Lies, Damned Lies, and Media Stories on LAPD Traffic Stop Data

The Los Angeles Police Department has released statistics which break down traffic stops — and how they are handled — by ethnic group.
You can take a look at the raw data in a big PDF file here, and if you’re even more ambitious, you can check out the consent decree the LAPD reached with the Justice Department which led to this data being produced.
The consensus among big media appears to be that the data shows evidence of racial bias:
LA Daily News: “Latino and African-American pedestrians and motorists are more likely to be stopped and searched by Los Angeles police officers than other population groups, according to data the city issued Monday.”
CNN: “The Los Angeles Police Department released raw data Monday that seem to show African-American and Hispanic drivers and pedestrians are stopped and searched more frequently than whites and Asians. ”
Reuters: “Black and Latino motorists are three times as likely to be asked to step out of their vehicles by Los Angeles police during traffic stops than other drivers, according to data made public on Monday.”
I approached these numbers — and the big media stories about them — with extreme caution.
First, it is important to remember the Golden Rule of ethnic statistics stories: Correlation Does Not Necessarily Imply Causation. If a greater percentage of ethnic group X is stopped for a certain type of violation than ethnic group Y, that doesn’t necessarily mean that the police are using ethnic group X as a determining factor on who to stop. It could mean something very different: like, for instance, the simple fact that perhaps the police are perfectly just, and ethnic group X actually is committing the crime in question more often.
So at any rate, I took a close look at the statistics directly, then checked out the media reporting on them. My conclusion: as I suspected, the media in some cases seem to be playing a bit fast and loose with the data. But that said, in at least one area, the LAPD has, as Ricky would say, got some ‘splaining to do.
I focused my analysis exclusively on Page 3 of the document, which provides data for the entire city based on the driver of the vehicle for the period of July 1 through November 30th 2002. The total number of traffic stops for drivers identified as White, Black, and Hispanic were as follows:
White: 68266
Black: 37081
Hispanic: 78687
Then, we have the statistics for how many drivers were asked to exit their vehicle. The absolute number is direct from the report; the percentage is my own math based on the total numbers above:
White: 4483 (6.6 %)
Black: 8175 (22.0 %)
Hispanic: 17688 (22.5 %)
Then, there is the data for whether a pat down / frisk search was conducted. Not suprisingly, this follows the trend above (obviously, if you’re not asked to exit the vehicle, then you are not going to be frisked):
White: 2227 (3.3 %)
Black: 5437 (14.7 %)
Hispanic: 9707 (12.3 %)
Doesn’t look so good, does it?
Now again, there could be other factors involved here. Were blacks and Hispanics stopped in disproportionate numbers for more “serious” offenses that might be more likely to prompt caution by the police? Based on the data provided by the LAPD, it doesn’t seem so: they break down stops by the initial reason for the stop, and I could see no big differences by ethnic group. The vast majority of stops are for “Vehicle Code Moving / Pedestrian Violation” (83.6% for whites, 77.3% for blacks, and 78.9% for Hispanics).
So, my conclusion: while this doesn’t prove racial bias on the part of the LAPD in how they treat drivers after they are stopped, it sure does provide a significant bit of evidence that points in that direction.
Now, with that said, some of the scaremongering on the part of Big Media does seem to oversell the story. Let’s go back to the leads of the LA Daily News and CNN:
LA Daily News: “Latino and African-American pedestrians and motorists are more likely to be stopped and searched by Los Angeles police officers than other population groups, according to data the city issued Monday.”
CNN: “The Los Angeles Police Department released raw data Monday that seem to show African-American and Hispanic drivers and pedestrians are stopped and searched more frequently than whites and Asians. ”
(Emphasis mine). The problem here is that the LAPD report doesn’t actually tell you anything at all about how likely a person of a certain ethnic group is to be stopped in the first place, because it doesn’t include any data about the composition of the LA population as a whole. Without that data, you can’t make any conclusions about how likely a person of a given ethnic group is to be stopped.
Now, if you match the LAPD survey data with census data, theoretically you can make this kind of conclusion (although you’d still be fudging to some degree, because census data doesn’t necessarily correspond to what percentage of the driving population is of a given ethnic group). It’s possible that CNN and the LA Daily News both did this analysis, and that’s what they based their leads on. But we don’t know that, because they don’t say. And it’s telling that while both stories include the “more likely to be stopped” claim in their leads, neither of them repeat that claim within the body of the story: they both focus on the asked-to-exit-the-vehicle statistics and other detailed numbers that are actually in the LAPD report. (For the record, Reuters doesn’t make the “more likely to be stopped” claim, to their credit).
This story certainly isn’t dead; there’s a lot of work that could be done here in crunching these numbers in far more sophisticated ways than I (or any big media folks I’ve seen thus far) have. In particular, that cross-referencing with census (or perhaps DMV? ) ethnic data sure would be nice. (And the LAPD survey does break down the data by sub-area of the city; I just focused on the high level for simplicity). The city itself is actually looking to hire a professional consultant to do their own analysis on the data, in fact, so we’ll certainly be hearing more about that in months to come.
Anyway, thus endeth my foray into the perilous world of ethnic statistics. Hopefully I haven’t screwed anything up too badly, and above all, my recommendation is to read the actual report itself if you want to really form your own opinion. Which, of course, is always the safest approach…
Additional Note: One other odd thing about the LAPD statistics: they don’t always add up. Literally. When you total up the subcategories for some subsections that you would think would add to the total number of stops, they don’t. For example, looking only at white drivers, the total number of stops listed is 68,266. But when I add up all the subcategories under “Initial Reason for Stop”, I get 70,797. I encountered similar discrepancies using other subcategories.
There’s not much in way of explanation of the numbers in the report, so perhaps I’m not understanding the categories correctly. But it does seem a bit weird….