Hi everyone,
I tested a traditional 2-1-1 MLM based on the Mplys syntax example of Preacher
(see https://www.statmodel.com/download/Preacher.pdf for more info).
I specified for standardized output in the syntax, which I received for the direct effects and path coefficients etc. The problem is that I did not get standardized estimates for the calculated indirect effects but only unstandardized coefficients. I was looking for the answer on why this is, but I can only seem to find papers who also report unstandardized estimates for indirect effects yet don't explain why.
So in short, why are standardized estimates for indirect effects of multilevel mediation models in Mplus not included in the output even though I requested it?
Cheers,
Maria
Maria Wiertsema In Mplus, the output for indirect effects in multilevel mediation models is typically provided in unstandardized form, even if you have requested standardized output. This is because indirect effects are typically computed as products of regression coefficients, and standardization is not a straightforward process for these coefficients.
The reason for not including standardized estimates for indirect effects in the output is that standardization can introduce complexities in multilevel models due to the hierarchical structure of the data. Standardization typically involves dividing coefficients by standard deviations, which may vary at different levels of the hierarchy.
In multilevel models, standardization at one level may not be meaningful or appropriate when considering the entire hierarchical structure. Therefore, Mplus, like other statistical software, usually provides unstandardized estimates for indirect effects to maintain the integrity of the multilevel model and to avoid potential misinterpretations that can arise from standardizing coefficients in a hierarchical context.
If you wish to obtain standardized indirect effects, you may need to manually compute them by dividing the unstandardized indirect effects by appropriate standard deviations of the variables involved. However, it's essential to exercise caution when standardizing coefficients in multilevel models and to consider the hierarchical nature of the data to ensure meaningful interpretations.
- Badges
- Science topic