2-way anova on unbalanced dataset

Is aov appropriate f开发者_开发知识库or unbalanced datasets. According to help ...provides a wrapper to lm for fitting linear models to balanced or unbalanced experimental designs. But later on it says aov is designed for balanced designs, and the results can be hard to interpret without balance.

How should I perform a 2-way anova on an unbalanced dataset in R?

I would like to reproduce the different results for type I and type III sum of squares of SAS output (when using proc glm). I remember we were using type III sum of squares for unbalanced datasets.

Thank you in advance.

Function anova (or summary.aov) will give you the so called type I (or sequential) sum of squares. To get type III sum of squares, you can use the Anova function from library car with parameter type="III". The difference between these two approaches in unbalanced datasets (and also sample R code to produce both tables) is presented in detail here.