Friday, June 20, 2008

Crime statistics

Interesting post by the Wall Street Journal's The Numbers Guy about the possible undercounting of murders in Houston, and he makes two points about crime statistics: "(1) They are self-reported, and therefore subject to fudging; and (2) Despite their weaknesses, the stats have a broad influence."

I spent a couple of years going to school and living in the Hyde Park section of Chicago (now famous as the home of Barack Obama). At the time, the bordering areas hadn't seen any gentrification, so Hyde Park was considered an enclave of safety in the midst of a pretty nasty place. And official Chicago Police Department (CPD) statistics bore that out.

Yet...the first day on campus we were "oriented," and a lot of time was spent on how careful we should be, and not just if we ventured out of the roughly 1 square mile of Hyde Park. It was made very clear, especially to female students, that caution needed to be exercised within the boundaries of the neighborhood.

Some of this was undoubtedly for the benefit of the urban-challenged among us; if you had spent your young life on the farm, then went to Iowa City for college, it would be natural to suppose that your street-smart instincts might be somewhat lacking even in a relatively safe area of Chicago. But the message still seemed somewhat at odds with the concept of Hyde Park as safe oasis.

As we soon learned, the "oasis" idea was not quite right. While the area was not anarchic lawlessness, neither was it the urban equivalent of a leafy suburb. It was part of a city, a sometimes-dangerous city, and vigilance was not to be discarded.

What was happening, and this was borne out by talking to people who had lived there longer, was that campus security was handling a lot of the crime that would, in other neighborhoods, be the sole province of CPD. And not simply handling it, but failing to report it (I did not know then, nor do I still know, what the requirement is for this reporting, so I cannot say that there was anything legally wrong with that practice).

Ever since that time, I have taken reported crime statistics with a huge grain of salt, and treated them like any other statistics. Try to understand any incentives people have for reporting them as they do (in the Hyde Park case, a big desire to make the university look more attractive), look for alternative explanations (the presence of a surprisingly large campus police force), and understand that, in the real world, small gradations are rarely significant.

No comments:

Clicky Web Analytics