Discussion questions#
Estimating random effects in a mixed effects model involves estimating the variance of the random distribution that generated the random effects (i.e., \(\eta \sim N(0, \Sigma_{\theta})\)). Why is this step necessary and how does it differentiate random effects from nuisance variables?
Yarkoni (2019) argued that mixed effects models should be standard practice for increasing generalizability of findings in psychology and neuroscience. How exactly does this lead towards better generalizability? Where could this potentially harm generalizability?