Hi I have a 2 level dataset (question nested within subj). My variables are: I am interested in a model is for the relations in level 1, i.e., I want to examine the within-subjects relationships of the variables: Reward precits correct both directly and indirectly via time Curiosity precits recall both directly and indirectly via time I used the following code (see below). The code was able to run, but I have few questions: 1. The output specifies correct as a continous variable, but it is actually a dichotomous variable. How do I change it, and does it matter? (reward is dichotomous too, BTW). 2. Other than that, does my code look good? 3. How do I calculate the indirect effect of reward (or OF curios) on correct via time? I know I can multiply the two relevant parameter estimates (e.g. time ON reward and correct On time), but how do I know whether this indirect connection is significant? Many thanks!!! Vered
VARIABLE: NAMES ARE subj exp sex age version question reward curios interest correct time; USEVARIABLES ARE subj reward curios correct time; CLUSTER IS subj; WITHIN = reward curios correct time;
MODEL: %WITHIN% correct ON reward curios time; time ON reward curios;
ANALYSIS: TYPE IS TWOLEVEL; ESTIMATOR IS MLM; ITERATIONS = 1000; CONVERGENCE = 0.000001;
See our Short Course Topic 11 video and handout as well as our book Regression And Mediation Analysis Using Mplus and its examples - all of this can be accessed via our website. There are special considerations for mediation with a binary outcome.