[R-sig-ME] Reconciling Near Identical AIC Values and Highly

[Elizabeth Crone]

Be careful of the p-values generated by summary() or single-model anova() for a lm/GLM/lmer model.  They are rarely for the hypotheses I want to test.  The most idiot-proof way to get p-values is with Anova() for one model, or anova() for two.

I hope this resolves your discrepancies.

Elizabeth Crone

Sent from my iPhone

On Apr 14, 2014, at 6:00, <r-sig-mixed-models-request at r-project.org> wrote:

> Reconciling Near Identical AIC Values and Highly