Crime Statistics at Big Universities
I have continued my efforts to make sense out of the Clery Act crime data and made a couple more interesting discoveries. In trying to limit noise and floor effects in the data, I’ve been focusing on large 4-year schools with dorms. After eliminating secondary campuses and mislabeled schools, I found 54 universities with at least 30,000 students fitting the other criteria. Even at those big schools, most of the tracked crimes rarely occur. For example, from 2008 to 2012, for any of those 54 schools, there was 1 murder (Virginia Tech, 2009) and 2 manslaughters (Florida State, 2011; Michigan State, 2012). Even data for the tracked crimes of aggravated assault, arson, car theft, non-forcible rape, and robbery at the 54 biggest campuses in the country are so infrequent that any incidents are difficult, if not impossible, to differentiate from noise. That leaves only two Clery Act crimes that are reported to occur with a high enough frequency such that we might hope to learn something comparative about on-campus crime: forcible rape and burglary.
In the “real world” (non-campus environment), crime rates are highly correlated on a national and per-jurisdiction basis. It was those strong correlations that allowed me to find outlier rape data in my recent study. The story with the 54 largest universities is quite different. I averaged the burglary and forcible rape crime rates (incidents per 100,000 people) for the 5 most recent years with data (2008 to 2012) for each of the 54 schools. I then computed the percentile of each school in relation to the other schools for both their burglary and rape rates. The results are on the graph below:
With the possible exception of a few clustered very safe and unsafe schools, the burglary and rape rates appear to be uncorrelated. The overall trends of the two crime rates are very different as well as seen in this graph:
I’m not sure what to make of the campus data. Burglary has decreased by about 14% while rape has increased over 60%. At first blush, I think I could draw one of four conclusions: 1) campus crime is radically different than in the rest of the U.S.; 2) the 54 large schools are not representative; 3) the data is just garbage; or 4) some unobserved variable is causing reports of rape and burglary to follow the observed patterns. I’m hoping 1) isn’t true, but can’t rule it out. 2) is possible, but the trend lines are also consistent across all schools. 3) doesn’t seem as likely for the big schools where there is less noise (but that doesn’t mean the data is accurate). I think 4) is likely to be the best explanation and events like the Freeh Report might explain some of the answer. I’m more than happy to hear other thoughts, though.