include mean(bernoulli) for getting the proportions for visualisation
change of outcome for one unit increase of predictor—means different things for different points of incdreasing, because of the non-linear transformation (logit()). so figure it out at the mean 0, + 1. then report it that way too.
mean(bernoulli)for getting the proportions for visualisation