Inference in an analysis of variance makes a claim about the population means of the groups represented in the data. Before considering such inferences, we need to verify the conditions of the associated regression model.
The conditions for regression that are relevant in ANOVA regression are
✓ No obvious lurking variable
✓ Evidently independent
✓ Similar variances
✓ Nearly normal
The linear condition is missing from this list because this condition is automatic for an ANOVA regression. Because the slopes in an ANOVA regression are differences between averages, there is no need for fitting a curve. As for lurking variables, this condition is also automatic—if the data are from a ...