Working on the random effects models now, and I just realized that we include school year as a linear predictor. Do we have good reasons to expect a one year increase in school year to be associated with a constant increase in the probability of turning out? Why would we? Should we switch this to be a set of fixed effects?
Working on the random effects models now, and I just realized that we include school year as a linear predictor. Do we have good reasons to expect a one year increase in school year to be associated with a constant increase in the probability of turning out? Why would we? Should we switch this to be a set of fixed effects?