Search This Blog

21.1.08

Shreveport Times irresponsibility could hurt policing

Maybe it’s because it appears to have dropped quite a chunk of change on getting the data for the stories. Or maybe it wanted to make a big splash for the Martin Luther King, Jr. holiday. Whatever the motivation, the Shreveport Times produced a series on “racial profiling” riddled with muddled thinking, inappropriate insinuation, and harmful public policy prescriptions.

The Times studied traffic citation data and found that black drivers were receiving perhaps twice as many citations – usually for minor offenses – as were white drivers in both Shreveport and Bossier City given population levels. On this alone, it concluded there must be something “troubling” about this outcome, without ever spelling out exactly what that is.

But instead of providing a fair and balanced analysis of the issue (the closest it ever came was one quote from my colleague at LSUS who pointed out the many factors that go into decisions made by police about traffic stops and admonished it to “do a little more work” before making any conclusion), instead it made the elementary mistake, intentionally or otherwise, about which I warned my statistics students on many occasions to avoid, treating association as if it were causation.

As an example of this error, suppose you observe that the larger a fire is, the more firemen there are that show up to fight it. As a result, you conclude firemen cause fires: where there are no fires there are no firemen, and the size of the fire grows directly with the number of firemen present. What you have done is erroneously posited a relationship between to things based solely on the evidence that they covary. They are associated, but that does not then imply they must have a causal relationship. Without paying close attention to plausible theory, any causal conclusions are recklessly made.

Causation is demonstrated only when there is theory that most plausibly explains why there must be a relationship and what affects what. On this issue, The Times and a number of “experts” (with the exception of an academician all vocal politically liberal elected officials or activists) used the association of race determining probability of ticketing and grabbed their ideological biases to give an immediate, unreflective, erroneous conclusion (that, again, they are reluctant to state unambiguously no doubt because it would sensitize readers to that bias and lead them to dismiss their argument): blacks are disproportionately ticketed because local police departments refuse to change irredeemably racist institutionalized policies and to punish racist behavior.

This is despite the fact that local departments (including Shreveport’s black police chief) steadfastly maintain that “hard” profiling (with race being the only standard being used to decide whether to make a stop) is forbidden. It is despite the fact that policies are in place to prevent hard profiling (which is dismissed by The Times and the consensus of its interviewed as unsuitable without releasing data to the state). It is despite the fact the departments handle and resolve complaints of this. It is despite the fact that in three crucial ways the analysis failed to explore evidence that could disconfirm their hypothesis.

One was to look at the conviction rates of the offenses cited. That data is not presented but chances are it was pretty high. In other words, legitimate crimes were being committed, but in one sense The Times is nevertheless blaming the police for being too vigilant. Its response probably would be that this shows selective enforcement, or that white motorists committing the same crimes were being given a pass because of either cut-and-dried racist attitudes among the police and/or because they were making race the primary factor in deciding about a traffic stop.

This leads to the second piece of evidence that The Times easily could have (or did?) gather – the race of the citing and/or arresting officer. One would expect that if hard profiling if not racist behavior was going on, you would find white officers ticketing significantly more black offenders than would black officers. Indeed, the data collected by The Times contained the names of the officers; unless the departments objected, it should have been little problem and just taken a few extra hours of coding to put that information into the database, and then analyze whether officer race made a difference in proportion of tickets issued by race.

But The Times didn’t do this. If it couldn’t because of departmental objections, it needed to act responsibly by noting this argument and tempering its own. If it could but didn’t, this was a simple, stupid mistake that greatly detracted from the series. If it could and did but chose not to report the results, it was disingenuous if not outright fraudulent. Even though their editor preaches about how there needs to be “transparency,” The Times appeared not to practice that, nor did it do sufficiently what he counsels the reading audience to do, analyze the numbers.

(Also interesting: The Times has a comment feature on its stories where registered users can leave notes which are moderated. For the stories in this series, the function was disabled – doesn’t seem like they want a whole lot of debate on this despite their editorializing in favor of it.)

This neglected the probable true explanation, the final bit of evidence ignored by The Times, for why we see differences in proportions cited by race is a possibility confirmed by every statistic and study on crime in America: like it or not, unfortunately blacks simply are more likely to commit crime, in almost every category of crime. It may well be that, for example, you have, given relative population, twice as many black motorists being cited for windows too heavily tinted or music played too loud because twice as many engage in that behavior. But, once again, The Times could, and maybe did, excuse itself from entertaining and investigating this notion by blindly believing crime rates are disproportionately higher for blacks because of that old bogeyman racism again – even as what federal government evidence does exists seems to indicate blacks are more likely to operate vehicles unsafely than whites.

While it is undesirable that The Times should choose to create an issue out of nothing or at least half-baked, the real damage can come as its insinuations could detract from law enforcement optimally carrying out its job. As summarized in a report about the issue of racial profiling which revealed the ideological politics behind the crusade against it in all forms and gave examples of the harm a prior assumption that racism was behind the use of race in fighting crime, “The anti-profiling crusade thrives on an ignorance of policing and a willful blindness to the demographics of crime,” that threatens to shatter the “commonality between law-abiding inner-city residents and the police.”

In the final analysis, it’s not bad that The Times looked into this issue, it’s that it did so in such an manner, neglecting a complete, thorough, and balanced analysis, that has the hallmarks of wanting to promote an agenda. One argument it made in its series was that the interpretation they supplied to the data when circulated would lead to greater distrust of the police and thus hamper crime fighting, unless police behavior subsequently changed. It was spectacularly irresponsible of The Times to float an easily-detected flawed argument, raising alarms about the results of such an incomplete, if not ideologically-driven investigation that threatens to lead to the outcome it so piously claims to want to prevent.

1 comment:

Beaumont said...

This rebuttal is well thought out and concise. Congratulations on exposing the media manipulation that occurs every day. I came here from a site called patterico.com where a commenter handled DRJ posted this link. Keep going.

Mike