Table 1. Advantages of estimating grouping factors as fixed (LM) vs random (LMM) effects.

This table describes some advantages of fitting linear models (LMs) vs linear mixed-effects models (LMMs), when a grouping factor can be estimated as either a fixed effect or random effect, respectively.

LM	LMM
Faster to compute	Can provide increased precision
Simpler to use (less prone to user error)	Can incorporate hierarchical grouping of data
Conceptually simpler, such that metrics like R2 are easier to compute	Are (often) conceptually correct models of the system
Avoid concerns about singular fits	Can share information across groups (partial pooling), which aids in the estimation of groups with few observations
Preclude users from making inappropriate generalizations to unobserved levels of the grouping variable	Allows generalization to unobserved levels of the grouping variable
Avoid assumptions about the distribution of the random effects	In a study with multiple analyses with varying numbers of grouping variable levels, it would be more consistent to use LMMs throughout rather than switching to LMs at some threshold value
Estimating the number of degrees of freedom used is straightforward	Random effects use fewer degrees of freedom