The complexity of linear mixed-effects (LME) models means that traditionaldiagnostics are rendered less effective. This is due to a breakdown ofasymptotic results, boundary issues, and visible patterns in residual plotsthat are introduced by the model fitting process. Some of these issues are wellknown and adjustments have been proposed. Working with LME models typicallyrequires that the analyst keeps track of all the special circumstances that mayarise. In this paper we illustrate a simpler but generally applicable approachto diagnosing LME models. We explain how to use new visual inference methodsfor these purposes. The approach provides a unified framework for diagnosingLME fits and for model selection. We illustrate the use of this approach onseveral commonly available data sets. A large-scale Amazon Turk study was usedto validate the methods. R code is provided for the analyses.
展开▼
