from Amazon
Monday, October 29, 2007
Zimbalist’s regression
I just finished writing a post on another thread that shows that you can get a regression to tell you anything you want. This is how I ended it:
If GP approach infinity, r approaches 1. If GP approached 0, r approaches 0. You see, the correlation tells you NOTHING, absolutely NOTHING, unless you know the sample size.
Read the rest of the link. So, here comes Andrew Zimbalist, who I must say I highly admire, at almost the Bill James level, and is quoted as saying:
“If you do a statistical analysis [of] the relationship between team payroll and team win percentage, you find that 75 to 80 percent of win percentage is determined by things other than payroll,” says Andrew Zimbalist, a noted sports economist and professor at Smith College in Northampton, Mass.
However, I said this a while ago:
Finally, here is the overall average, 1992-2005, of team wins and payroll index. Correlation? r = .70. There is an extremely high correlation between wins and payroll.
...
So, the relationship between payroll and wins is a little tricky and it depends on exactly what the question is. There’s no question that the driver to wins is the base talent level. That talent level is not necessarily going to be paid at the proper levels year in and year out.
You can get a correlation to tell you r=.50 as Zimbalist is saying, or r=.70 as I’m saying, or anything at all that you want. It’s a matter of the sample size.