper.
See the listing of Mplus scripts on page 22 of this article or see the UG on Model Indirect and MOD on pages 759-766. See also our Topic 11 Short Course video and handout on our website describing indirect and direct effects with a binary outcome. |
|
|
| Back to top |
|
|
