That is, until I read Baseball Between the Numbers. Edited by Jonah Keri and featuring contributions from a variety of authors from Baseball Prospectus, the goal of this book seems to have been to provide a near-comprehensive view of current baseball research on all facets of the game. They sought to cover everything from performance analysis and on-field strategy to long-term team development and the business side of baseball. They wanted to communicate these findings accurately, all the while keeping the writing approachable to the average intelligent fan. They are remarkably successful in accomplishing these goals. While certainly not perfect, this book is a wonderful resource to someone, like myself, who is trying to learn about all the great work that has been done over the past several decades to better understand baseball.
Here are a few snippits of things I learned while reading this book.
- Closers should be brought in earlier in the ballgame when the situation allows it, though using a closer in a standard closing situation isn't a terrible use of one's ace reliever.
- Earned Run Average is probably not as good a measure of pitching performance as runs allowed per nine innings.
- Teams leading late in the ballgame can be well-served by playing for one run.
- Bunting a runner to third with zero outs is often a better use of an out than bunting a runner to second with zero outs.
- Players do, in fact, perform better in their "walk" year prior to free agency.
- New stadiums are almost always a great deal for the team, but a bad deal for the city that pays for it.
- Some kinds of players lose effectiveness sooner than other players--it's predictable, to some degree.
- There actually is a measurable "clutch" skill that differs among players...though it's really small.
- The playoffs aren't a complete crapshoot--particular types of teams do tend to do better than others.
All that said, there are some notable critiques that one could levy toward this book. First, if one didn't know any better, one could easily walk away from this thinking that all the great baseball research has been done by either Bill James or Baseball Prospectus. It is true that there is some excellent research that goes on at that site, but this text ignores much of the work done outside of their group--and sometimes to their detriment. For example, all discussion of defensive statistics relies on Clay Davenport's defensive translations which, while certainly better than fielding percentage and range factor, fall short when compared to more detailed play-by-play metrics such as Zone Rating, Mitchel Lichtman's UZR, or David Pinto's PMR, all of which have been around and available for many years. There is only a brief mention of these alternatives in the chapter on fielding, and they are said to report results that are simply "very close" to that reported by Davenport's translations. In fact, differences do exist, and they can be substantial.
Those concerns speak to my other major issue with the book, and that is its tone. While not the case in all essays, a substantial number of the articles are written in a tone that is overwhelmingly authoritative--and unfortunately, far more so than the data justify. For example, after failing to detect significant year-to-year correlations between rates at which hitters ground into double plays, James Click writes that "Anyone who's seen many catchers lumber down to first knows that beating out a double play can't be entirely random. Yet the lack of any year-to-year consistency assures us that it is." What should be said is that the effect, if it exists, has not been large enough for us to detect, and is likely confounded by a variety of factors. In fact, there may still be differences, and you may indeed encounter players with unusually high or low double play rates--it's just that the effect is not as strong or common as one might expect it to be.
In this case, as well as other cases in this book, the conclusion seems to me to simply be too strong for what the data can show. There is always a certain degree of uncertainty in data analysis of any kind, and it is a mistake when researchers, be they amateurs or professionals, overstate the strength of their conclusions. In fact, it actually hurts one's arguments when one does this, because when a rare exception comes along that bucks the population trend, it makes one's arguments seem to lose all credibility. It is far better to take the cautious approach and stay within the bounds of one's data, which includes identifying and discussing potential shortfalls of one's analysis. This problem, of course, is not limited to baseball research, and is something that I also see from time to time among papers that I read as part of my professional work as a biologist.
Despite these criticisms, the scope, readability, and, with a few exceptions, accuracy, of this book cause me to highly recommend Between the Numbers. It serves as a great primer, and a great review, of much of the modern research in baseball, and yet it remains highly readable. Any fan interested in using statistics to understand the game of baseball would be well served by picking up this book and giving it a thorough read. Just be sure to read it with the same critical eye that these authors turn towards the conventional wisdom of baseball.