On the behaviour of marginal and conditional AIC in linear mixed models

In linear mixed models, model selection frequently includes the selection of random effects. Two versions of the Akaike information criterion, <sc>aic</sc>, have been used, based either on the marginal or on the conditional distribution. We show that the marginal <sc>aic</sc> is not an asymptotically unbiased estimator of the Akaike information, and favours smaller models without random effects. For the conditional <sc>aic</sc>, we show that ignoring estimation uncertainty in the random effects covariance matrix, as is common practice, induces a bias that can lead to the selection of any random effect not predicted to be exactly zero. We derive an analytic representation of a corrected version of the conditional <sc>aic</sc>, which avoids the high computational cost and imprecision of available numerical approximations. An implementation in an R package (R Development Core Team, 2010) is provided. All theoretical results are illustrated in simulation studies, and their impact in practice is investigated in an analysis of childhood malnutrition in Zambia. Copyright 2010, Oxford University Press.