A simulated binary outcome with two latent types of learners, for examples
of fit_glm() and subgroup analysis. Forty learners are observed on 80
days. For the 20 engaged learners the log-odds of submitting an assignment
rise with daily motivation (slope 1.2); for the 20 avoidant learners they
fall (slope -0.6). Each learner's slope varies around the slope of the type
(SD 0.15), and intercepts vary around -0.2 (SD 0.3). Daily motivation
fluctuates with SD 1 around a stable average drawn with SD 0.5. The data
are generated by data-raw/submissions.R in the package sources.