/cdn.vox-cdn.com/uploads/chorus_image/image/37520842/154300792.0.jpg)
Those who follow the league closely look carefully at the average attendance figures published from the WNBA and other sources. (Note: Before any argument starts about the difference between the numbers below and what you see on television - peace. That's a topic for another essay. For the sake of argument, we'll take the numbers below as a given.) In particular interest is whether or not the averages for some given season are moving up or down.
In mathematics, however, there are other averages than the standard arithmetic average, which is the sum of the values divided by the number of values. The goal of an arithmetic average is to get a single number that represents a whole, but the average number is prone to be affected by very large or very small numbers. Games held on different days, at different times, for different reasons (kids' day for example) can skew the arithmetic average up and down.
There are other averages which might be better suited to measuring WNBA attendance. One is the median, the number which splits a set of numbers into a lower half and an upper half. Finding the median of a set of numbers might be more difficult, but since only 204 games are played in a WNBA season, it isn't too hard to do. Furthermore, cases like the Atlanta Dream moving to another arena for four games (which had low attendance) or cases like kids days (which have high values) don't affect the median very much.
So let's look at the median attendance numbers for each team in the 2013 and 2014 season, and the median attendance value for the WNBA as a whole in those two years.
2014 | 2013 | Difference | ||
Phoenix | 9155 | 8464 | 691 | |
Minnesota | 9104 | 9032 | 72 | |
New York | 8482 | 6621 | 1861 | |
Los Angeles | 8110 | 10090 | -1980 | |
Indiana | 7925 | 7934 | -9 | |
Washington | 7229 | 7092 | 137 | |
San Antonio | 7022 | 7486 | -464 | |
Seattle | 6179 | 6738 | -559 | |
Chicago | 6107 | 6047 | 60 | |
Connecticut | 5881 | 6285 | -404 | |
Tulsa | 5540 | 5452 | 88 | |
Atlanta | 5486 | 5359 | 127 | |
WNBA | 7348 | 7021 | 327 |
The numbers look good. The median attendance of a WNBA game has moved up by 327. The teams receiving the biggest jump in median attendance is New York, and that jump undoubtedly comes from moving back to Madison Square Garden. Los Angeles suffers the biggest loss, and that undoubtedly comes from the beginning of 2014 when it looked like the Sparks wouldn't find a buyer. Even though Atlanta has the worst median attendance of both 2013 and 2014, Atlanta's median attendance has gone up by 127, and Phoenix - in third place in 2013 - has now moved to the top most likely because of their amazing season.
Since I have data from 2013 and 2014 broken out by game, by date and by time, it might be interesting to look at "chronological" attendance medians. First up is the median attendance across the WNBA by day of the week.
2014 | 2013 | 2014 | 2013 | ||
Sunday | 7518.5 | 6828.0 | 46 | 47 | |
Monday | 6235.5 | 7381.0 | 2 | 3 | |
Tuesday | 7347.5 | 6860.0 | 36 | 28 | |
Wednesday | 7925.0 | 7516.5 | 19 | 20 | |
Thursday | 6931.5 | 7519.0 | 26 | 20 | |
Friday | 7137.0 | 6704.0 | 46 | 49 | |
Saturday | 7649.0 | 7870.0 | 29 | 37 | |
WNBA | 7381.0 | 7021.0 | 204 | 204 |
Note that the Monday attendance is misleading due to the small sample size. Most WNBA games are played on Tuesday and Friday. The early start to the season might have helped overall Sunday attendance this season, as the Dream didn't have to compete with pro football for the most part.
Another interesting graph to look at is the median attendance for games based on game start time. All game times listed below are the home team's local time.
2014 | 2013 | 2014 | 2013 | ||
11 am, 11:30 am | 10973.0 | 8911.0 | 9 | 9 | |
12 noon, 12:30 pm | 8770.0 | 10192.5 | 13 | 8 | |
1 pm, 2 pm, 2:30 pm | 7253.0 | 8023.5 | 8 | 6 | |
3 pm, 3:30 pm | 7023.0 | 6560.0 | 13 | 19 | |
4 pm, 5 pm, 5:30 pm | 7229.0 | 7004.0 | 11 | 17 | |
6 pm | 7071.0 | 7266.5 | 12 | 14 | |
7 pm | 7199.5 | 6889.0 | 112 | 105 | |
7:30 pm | 7158.0 | 6607.0 | 22 | 21 | |
8 pm | 7877.5 | 6490.0 | 4 | 5 | |
All times | 7381.0 | 7021.0 | 204 | 204 |
The overwhelming number of game starts - over half of all games played in the WNBA - take place at 7 pm local time. Note the high numbers for early morning games. These are skewed by the fact that these are probably kids' day games that take place during summer vacation.
Can any of this data be used for predictive purposes? Possibly. It might behoove a team with weak attendance to try to schedule as many games as possible on days and times when attendance is generally good. But in order to solve the problem of attendance in the WNBA - to give the team the strongest of fighting chances as possible - the problem will be a lot more complicated than just deciding when to hold the games.