All,
I need some help generating a graph to look the way I want using estimates based on a mixed effects logistic regression model. I am estimating the probability of drug use over time, treating time as a categorical predictor (because I want to examine the effects of time relative to one specific time point). Observations are nested within participant and the model includes a random intercept term for each participant. Post-estimation, I want to show the marginal effects over time by serostatus (3 categories) and I want this plot repeated 3 times, once per recruitment cohort. So graph one represents the effects of time by serostatus for cohort 1, the second represents time by serostatus for cohort two, etc.
What I want looks like this graph: https://www.statcon.de/shop/images/stata/stata_05.gif
where each individual plot would be side by side with one title and one legend and the y-axis labeled only once. The best I have gotten so far is 3 graphs on the same plot but with two at the top and one at the bottom and to the left, leaving a gap on the lower right side. Can anyone help me with the margins and margins plot commands to get a graph that looks like this example graph? If there is a way to use predict to get the correct values and then combine the graphs using twoway, that would work also (I think).
Thanks in advance for any help.
Regards,
James
I need some help generating a graph to look the way I want using estimates based on a mixed effects logistic regression model. I am estimating the probability of drug use over time, treating time as a categorical predictor (because I want to examine the effects of time relative to one specific time point). Observations are nested within participant and the model includes a random intercept term for each participant. Post-estimation, I want to show the marginal effects over time by serostatus (3 categories) and I want this plot repeated 3 times, once per recruitment cohort. So graph one represents the effects of time by serostatus for cohort 1, the second represents time by serostatus for cohort two, etc.
What I want looks like this graph: https://www.statcon.de/shop/images/stata/stata_05.gif
where each individual plot would be side by side with one title and one legend and the y-axis labeled only once. The best I have gotten so far is 3 graphs on the same plot but with two at the top and one at the bottom and to the left, leaving a gap on the lower right side. Can anyone help me with the margins and margins plot commands to get a graph that looks like this example graph? If there is a way to use predict to get the correct values and then combine the graphs using twoway, that would work also (I think).
Thanks in advance for any help.
Regards,
James
