From Russell Carleton at Baseball Prospectus on October 7, 2014, with mention of SABR member Jacob Pomrenke:
MLB formally banned (well… they started testing for) amphetamines in the 2007 season, so I took all regular season games that took place between 2003 and 2013. That should give us plenty of space on either side of the ban.
The great asymmetry in a day game after a night game (Jacob suggested DGANG as an abbreviation) is that the hitters in the game played the night before, while the starting pitchers were both likely explicitly given the night off, sent home early, and told to rest up. The manager does sometimes rotate in a bench player or two during the day game (more on that in a bit), but it’s likely that the starter is fresh while most of the lineup he’s facing has that not-so-fresh feeling.
Once again, I went to my trusty log-odds ratio method to control for our expectations of each batter/pitcher matchup. For those unfamiliar, it’s a way of controlling for the fact that we know that some matchups (Rob Deer vs. Roger Clemens) are more likely to result in a strikeout than others. I looked at the major outcomes for a plate appearance (strikeout, walk, HBP, single, double/triple. HR, and out in play). So, all of these findings control for the talent level of the batter and pitcher in each plate appearance.
How big is the DGANG effect? Well, in general, hitters in DGANGs show an increase in strikeouts and outs in play and a drop in singles, doubles, triples, although the shifts were in the range of a few tenths of a percentage point above (or below) expectations, enough to drive overall OBP down by 6-7 points. (All of these analyses were using binary logit regression at the plate appearance level.)
But did the ban on amphetamines make a difference? To get at that, I selected only plate appearances in DGANGs in which the batter was facing a starter (relievers may have thrown the night before, the starter almost certainly did not). I controlled for batter-pitcher matchup (min 250 PA in that season for both), the starter’s pitch count, whether there were runners on, and whether the pitcher had a platoon advantage. I also looked to see whether the hitter in the day game had been in the starting lineup last night. Maybe a fresh player has an advantage. I also coded whether the plate appearance took place in a year before the testing program went into effect (2003-2006) or after (2007-2013).
Read the full article here (subscription required): http://www.baseballprospectus.com/article.php?articleid=24810
Originally published: October 7, 2014. Last Updated: October 7, 2014.