Monday, October 11, 2010

Effect and Cause

People love crime.

It's true. Anything crime-related always gets a ton of hits on the website. I've talked about the popularity of crime databases before.

There are really practical reasons for this. For instance, the U.S. Department of Justice's National Sex Offender database gives moms a reason to worry and check up on the history of piano teachers and whatnot before leaving their child alone with them. It also provides a real economic incentive: if you live near sex offenders, you can find that with a few simple clicks, which can potentially scare away buyers from a piece of property you're trying to sell.

And considering just about every police interaction with the public is required by law to be public record, and at least in Chapel Hill they're typed up in a very formulaic way, there's a lot of good data potential.

But they are something that it's important to keep a close eye on.

Just the other day I was typing up crime reports for The Daily Tar Heel and noticed something odd. It was a break in like any other. Under items stolen, it listed $1,000 cash and a $75 Apple laptop.

There seems to be a very logical explanation for this: the police officer switched the two amounts.

But we don't know that. That's something we'd have to confirm.

So I called the Police Department to see if they could explain it, but of course, the only person who could answer my question left at 5 p.m. So I simply had to replace it with a Carrboro police report.

I've been doing these reports since February, and this stuff happens all the time.

This raises the point: when you've got a $75 laptop and $1,000 in cash stolen, you've either got a good story or bad data.

We got some bad data.

Where do we go from here?

This was an easy one to sport. But how often does this happen where it's not so obvious? When we just go by reports, how often do we insert fact errors and never, ever know it?

Well, we won't ever know it. No one's going to send us a correction for switching the values of the two different bicycles when one's $120 and one's $130. But it's still a fact error, even if it seems insignificant.

But there are ways this could seriously confuse people. What if police mix up the location of the incident, saying it's "Smith Street" instead of "Smith Avenue"? If we put that on a crime map, those people on Smith Street could get really scared over something miles away. But it's exactly what the police told us.

If we get something wrong in a police report, as long as it doesn't libel someone, we don't normally get corrections and we never know. It passes onto the next day's reports and people forget about it.

But with things like databases, that's preserved forever. And we use it for trends. Trends that could be false.

No comments:

Post a Comment