Data analysis can be a tricky business fraught with pitfalls to trap the unwary. In teaching data analysis I have found that well-constructed data sets can illustrate some of these pitfalls effectively. My general strategy is to present students with the data in a computer lab setting and ask them to do an analysis then and there. I do not tell them the data present a pitfall. Whether anyone successfully avoids the pitfall or not I immediately follow up with a discussion of the data and the pitfall.
In this note I present three data sets I have used successfully in acquainting both undergraduate and graduate students with some of these pitfalls. The first is due to Anscombe [1] and is widely known as Anscombe’s quartet. The other two are variations I have created to illustrate other pitfalls.
