The functions aov() and lm() can indeed be used to fit the same model and aov() has the advantage that when we summarise the fitted model object the output includes analysis of variance tables as the user might be expecting to see with the partitions based on the model terms.

Thanks for the suggesting about extending the blog post and I will look in to describing how to create graphs with means and confidence intervals and other usual ways to interrogate the model.

Ralph

My intention is to convert the blog posts into pdf documents for easier consumption online and offline. As you can see from supplementary material section of the website I have (slowly) started this process with producing graphs of various different types. If there are particular statistical methods that you feel should be covered then that is useful for me to know where to focus efforts in blogging.

Ralph

I personally use aov() and draw the diagnostic plot of that model.