Fan Bases and Winning Games

Seem like they're one and the same, right? If you win, people will come, right? Well yeah, but that's not the biggest reason.

Looking at the NL for the last 4 seasons, the biggest determinant of attendance was team payroll - if a team spent a certain amount of money, so many fans would come. Now, of course, this may actually be a faulty cause and effect; it might be that teams know their market and get what their attendance will afford. But still, take a look at this:

Regression Analysis: Att versus W, Pay
The regression equation is
Att = 199189 + 10498 W + 0.0222 Pay

Predictor - P
Constant - 0.608
W - 0.050
Pay - 0.000

S = 374825 R-Sq = 70.5% R-Sq(adj) = 69.5%

That is a very tidy regression. That is also a very low P-value for pay, indicating that we are very very sure that payroll affects attendance. Wins would appear to have some impact as well on atendance, but not entirely.

But this is more telling:

Regression Analysis: W versus Pay, RD
The regression equation is
W = 81.2 - 0.000000 Pay + 0.0969 RD

Predictor - P
Constant - 0.000
Pay - 0.975
RD - 0.000

S = 4.02786 R-Sq = 84.0% R-Sq(adj) = 83.5%

Taking into account the effect of payroll on run differential, there is some effect, so you are going to run into some multicollinearity, but still, the R-squared value is 20.7%.

It's really hard to define a model that actually tells you what you want to know, but my intuition seems to support the following: payroll is for putting fans in the seats, not winning ballgames. The Yankees have a $200,000,000 payroll, sure, but they're wasting money on guys like Carl Pavano in all that. Having good drafting and scouting helps. Being able to sign free agents is nice, but there are few free agents that are really worth what they might cost.

It is with that in mind that I ask whether one can really be sure if Matt Kemp is better than someone like Torii Hunter or Andruw Jones. He's certainly better than Juan Pierre.