I’m just curious…
How easy is it to determine the race of drivers using the methodology these researchers employed?
How would you separate brown-skinned blacks from brown-skinned Southeast Asians, for instance?
And light-skinned black people can look like anything. White. Hispanic. Asian.
The researchers used a device that would take photos through the windshield rather than through the passenger/driverside windows (where cops sitting on the side of the road have visual access). One can argue that the windshield is more subject to sun glare, shadows, and other confounders. I saw a glimpse of the kind of photos the researchers used on the news, and I would have a hard time classifying the race of the drivers going on just those things alone. Granted, I’m not a ID expert, but I think there’s lots of room for ambiguity.
Also, I wonder how many cops find perps just by hanging on the side of the road, waiting for them to speed by. They now have those radar guns that don’t require you to be stationary in order to clock speeds. Is it possible that officers can zoom in and out of traffic, looking for (and at) people to pull over? I’ve never really thought about it before, but I’m sure it’s possible.
Another problem with the methodology is that anyone clocked at least 15 mph over the speed limit was considered speeding. This is in fact speeding, but in most places–particularly the NJ Turnpike–EVERYBODY goes this fast. I’m a slow-poke driver, but even I find myself going heavy on the gas on the Turnpike, just so that I don’t get tail-gated and cussed at by drivers behind me. If there are more white drivers than black, it seems to me that you’re statistically more likely to find whites going under this limit than blacks.
Think about it. If you have 100 drivers–70 white, 30 black–then all 30 black drivers could be speeding just to keep up with the 30 white drivers who keep riding up on their tail. In other words, we aren’t really talking about completely independent variables when it comes to comparing the behavior of white drivers and black drivers on the same road.
**
I know this is the assumption you are working from, but problems with this assumption are at the heart of the controversy, I believe. I would have more confidence in the data if they had verified their determinations by tracing licence plates numbers to informaton on driver’s licenses. That way we could be sure that someone they designated as black or white really wasn’t something else.
I think the Feds are right to be cautious about this. First off, racial profiling has been admitted by NJ state troopers, so I think it’s too late to try to cover their ass with a study–however strong the data are. Secondly, the very fact that NJ Justice Department commissioned this study–when there are obviously political implications involved–should raise eyebrows. I’m not saying the researchers did shoddy or unethical work, but you always have to be cautious about data which favor the agency who’s funding the research. That’s just a general rule of ethics. And lastly, it would be horrible PR for the state if questionable methods were come to light. And NJ doesn’t need any more bad PR, especially regarding the issue of racial profiling.