Message/Author |
|
|
Hello, I’d like to plot a cross-level interaction following ex9.2b in the user’s guide. Deviating from ex9.2b my x-variable is dichotomous (coded ‘0’ & ‘1’). I’d like to plot the effect of w (between-variable, continuous) on s (random slope of x) for two different categories of x. How should I adjust the syntax, given that loop represents variation in x, but there are only two categories? Thanks for replying! |
|
|
s ON w is a between-level relationship. but x in ex9.2b is a within-level variable so it can't moderate this relationship. Is your binary x a between-level variable? |
|
|
Hello Is it possible to use example 9.2 to plot a crosslevel interaction when it is a multilevel logistic regression model? I am not sure how to interpret the effects because they are log odds. Thanks, Miranda |
|
|
For a plot example, see http://www.statmodel.com/download/Cross-level%20effect%20plotted%20by%20LOOP.pdf If you think it is simpler to understand, you can do this plot for odds instead of logodds. You just have to express the odds in Model Constraint. |
|
|
Thank you for the link. Do I understand it correctly that this input will plot the simple slopes on the y axis, that is, the effect of Y ON X, for different values of the moderator (x axis)? Is it possible to plot actual data points, based on the simple slopes. Thus, plotting the relation between x (on x axis) and outcome y (y axis) for different values of moderator (say, -1SD and +1SD shown as two separate lines)? If so, how would I set this up in Mplus language? |
|
|
Yes, using the LOOP and PLOT options in Model Constraints, the PLOT option can have more than one argument so that more than one thing is plotted in the graph for the same x-axis. See for example this script from our RMA book: http://www.statmodel.com/mplusbook/chap1/ex1.8_Part3.html |
|
Back to top |