Zhang Yue posted on Saturday, November 15, 2008 - 8:51 pm
By now,there are three methods for estimate the mediantion effect in multilevel-mediation-model,where x,m and y are all lower-level variables.
I know that Mplus can do this ,which directly estimate the covariance of random effect aj and bj .But I want to know by which parameter estimation method Mplus uses to do so,and what else parameter estimation method I can use in multilevel-mediation-model in Mplus.
I also want to know how can Mplus directly estimate the covariance of random effect aj and bj ,would you send me more message for this ?
Two-level mediation modeling with random slopes is done by ML in Mplus. If you only have random intercepts you can also choose weighted least squares estimation. The covariance betweeen aj and bj is a level 2 parameter that is estimated by ML in line with regular hierarchical linear modeling.