Dear community,
Unfortunately, we don't work with SAS very often at the university. Therefore, I have been attempting to code a multilevel logistic regression using proc glimmix over the past few days. I successfully defined the model, but I am now facing challenges in visualizing the plot.
The model is as follows:
/*Random Intercept, fixed slope model*/
proc glimmix data=muslim method=LAPLACE empirical;
class country_1 sex (ref="0") believe_god (ref="0");
model anti_muslim(ref="0") = age sex edu rlscale2 believe_god attend_rel
/link = logit distribution=binary solution alpha=0.05;
random intercept / subject = country_1 type = un;
output out = predict pred = logitpred;
run;
I am grouping the data by countries (N=34), and my outcome variable is dichotomous (responding to the question: "Would you like to have a Muslim as your neighbor?" where 0=no). Additionally, I have a few Level 1 predictors (age, sex, etc.). My goal is to create a spaghetti plot with the predicted outcome on the y-axis and the predictor on the x-axis. I believe that by doing this, I should obtain a graph with 34 lines representing each country's development over the predictor.
My attempted approach is as follows:
proc sort data = muslim;
by age sex edu rlscale2 believe_god attend_rel;
run;
data predict;
set predict;
pred = logistic(logitpred);
run;
title "Plot";
proc sgplot;
series y = pred x = age / group = country_1;
run;
Unfortunately, the resulting visualized plot is quite confusing. My suspicion is that SAS created a line for every individual (Level 1 unit).
I am unsure of what might be causing this issue and would greatly appreciate any assistance.
One more piece of information: There are missing values in the predictors. As a result, when I save the predicted values as shown above, there are numerous missing values in that variable.
Thank you in advance for any help you can provide and if you need any information please ask!
