December 28, 2004
An Attendance Study
Eight years into the concept, MLB stands by its creation of interleague play, usually with a variation on this theme: "Fans love interleague play. In 2004, interleague play raised attendance by about 10%." Now, on one level this is accurate; interleague games had an average attendance of 32,914, while the intraleague games averaged just 30,112, so the interleague average is 9.3% higher. However, as Joe Sheehan has frequently pointed out, the data driving this claim is flawed and stacked in favor of interleague play. The primary problem is the issue of when games are played during the year.
There are two major divisions that can be made in terms of the schedule, both of which can have significant impacts on the number of people who will show up to a game. The first is the time of year, the summer months versus spring and fall months. Games played in June, July, and August are more likely to draw good crowds than games in April, May or September are. The weather is far more likely to be conducive to doing something outdoors during the summer. In addition, during the summer kids are out of school, and therefore families are more likely to come out to a game, especially a night game. The second division is weekday games versus weekend games, with weekend games obviously being more likely to draw a larger crowd. Since most Friday games are evening games and Sunday games are afternoon games, we'll call games on Fridays, Saturdays, and Sundays weekend games.
Putting these two factors together, you get that weekend games during the summer are likely to be the best for attendance while non-summer weekday games will be the worst. Let's look at the distribution of interleague and regular games using these distinctions for a typical team's schedule.
Interleague Regular Summer weekend 67% 21% Summer weekday 33% 25% Non-summer weekend 0% 25% Non-summer weekday 0% 29%The interleague games are heavily skewed towards the optimal calendar slots and as a result, the other games are skewed the opposite direction.
If we accept the idea that the straight comparison is a bad idea, the next question is how to get a better comparison. We can do that by using a "paired" comparison, i.e. we match up interleague games with other games based on the divisions mentioned above. For most teams this means using the first three weekday games after June 7, this year's start of interleague play, and the first six weekend games after that date. Some teams played fewer interleague games and in those cases the number of regular games used was reduced in order to match the interleague schedule. In addition, the last interleague game, an August 30 makeup game between the Phillies and the White Sox, was dropped because there was no good equivalent game to match it with.
So if we used that paired comparison, does interleague play still increase attendance? Yes, but by nowhere near as much as its boosters would like to proclaim. This year's interleague games averaged 33,023 fans, while the matching regular games averaged 32,453, an increase of 1.8% for the interleague games. Indeed this would seem to indicate that more than 80% of the supposed benefit of interleague play is merely a reflection of when the games are played.
An obvious possibility is that last year was just a fluke year or maybe the regular games which happened to match up with the interleague games featured unusually high attendance. Thanks to Retrosheet, it's possible to repeat the process for previous years. As with the 2004 data, the first Monday in June is taken as the starting point for the summer months and the occasional oddly placed makeup game was ignored. The results for the past four seasons:
Year Claimed gain Matched-pair gain 2001 13.0% 4.8% 2002 15.1% 6.0% 2003 11.5% 1.0% 2004 9.3% 1.8%Clearly last season was not a fluke, but a reflection of a very sharp difference between the claimed benefits of interleague play and the actual boost in attendance. Yes, there does seem to be some boost caused by interleague play, but it's nowhere near the double-digit percentage jump that its backers like to claim. 2002 was the year that the "rotating" interleague scheduling started, so each team played teams it hadn't faced before, which may account for the jump in that year.
There also seems to be considerable variation among teams in how much benefit interleague play provides. Some poorly drawing teams will see as much as a 50-60% jump in attendance for interleague play in some years, but others will see a 30-40% decline. Montreal managed to go from a 59.3% increase in 2001 to a 27.2% decrease in 2002! There are some theories floating around about which teams fare better in interleague play. In order to get something vaguely resembling a sane sample size we can divide the teams into pools and compare how each pool does.
One possibility is that the benefits of interleague play are driven by the rivalry games that are scheduled every year. Not every team has a natural rival, however, leading to some bizarre scheduling along the way. (The claims in 2001 that Seattle and San Diego were rivals sticks out as a particularly egregious example.) For the purposes of making a divide, we'll call the rivalry teams the ones who have an interleague opponent in the same state who they always wind up playing. This gives us eight pairs: Mets/Yankees, White Sox/Cubs, Dodgers/Angels, A's/Giants, Marlins/Devil Rays, Indians/Reds, Cardinals/Royals and Rangers/Astros. If we look at the results, we see that it pays to have a rival for interleague play.
Year Teams with rivals Teams without rivals 2001 8.3% 1.0% 2002 8.0% 3.9% 2003 1.7% no change 2004 2.0% 1.5%2001 was the year in which the Astros and Rangers started playing each other, which may account for the large difference between the two groups that year. In general, however, even the teams with rivals aren't reaching the double-digit percentage increase level.
Some might quibble with the inclusion of pairings like Indians/Reds or Marlins/Devil Rays, so we can narrow things down even further. The games between teams in the two-team markets--New York, Chicago, metro L.A. and the Bay Area--seem to be the games that draw the greatest amount of attention, so lets group those teams together and look again.
Year Two-team markets One-team markets 2001 10.9% 2.3% 2002 9.2% 4.5% 2003 3.2% -0.2% 2004 2.0% 1.6%So the teams sharing a market definitely seem to benefit more from interleague play than other teams do. Somehow, I doubt that an additional boost for large market teams was quite what MLB in mind when they started interleague play.
A variation on the idea of who benefits is the idea that there are certain hot teams that large numbers of fans want to see. To start with, let's consider the teams that host the defending league champions. Here we can add an extra detail to our test. In addition to separating out the teams that played the opposite league's champ in interleague play, we can also create a group of teams who played their own league's champion during their comparison games. Finally we have a third group consisting of teams who played neither or who played both, the latter included because the theory is that the benefits should more or less cancel out.
Year Opposite league Neither/both Same league 2001 16.7% 4.1% -6.3% 2002 8.5% 7.2% -4.8% 2003 11.1% 0.2% -1.6% 2004 12.5% -0.5% 0.3%The negative percentages mean that the teams in question actually drew better in the regular games compared to the interleague games. With the exception of last season, it's pretty clear that the quality of the opponent was a bigger drawing card than which league the opponent was from.
We can take this one step further and look solely at the effect of one team, namely the Yankees. Looking over the data, a pattern quickly emerges that whenever the Yankees come to town, attendance seems to jump unless the team was routinely selling out the ballpark already. Repeating our divisions from the last comparison we get some very striking results.
Year NL playing NYY No NYY AL playing NYY 2001 26.7% 5.5% -20.9% 2002 14.4% 6.3% -5.2% 2003 7.9% no change 2.7% 2004 10.9% 2.0% -9.0%Clearly the Yankees are indeed a big draw virtually anywhere they go, with Toronto seeming to be the exception in 2003. So while some AL teams might get a bump from interleague games, some teams are losing out because they are losing potential games against the Yankees, ones which seem to be as good or better a draw. Furthermore, it could be argued that the rivalry numbers are somewhat inflated because they include the Mets benefiting from playing the Yankees.
An overall pattern that seems to emerge from all of this is that hosting strong teams is always good for attendance, with the benefits being fairly consistent over time. On the other hand, the other comparisons show a general decline over time as interleague play becomes less of a novelty. Given that the seasonal differences in attendance have been known to baseball front offices for a long time, I suspect some variation on this analysis has been done by many of the teams. The inclusion of interleague play has a cost in making the scheduling process more difficult and making any weather-related postponements harder to make up, so there is a point where a small bump in attendance is not enough to cover the costs.
As interleague play continues to lose its novelty, we may start to hear rumblings that it's not worth the effort, and those rumblings may begin sooner than we expect.