I have taught using this text 5 times now. I like the concept very much. My goals are to teach the students how to choose appropriate statistical tests and how to write up their results professionally. I very much like the concept of "project" exercises with real data sets.
Unfortunately, many of the datasets contain faulty data. I doubt that I have a full list of example, but Chapter 2, Problem 23 supposedly contains percent change in fatalities in states that retained and did not retain 55 mile per hour speed limits between 1995 and 1996. Unfortunately, the data reported have nothing in common with the real data. For instance, there were 87 fatalities in Alaska in 1995 and 81 in 1996 leading to a 7% decrease. The number used in the text is a 29% decrease. Other examples include Chapter 8, Problem 20 which is admittedly an approximation but the authors did not read the scale on the New York Times graphic correctly which places the supposed outlier rather differently than if the correct scale for the data is used. In Chapter 11, problem 24, there is an excessively heavy grasshopper mouse among other errant species weights.
The concept is great, a few data entry errors can be an effective learning tool, but too many blatantly incorrect datasets that lead to conclusions diametrically opposed to the real data seems sloppy to me.