5
$\begingroup$

What is the consequence of using an ordinary linear regression model while a linear mixed effect model is appropriate? Will the estimates and inferences be misleading?

$\endgroup$
8
$\begingroup$

We often use mixed models to account for non-independence of observations, which is one of the usual assumptions of linear regression. This often occurs due to clustering of observations, such as repeated measures within subjects. When observations within a subject are more similar to each other than to observations in other subjects, we will have non-independence so we fit random intercepts for subjects to acount for this. If we do not fit random intercepts, or account for this non-independence in other ways (eg with fixed effects for subjects, or with generalized estimating equations) then point estimates will still be unbiased and consistent, however standard errors will be wrong.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.