Putting Ray Lewis' crime prediction to the test
[UPDATE, Wednesday 11:40 AM: I asked one of the country's leading criminologists, Rick Rosenfeld of the University of Missouri-St. Louis, to evaluate my analysis. He got back to me today and here's his response: "Your method of testing the theory that football reduces crime is perfectly fine. You were right not to rely on the single result involving the bye week. The other week-to-week comparisons are valid. Nice!" So there you go.]
So the Ravens' Ray Lewis says crime will increase in the event of an NFL lockout. "Do the research," he told ESPN. "Watch how much evil, which we call crime, watch how much crime picks up when we take away the game."
Can we quantify this theory? Well, we can sure as heck try.
In a highly unscientific analysis, I took data from the city's Open Baltimore web site that allows us to look at crimes recorded by day, and maybe Lewis is on to something: As it turns out, the Ravens' bye week was the Sunday during the NFL season that saw the highest number of crimes. The following week, when the team resumed play against the Dolphins, was all quiet, relatively speaking.
Week 7 (Oct. 24 vs. Buffalo Bills) - 128 crimes recorded by police
Week 8 (Oct. 31 - Ravens' bye week) - 158 crimes
Week 9 (Nov. 7 - vs. Miami Dolphins) - 102 crimes
That's a 23 percent increase during the bye week, followed by a 35 percent decrease once playing resumed. Only two other NFL Sundays saw more than 130 total crimes recorded (Week 3, 6, and 17), and most were in the 102-125 range.
Did I mention this is unscientific?
Beyond the bye week test, other attempts to measure this issue seem to cast doubt on whether there's a correlation. A comparison of the first few weeks of the season to the weeks that preceded kickoff Sunday - and similarly, the playoff weeks compared with the weeks that followed the team's exit - seems to dump some cold water on the analysis:
Aug 15, 22, 29 and Sept. 6 (the Sundays comprising the pre-season), and Sept. 12, 19, 26, and Oct. 3 (the first four Sundays of the NFL season), saw the exact same number of crimes: an average of 130.5 incidents recorded by police. The final weekend of the season and the weekends of the Ravens' two playoff games actually saw more crime than the three weekends after they were knocked out - an average of 222 crimes while they were in contention, and 195 after they were sent packing.
There's one more scenario to consider. There's been two prior work stoppages, in the 1980s. While we don't have detailed data for those years, a month-by-month look at homicides shows that there was an increase in homicides during the months of the 1982 strike, with killings increasing from 33 to 44. During the 1987 strike, there wasn't any noticeable difference, though Baltimore also didn't have a football team by that time.
What's that, you say? This analysis fails to take into account weather, school schedules, police deployments, the fact that the team played some of its games on Mondays or Thursdays, and a whole host of other mitigating factors?Well, to that I say: I told you it was unscientific.