Skip to main content
4 events
when toggle format what by license comment
Jun 27 at 14:50 comment added Ben Ogorek One more comment. After banging my head against the wall for while, I realized that the procedure above does simulate the residual error term, at least not with predict(fitted_model, newdata = new_data, re.form = NULL). I added an rnorm with the residual sd from the mixed model estimate and now my coverage is correct.
Jun 26 at 15:49 comment added Ben Ogorek Really nice answer. It's annoying to me that the lme4's predict function needs newdata with random effects factors from the original data set. I'm going to simulate them, so I don't want to hard code them in! Well I swapped in two to make sure and type=parametric and use.u = FALSE is what I needed. Thank you! I should have been a Bayesian.
Jul 1, 2021 at 21:55 history bounty ended Björn
Jul 1, 2021 at 13:37 history answered EdM CC BY-SA 4.0