Sunday, August 02, 2009

Decline of racial profiling

I originally heard about this research at a talk at work. It was sort of cool how they did the analysis, because you can't really setup an experiment to test a hypothesis. You have to create one later from the existing dataset that can stand up to analysis.

So to make it the same time of day, they used a time that was dark part of the year and light during the other part. Daylight savings time helped out alot with that. Assuming that when it is dark you can't tell the difference between people (which you can't), then you can compare when they can tell race and when they can't.

I prefer this sort of data analysis to the anecdotes like in the comments. I am sure that some racial profiling exists, but it is shown to not be a rampant epidemic from the numbers. Some people suggest it is everywhere from a gut feeling which turns out to be false. (Calling you out, Obama!)

No comments: