Subject: Re: [isostat] stat in econ/public policy books, etc. |

From: "Steve C. Wang" |

Date: Thu, 27 Jul 2006 14:14:39 -0400 |

To: isostat@oberlin.edu |

Last week I asked for suggestions for general-interest (i.e., non-textbook) reading on stats in econ or public policy for a student of mine. It was suggested to me that I summarize the responses I received, so here they are (with names removed). My original request:

I have a student who is interested in applications of statistics in economics and public policy, and he's looking for a book to read over the summer. He's looking for something that can be read cover-to-cover, as opposed to a textbook or reference book. He's also looking for something that would convey an idea of how stats is used in those fields, rather than teaching how to carry out the actual methodology -- sort of like a Stephen Jay Gould book, but for stats. My first thoughts were "Statistics: A Guide to the Unknown", and "Freakonomics" by Steven Levitt. Any other ideas?

The replies:

There are some older books - one by Federer comes to mind - it may be hard to find. Dekker published it. Another option might be on the history of statistics. Hald has a wonderful two volume set (Wiley), and there is the more recent history by Steve Stigler. F.N. David had a book on Games, Gods and Gambling (Hafner?). These aren't specifically econ/public policy. Another issue might be that a "read-through" is not likely to take. Statistics isn't a spectator sport, so some questions/projects etc. in addition could be helpful. For example, having the student dig out some data in his/her field and write a report on it. Maybe do some graphs You didn't mention if the student has had a course in statistics yet. If not, this would be a natural step for the fall. I don't like the idea of "learning" statistics as a reading course, although a lot of distance learning is working that way. The difference is that homework assignments and exams are usually required.

How about "Against the Gods: The Remarkable Story of Risk" by Bernstein. It's a very entertaining book on the history of probability, statistics and their application in economics.

I think the Tanur book is a good choice. It's dated, but the case studies are still relevant and interesting. I was talking to one of my econ students about freakonomics, and we both described cringing at the conclusions made at various points. It is a fun read, and explores a lot of statistical problems you wouldn't normally think of an economist tackling (did I tell you Steve Levitt was in my graduating class of 80

>from St. Paul Academy?). Concepts and controversies is also good (as

you know) but is a little more of a text book and maybe too elementary.

Freakonomics is great, and I think the actual economic research is very accessible. I'd recommend that he read the book and the actual papers that support the claims.

I know a book that's very good for economics majors. But it's a textbook. The book is: Applied regression analysis, by Terry Dielman, Thomson, Brooks/Cole. Again, this is NOT exactly what you look, as I am using it in my regression course, a good text book.

The first thing coming to mind would be to do a search using "econometrics" and see where such leads. With respect to Steve's comment regarding cover-to-cover reading of a textbook, I believe that if we are dealing with a well-written book (such as any of the Fab Five), then it would be a good idea for the first-time teacher to read such cover-to-cover during the summer before first teaching the course. Relatedly, if I had to pick one textbook to recommend to a teacher who will eventually be teaching APStat that is in line with Steve's (unrelated) request and best reads like a nontextbook, then I would recommend David S. Moore and William I. Notz's Statistics: Concepts and Controversies (Sixth Edition), where the three-word subtitle expresses one big reason why well. In a nutshell, ignoring the exercise sets (which may be difficult to do because, although they're fairly straightforward, they're somewhat intriguing), the book reads like what it is, namely a 24-chapter book of commentary. PS: For those who are unaware, is was Dr. Moore's SCC that was the foundation for the much of the way things are presented in APStat as well as in the typical intro college class today.

Yes, Freakonomics comes highly recommended. Levitt is a world class economist and a really good econometrician. My son read it after he graduated from high school and really liked it. Lots of good inferential statistics.

Against the Gods, the Remarkable Story of Risk by Peter Bernstein is excellent and easy to read.

Steve: Your two choices: "Stats, Guide to Unknown" and "Freakonomics", are both good, for different reasons. Freako is very amorphous but inspiring (I almost objected to it.) You might also consider the special issue of Statistical Science, described below (I wrote it up for College Math. Journal column, "Media Highlights".) If anything else comes to mind, I'll relay it, though I suspect you'll get plenty of suggestions from isostat. Special Section: How to Lie with Statistics Turns Fifty. Statistical Science 20:3 (August 2005) The fiftieth birthday of Darrell Huff's famous "How to Lie with Statistics" provided the occasion for these sequels. The title of Huff’s book is both a blessing and a curse. To its credit, such a come-on attracted a huge readership, helping to make this the best selling statistics book of all time. Unfortunately, it suggests a wrong-headed view of statistics and unfairly imputes malice as cause for the numerous errors that do infect the use of data. The present collection of articles does an excellent job of showing the manifold ways things can go wrong or right in the collection, presentation, use, and implications of data. [...]

Steve: Against the Gods is indeed a good choice. Also worth a look in that general vicinity are the following two, picked almost at random off my shelf of such: Brian S. Everitt: "Chance Rules" 1999 Springer (Copernicus imprint) H> W> Lewis: "Why Flip a Coin" 1997 Wiley For this latter see Chance News 6.11, part 2 of 3 parts for a review.

Another, that one of my business school students really liked, is "Fooled by Randomness: The Hidden Role of Chance in Life and in the Markets" by Nassim Nicholas Taleb. About probability and the financial markets.

Freakonomics has at least one really bad analysis in it, but maybe that's ok. There are a couple of other options... Statistics and Public Policy by Fairley and Mosteller (a collection of articles, like SAGTU) Summing Up by Pillemer and Light (a reviewing literature/meta-analysis focus)

Against the Gods: The Remarkable Story of Risk.

Since you mention Stephen Jay Gould, you might suggest his "Full House", especially Parts 2 & 3 as a wonderfully clear discussion of variation in many fields. It's a big old, but Robert Behn & Vaupel's "Quick Analysis for Busy Decision Makers" is a very readable treatment of ideas in decision making under risk and uncertainty in a public policy setting. Freakonomics is an excellent choice as you note.

I'm not sure if this is relevant, but Edward Tufte has a book titled "data Analysis for Politics and Policy".

Data Analysis for Politics and Policy, Edward Tufte, Prentice Hall, 1974. Very short, very good, very expensive. A reader who has not taken a statistics course will have to skip some technical sections but this deals with meaty applications and meaty statistics. Recommended for every statistics teacher.

-- -------------------------------------------------------- Steve C. Wang Assistant Professor of Statistics Swarthmore College http://www.swarthmore.edu/NatSci/swang1