Message/Author |
|
|
Hello, I am performing SEM analyses in MPlus and was wondering if there's a syntax to get effect sizes for various associations? Thank you |
|
|
You can use Model Constraint to express any effect size definition you want in terms of model parameters. |
|
|
Hi Linda and Bengt, I am trying to compute an effect size via MODEL CONSTRAINT in a multigroup model. Question 1: Is there a way of labeling the number of observations in each group, so I can use those numbers directly in MODEL CONSTRAINT? Question 2: When adding the constraints to compute the effect size, the results contain a lot of "NaN"s. What can I do? Thanks! |
|
|
Q1. No, but you can enter the numbers directly. Q2. We have to see the output. Send to Support along with license number. Be sure you used 7.4. |
|
Mplus User posted on Wednesday, May 31, 2017 - 11:05 am
|
|
|
I am hoping to report effect sizes for my mediation analyses. In a personal conversation with Kris Preacher, he tentatively recommended reporting the completely standardized effect for a mediation analysis. See p. 99 of the following article for the mathematical formula: http://quantpsy.org/pubs/preacher_kelley_2011.pdf I'm wondering if Mplus produces the completely standardized mediation effect. If I run a mediation analysis using bias-corrected bootstrapping and stdyx, is the standardized coefficient for the mediation effect with confidence intervals equivalent to the completely standardized effect? |
|
|
Yes, Mplus uses eqn (9). |
|
Back to top |