MLS Attendance Analysis

[voros]

I've begun to post the results of my analysis of the attendance of the 1,839 regular season MLS games.

http://numeridicalcio.wordpress.com/2006/09/12/mls-attendance-analysis-intro/
http://numeridicalcio.wordpress.com/2006/09/14/mls-attendance-analysis-part-i/

[kenntomasch]

Great work, voros.

Even if I have precious little idea what any of it means.

It seems to me, math-challenged as I am, that the #1 variable on your list is Saturday matches. Which lends credence to the notion that dates are hugely important.

[voros]

Great work, voros.

Even if I have precious little idea what any of it means.

It seems to me, math-challenged as I am, that the #1 variable on your list is Saturday matches. Which lends credence to the notion that dates are hugely important.
Right, it is worth pointing out that significance is based on not just the strength of a relationship but the frequency it occurs.

The variables for season opener, Freddy Adu and SSS are all actually slightly stronger in effect than the saturday effect. The saturday effect, however, is more prevalent throughout the sample, and so it gives us much more statistical significance. I'll get to a lot of this as I post explanations for each of the variables.

[JohnR]

Voros -

I forwarded your analysis along to an Italian friend of mine, who is not only a whacked-out Juve/Azzuri fan, but who is also a Ph.D. econometrician. He will enjoy your analysis and possibly have some stat-related thoughts.

I could make a stab at analysis but I would be in a bit over my statistical head.

[voros]

Great work, voros.

Even if I have precious little idea what any of it means.
Yeah, I'm sorry. It's tough because I'm doing something very specific and I'd like to explain what it is my doing, but obviously it often reads like Greek. Combine that with my tendency for verbosity and it can become dense at times.

The variable explanations should help as with all of the general explaining out of the way now, I can just say what happened.

[voros]

I could make a stab at analysis but I would be in a bit over my statistical head.
Believe it or not, I probably am too. But I've tried to be both fair and logical, and have tried to stress that by no means have I solved any type of vast puzzle.

[JohnR]

My buddy says that you clearly understand how to run such studies and you don't need his advice. Only thing he says is that if you had a variable for actual weather conditions as opposed to average weather stats that would probably be instructive, but he appreciates the difficulty of obtaining that data.

[wolfp10]

Have you considered trying to have this published in an academic or somewhat mainstream journal or magazine?

BTW, for day to day weather conditions, Weather Underground has an excellent almanac which contains a very lengthy archive of weather data.

[cpwilson80]

Incredible work. Can you please cut and paste the variables that lacked significance and email them to MLS GMs? :)