Thomas Lumley • 1/9/2024

Asymptotics for linear mixed models

This technical article analyzes the asymptotic properties of parameter estimates in Normal linear mixed models. It examines the loglikelihood as a quadratic form in Gaussian variables, discussing its distribution as a linear combination of non-central chi-squared variables and the conditions needed for asymptotic normality of the estimates, including the Lindeberg condition and implications of model misspecification.

0 comments

#statistics #Asymptotic Theory #Maximum Likelihood