Maximal model

The “maximal model” is the version of the model that contains all possible random intercepts and random slopes that the data can support.

A maximal model does not need to contain every possible fixed effect. The term “maximal model” refers only to the random effect structure, not the fixed effects. (As always, fixed effects should be chosen based on your RQ.)

For how to identify the maximal model for your data, see Identify possible random effects.

If you try to fit the maximal model to your data, you might find that it fails to converge (see Troubleshoot common issues). If this is the case, you may need to change the model’s optimiser and/or simplify the random effect structure to a non-maximal version that does converge.

Linked flash cards