RebKell's Junkie Boards

[Phil]

Anyone up for some analysis? Not much, I'm not looking to do an in depth study; I was going to write something about marketing, and I wanted to tease it with the "fact" that stronger teams have much better attendance.

Which turns out to be true, but much less strongly than I would have guessed.

I plotted last year's attendance against the teams Sagarin rating. As expected, the higher the Sagarin rating, the higher the attendance, but the correlation isn't all that strong. A linear fit has a correlation of about 24%. That's pretty weak tea, and not improved by using a log, exponential or power fit.

(I also did a comparison of Sagarin rank versus attendance, which is best fit by a power curve, but that doesn't look like a better metric.)

A couple things surprised me. Rutgers isn't among the top 50 in attendance, so one of the reasons I am posting this is to make sure I am not missing anything.

I can already see I ought to use a multi-year measure of strength, which I may do.

Would love some reactions or suggestions.

You can see the data in Google docs at:
https://docs.google.com/spreadsheets/d/1qTS3W8TMx56ytPqEaqsfHz3_aKYDm9L1O0oW8HRhXrw/edit#gid=996592021

Google docs doesn't include trend lines; you can see the original trend lines at:
https://docs.google.com/file/d/0B3pjvR_kW6qvcjg1MUktN3ZzOFE/edit

[beknighted]

[Fighting Artichoke]

I think you need to look at more years. Since it wouldn't require much more work, it would be interesting to compare using each team*year as a data point as well as the more valid mean attendance vs. mean Sagarin rank with each school showing up as a single data point. For the year*team analysis, each year a team would count as a single data point, so there would be psuedoreplication, of course.

One more factor to consider is the venue size (available at WBBState.com). For example, some schools have small venues, so they cannot get large attendance numbers. You could try percent capacity vs. raw attendance.

Looking at your data, it seems to me that Tennessee fans are the best, with Louisville and Iowa State also amazing, as these fans show up disproportionately relative to their teams' success. Duke's numbers are very low compared other successful programs, as their stadium was barely above half their capacity. Could it be JPM fatigue? Has their attendance plummeted recently, or was it always that low?

[Phil]

Yes, I've been mulling the need for more years, just too lazy to pull together the data.

I'm thinking I need more years of team strength, more so than multiple years of attendance, but I guess I should look at both.

I'm not buying the capacity argument. If anyone is selling out, it would be worth looking at but I don't think arena capacity is often a limiting factor. Louisville sold out I think, but I doubt there are many examples.


In terms of who did best, I'd agree it is Tennessee, if we look at the raw count above expected. However, if we look at the ratio of actual attendance to expected attendance, Tennessee still does well, but New Mexico, Iowa State and Louisville are slightly better.

I'll post the numbers shortly.

[Fighting Artichoke]

[Phil]

Rank Actual Predicted Excess Ratio Team
1 5,914 2,664 3,250 2.22 New Mexico
2 9,916 4,696 5,220 2.11 Iowa St.
3 9,469 4,869 4,600 1.94 Louisville
4 11,038 5,703 5,335 1.94 Tennessee
5 7,388 4,792 2,596 1.54 Purdue
6 7,145 4,953 2,192 1.44 Michigan St.
7 3,885 2,825 1,060 1.38 Texas Tech
8 3,932 2,941 991 1.34 Toledo
9 5,430 4,115 1,315 1.32 Ohio St.
10 8,694 6,842 1,852 1.27 Notre Dame
11 6,161 4,990 1,171 1.23 Nebraska
12 2,977 2,443 534 1.22 Missouri St.
13 5,632 4,786 846 1.18 Oklahoma
14 4,196 3,633 563 1.15 Wisconsin
15 6,371 5,520 851 1.15 South Carolina
16 6,881 5,989 892 1.15 Baylor
17 6,727 5,910 817 1.14 Louisville
18 8,313 7,525 788 1.10 Connecticut
19 5,417 4,929 488 1.10 Penn St.
20 4,826 4,539 287 1.06 Middle Tennessee

[Phil]

In the table above
Actual - the actual average attendance last season
Predicted - the expected attendance if the attendance followed the best fit linear model of attendance as a function of strength.
Excess - the amount by which the actual exceeds the predicted (subtration)
Ratio - the ratio of the actual to predicted
Team - Team name

[Phil]

[Fighting Artichoke]

They removed the bench seats and replaced them with individual seats with backs, I believe, hence the reduced capacity. And the ticket prices, at least for the women, did not increase.

[Stephen Shirley]

Anyone have the ticket prices for those programs?

One of the things that I have noticed is a lot of Nashville area teams offering WBB SEASON tickets for as little as $25.

Meanwhile the cheapest MT season ticket will still set you back at least $110.

[Fighting Artichoke]

By the way, I checked the regular season games for last season and there were only 3 sellouts. Turns out the attendance was pretty steady for the season.

[Phil]

[ClayK]

Great work on this ... wonderful stuff.

Another issue, of course, is reported attendance versus actual attendance.

Just like every other sport, women's basketball routinely inflates attendance, either by counting season ticket holders who don't show, or by adding phantom numbers to the reported total.

That's not bad or good, but it relates here because some programs may

a) report accurately
b) fudge a little
c) fudge a lot
d) have luxury boxes and count those as attendance
e) manipulate the numbers in some other way

What might be a better measure is revenue, though that too is subject to the whims of the athletic department administration.

_________________
Oṃ Tāre Tuttāre Ture Svāhā

[beknighted]

[Phil]

I thought about mentioning reported versus actual, but decided not to. Several years ago, I was looking at attendance, and noticed that Tennessee was one of the very few teams to report both figures. I applaud them for the honesty, but the data did show that reported exceeded actual by roughly 3,000 per game, which is meaningful. Their reported attendance far exceeds, for examples. Notre Dames, but unless Notre Dame has a material number of no-shows, Notre Dame actually might have more cheeks in the seats. The Tennessee attendance is very impressive, even after adjustment, but their reported numbers don't tell the whole story. One challenge is that most teams do not report (at least in a way I could find) their turnstile numbers, so it is hard to tell whether Tennessee is an outlier or not.

As benighted beknighted notes, that won't change the trend lines if the practice is consistent over time, but it does distort the measure of attendance as a function of strength.

I also agree with beknighted's point about a lag. I am thinking about looking at attendance as a lagged variable. Anecdotally, I recall a lot of stories of teams who achieve some measure of success, (first NCAA, first Sweet 16, first FF, first NC) and reporting an increase in season tickets the next year.

[Dennis1361]

I have yet to see an attendance comparison where there are alternatives. For example Corvalis Oregon or Eugene are hardly comparable to sa the SF Bay Area

[GlennMacGrady]

[Phil]

I was thinking though other variables affecting attendance, and thinking about UCLA to illustrate two opposite points:

1. The availability of other options to spend an evening
2. The number of people living close enough to the arena

In many cases, those will roughly cancel.

[ClayK]

[Phil]

I looked at some older years, 2008,2009, 20010, 2011. I didn't have Sagarin ratings for those years, so I used the RPI ranking. RPI is an imperfect measure of strength, but it is decent as a simplifying case. I have some spreadsheets, but they are not very cleaned up, so I do not plan to post them.

The results for each year were close to each other, which is comforting, suggesting the results aren't random results.

As before, the correlation wasn't all that high - about 25%, surpassingly consistent over time.

I tried modeling the attendance as a function of the prior year's strength. It worked OK, but didn't improve the fit. My working hypothesis is that strength in one year may drive the seasons tickets for the subsequent year, but season's tickets are only a portion of total sales. Non-season ticket sales are probably more a function of current year's strength than prior years. I don't have good access to season ticket sales for schools, so not easy to test, but the limited analysis I did suggests it isn't a major factor.

I tried to do a multi year analysis; looking at an average rating over several years, to see how that predicts attendance. It didn't work well, many due to data challenges (the list of teams keeps changing each year). It still may be worth doing, but I'm abandoning ti at the moment.

My main goal was not a model of attendance itself, but a measure of the effectiveness of marketing efforts by schools. I wanted to remove the effect of team strength, as the marketing department doesn't have much control over than.

There are other factors not under the control of the marketing department, such as the population within driving distance, so I am happy that people have been identifying some of those factors.

[ClayK]

This is great stuff. You might want to look at the Ackerman Report, which Val Ackerman did for the NCAA (http://www.ncaa.org/sites/default/files/NCAAWBBWHITEPAPER.pdf).

I think there are many factors that contribute to attendance, and one of them might be the strength of high school girls' basketball in a particular area. Some states and regions have more of an interest in the high school game than others, and at least in the Bay Area, I think that interest translates to some degree to the colleges. (That said, the high school players can't go that often because of their own practice and game schedules.)

As for marketing, that can only get fans to try the product; after that, they have to like it enough to come back. So if a team has been solidly marketed for a few years, then I would suspect most possible customers have been exposed to the school's product.

At that point, such factors as competitiveness, cost, convenience, in-arena experience, etc. come into play, not to mention the overall alumni interest in the school.

The biggest issue for women's collegiate basketball, though, is that the students don't come, even if it's free. Very, very few schools draw student fans in numbers even close to the men's team, even if the men are much worse competitively. And female students are no better, going to the men's games but not attending the women's games.

_________________
Oṃ Tāre Tuttāre Ture Svāhā

[LegoMyEggo]

I haven't read through every post here, so forgive me if being redundant. I look at Knoxville, TN or Ames, IA as examples of locations where there is very little competition for the sports dollar. By comparison, a place like Maryland, UCLA, etc. are surrounded by a TON of opportunities for the public to choose from. Of course, those places also offer a lot more choices for students to entertain themselves with too. It is an interesting study.