Continuing with Major League Baseball today: is team payroll the most important driver of on-field team performance? Historically (1988 - 2011) MLB's payroll and performance relationship has been statistically significant, but not that strong (common variation between relative payroll and winning percent is less than 20%). So I decided to grab the MLB payroll data from USA Today and ran the regression examining payroll and performance for just the 2013 MLB regular season. How do you do this? Here is a step-by-step guide.
Admittedly, one season is a small sample (30 observations), but what I find is that payroll "explains" about 7.7% of performance (using the adjusted r-squared) and that payroll itself is statistically significant at only the 90% confidence level - (i.e. t-stat is less than two in absolute value). Typically I would conclude that if the t-stat is less than two in absolute value (as is the case here), that the independent variable (payroll) is statistically insignificant (i.e. statistically is zero - thus having zero effect). Yet even if you are OK with the weak statistically significant results, the last nail in the coffin is the very weak impact that payroll has on winning percent for 2013. So, if there was a $100 million increase in payroll that would only increase wins by 8.72. Since more than half of the teams spent less than $100 million in total last year, this is a rather large increase in spending to achieve an increase in winning percent of approximately 0.05 more towards winning percent.