I was absent-mindedly keeping track of the Seattle and Boston doubleheader today when I decided that it felt to me that the game was progressing quickly. The kernel of a thought gave rise to me pondering other doubleheaders that I had watched and I had a fuzzy recollection of those also moving quickly as well. Thus, a hypothesis was born. Knowing that they have to play two games in one day, do players attempt to play at a quicker pace?
Far be it for me simply to toss out a hypothesis without taking steps to try to prove it. After all, I wouldn’t want to be like Damien Cox. Lucky for me —and for all of us— we live in a reality where Retrosheet exists and is free and totally awesome.
Testing the theory is a bit more complicated than simply comparing the average game duration for doubleheaders against single game days. There’s historical context that needs to be applied. Games take longer to complete now than they used to and there used to be more doubleheaders. Therefore taking the average doubleheader duration would result in a skew toward earlier seasons, which had smaller durations.
To get around that, I broke up each season into a separate entry for normal games and doubleheader games and compared them down the line. Over the years 1952 through 2009, the average doubleheader game took 99.8% as long to finish as a normal game. Technically, they do appear to take a shorter amount of time to compete but the margin is so incredibly small and the sample size on doubleheader games small enough that I feel this is not a statistically large enough difference to prove the hypothesis.
It might be tempting to think that nothing was learned, but that’s not true. Proving a theory wrong —or at least showing it currently impossible to prove correct— is still added information. Besides, I made this cool line chart!
I find it interesting that game times were fairly stable from 1952 (147 minutes) through 1976 (149 minutes) and from 1991 (174 minutes) to present day (175 minutes). That was one steep and steady climb.