Students in our sabermetrics course often ask, “Where can I find a good hypothesis?” The simple answer to that question is, “They’re everywhere you look!”
Sometimes a hypothesis falls out of a good observation. It may be something you notice while watching a game at the park or on TV. For example, you might be surprised to see that Tony LaRussa has penciled his pitcher in the number eight spot of the batting order. Tony obviously thinks that this lineup will yield more runs than when the pitcher is hitting last. So there’s your hypothesis – does bumping a pitcher up in the lineup result in a different number of runs scored? Of course there are many approaches to answering that question, and there are many variations of this hypothesis we could explore. But it shows a simple example of a nice hypothesis.
Many good research hypotheses come from observations in the popular press. Here’s an example: last year the Boston Globe noted, sans context, that Jason Varitek was batting .375 (36 hits in 96 at-bats) after a day off. That raises some interesting questions, one of which might be, do players have a higher batting average in games after they take a day off? Or perhaps, do catchers have a higher batting average in games following off days? Now, you might think the answers to those questions are uninteresting or unimportant, but that doesn’t matter for this exercise. The important point is that in and of themselves they are good hypotheses.
Finding a good hypothesis by perusing the baseball pages of your favorite daily can be a little more difficult, because you sometimes have to unwrap a few layers of spin or misinformation. I find it useful to always stop and think critically when I see a number in print, or when I read a statement that might be untrue or misleading. For example, we can revisit the recent column in the Chicago Tribune, in which we determined that the writer’s ostensible hypothesis was that Mark Buehrle’s league-leading innings total from last year led to a decline in his pitching stats. So here’s the good hypothesis: is the number of innings thrown by a starting pitcher related to a change in his performance the next season?
For some good research ideas, the only explanation for their emergence is “Eureka!” You can’t predict when such an idea will hit you. . .but perhaps you can increase the chance of having such a moment by devouring all the baseball media you can, and thinking critically about the information you’re being presented.
Finally, it’s worth noting that you could state each the above hypothesis as a statement instead of a question. We’ll do that next, when we look at moving from hypotheses to models to graphics (and sometimes back again).