For modeling vaccine efficacy across all strata with random effects, despite only have flu-shot data for some subjects, we need to make the model understand flu_shot == NaN as flu_shot=="dontknow". The in modeling, we make no the reference state, so the log-odds of efficacy is given by the factor flu_shot == yes and we can provide an averaged prediction for don't knows.
For modeling vaccine efficacy across all strata with random effects, despite only have flu-shot data for some subjects, we need to make the model understand
flu_shot == NaN
asflu_shot=="dontknow"
. The in modeling, we makeno
the reference state, so the log-odds of efficacy is given by the factorflu_shot == yes
and we can provide an averaged prediction for don't knows.