Predicting probability of y <= threshold, after multilevel glm

[Amrit Ghimrie]

Hello everyone,
I ran a multilevel GLM using the megrim gamma family with a log link with random effects. After the model, I want to predict the probabilities that dependent_variable (y) would be less than or equal to a predefined threshold at different levels (0 and 1) of some binary explanatory variables holding other explanatory variables at the mean. I am struggling to get the probability. Could someone help me out with this?

[Maarten Buis]

What do you want to do with the random effects? Do you want to fix it at some value or do you want to treat that as an extra source of randomness?

[Amrit Ghimrie]

The predicted probabilities are, by default, averaged across the random intercepts' distribution if not mentioned explicitly. Isn't this similar to predicting the outcome value by defining some value to one predictor and leaving others at mean? Please enlighten me if I am wrong. And if random effects are also to be taken into account, could you please elaborate as I am just learning.