I think Bob Hayden's recent posts about statistics textbooks are very helpful, and I want to add a brief note of my own. I used M&M a year ago last fall. (Since that time, I have done the corresponding course only once, and that time opted to use Moore's lower-level but more recent book, which Bob referred to as BPS.)
1.5 years later, two things about M&M stand out in my mind:
(1) The data sets were chosen thoughtfully and with great care. There's a statistical "lesson" associated with almost all of them. And, as a bonus, many of them lead to extra-statistical insights. Look for example:
at p.142/32, where the ability to predict urban pollution from a nearby rural location makes students wonder if monitoring stations are needed at both sites;
at p.140/28, where the ability to predict blood flow through the lining of the stomach makes the virtues of a non-invasive procedure palpable;
or at p.116/15, where one examines the relationship between pecking order in chickens and body weight. (I had my students do this one in class, in groups, and turn their work in. One group proudly summarized their observations this way: "Fat chickens don't cut it!")
This is sharp contrast to the problem/data sets I find in other books I use, or have used. For example, I'm doing an elementary biostatistics course right now. The book we're using finally, in its 6th edition, got around to supplying attributions and other contextual information to many of their problems. But it often does not help. Consider this one:
"Seventeen patients admitted to the Aberdeen Teaching Hospitals in Scotland between 1980 and mid-1988 were diagnosed as having pyogenic liver abscess. Nine of the patients died. In an article in the journal _Age and Aging_, Sridharan et al. (A-7) state that 'The high fatality of pyogenic liver abscess seems to be at least in part due to a lack of clinical suspicion.' The following are the ages of the subjects in the study: ..."
There follows a list of 17 ages, and the student was to compute mean, median, mode, range, variance, st dev, and coefficient of variation. The data were NOT separated into the ages of those who died and those who did not, which would have been natural enough. So what was the point of supplying context here, if students could do nothing with it? This is typical of the book (and others like it): the APPEARANCE of real data, contextual information, attribution--in a vacuous problem.
(2) M&M is loaded with sound, practical advice--much of it easy-to-use "rules of thumb"--that you can find almost nowhere else (in elementary textbooks, that is):
"... the F test and other procedures for standard deviations are extremely sensitive to nonnormal distributions. ... It is difficult ... to tell whether a significant F-value is evidence of unequal population spreads or simply evidence that the populations are not normal.
... we do not recommend use of inference about population standard deviations in basic statistical practice."
or see p.723 where, in connection with ANOVA, we find "If the ratio of the largest sample standard deviation to the smallest sample standard deviation is less than 2, we can use methods based on the assumption of equal standard deviations and our results will be approximately correct."
So, in my opinion, there you have it: outstanding data sets and problems; sound, easy-to-use practical advice available nowhere else-- except perhaps in Moore' BPS. (I used BPS this past fall, by the way, and liked it. But I didn't think it had the substance of M&M.) All I know is that the course I did 1.5 years ago with M&M was more fun than any course I can remember since leaving high school teaching nearly 30 years ago. And it wasn't because I suddenly became clever; it was mostly because I had an outstanding guide on whom I could rely. It didn't matter that the reading level was high (I could help interpret, when needed), or that the book was not visually trendy (my students never were conscious of that, as far as I could tell). I'm going back to it next fall.
============================================== Bruce King Department of Mathematics and Computer Science Western Connecticut State University 181 White Street Danbury, CT 06810 (firstname.lastname@example.org)