Jul

24

2013

Thabiti Anyabwile|10:21 am CT

Why Statistics Don’t Justify Our Prejudice or Our Profiling

Over the past several days I’ve had a number of exchanges with good people perplexed about what to do with “racial” profiling. Most of these persons have focused not so much on public policy but on their own hearts and fears. They’ve been concerned about their own reactions in situations that, to them, require some level of profiling. They think they’re being “rational” in their profiling or prejudice. And that’s what bothers them most. They think the failure to profile represents an irresponsible risk, and yet they see the injustice—potential and real—of profiling and stereotyping.

Most all of these people are white, work in office settings and have advanced degrees. But few of them actually work with statistics for a living or have much training in their use. Yet, the sole factor that makes them feel rational and justified in their profiling are national crime statistics (in fact, it’s not an actual statistic at all, but a general sense that African American men commit more crimes). They’re left wondering, How do I account for disproportionate rates of crime when it comes to my personal interaction with African-American men?

I’ve been asked this enough and seen it on enough media outlets that I thought I’d offer a couple comments. Take them for what they’re worth.

A Few Common Mistakes with “the Statistics”

Do national crime statistics provide any meaningful information for personal safety? Most people would like to assume so. But, depending on how you use the statistic and what conclusions you draw, you’re actually quite likely to further misrepresent people and worsen the problem of “racial” profiling. Here are five ways misusing statistics create more problems.

Aggregate statistics don’t predict actual situations. In introductory statistics classes, there’s usually some illustration meant to make the student aware of the limits of correlations, that most basic of statistics. You’re told about worldwide ice cream sales spiking in December when temperatures are lowest outside. Then you’re asked if winter causes hunger for ice cream. Of course the answer is no. Correlations do not indicate causality. Some further stats are then used to explain why ice cream sales go up in the winter and we get to see the limits of correlation. When people refer to aggregate statistics as justification for their profiling, they’re making unwarranted predictions about their interaction with African-American men. In this case, the statistic neither causes nor correlates with the lived reality of most white people.

Aggregate statistics include redundancies. How many offenders have committed multiple crimes recorded in these aggregate stats? Significant numbers of offenders commit other crimes after their release. Multiple crimes can be attributed to one individual or group of persons. Take, for example, the incredible rates of violent crime in Chicago right now. Those aren’t crimes randomly committed by random individuals. It’s highly likely that a smaller set of people connected by some other factors (i.e., gangs, drugs, revenge) are responsible for the surge in assaults. If the crime statistics include multiple offenses committed by a smaller number of persons, then it’s not rational to view every African-American male as though he’s a likely criminal ready to assault you. This is called over-generalizing. It’s applying a statistic about some people with particular characteristics to all people whether or not they  share the key characteristics.

Aggregate statistics include geographical concentrations. Consider again the statistics on violent crime in Chicago right now. No doubt Chicago’s crime rate disproportionately contributes to any national crime statistics. Yet few people citing statistics as justification for their informal profiling take that into account. They cite the statistic as though the crime that’s likely in Chicago is just as likely in Des Moines or Greenville. But that’s not true in any measure. The reality is that most people commit crimes where they live. We can’t generalize from one neighborhood to another in this way, especially if we’re generalizing to really dissimilar neighborhoods that lie at some distance from Chicago.

Aggregate statistics include systemic anomalies. The disparities in criminal justice procedures are well-documented. African-American men—especially poor black men—receive stiffer penalties and are convicted at higher rates than other men committing the same crimes. If you want to think more about this, you might consider reading Michelle Alexander’s The New Jim Crow. When we tout “the statistics” we need to recognize that we’re citing systemic biases as well as criminal behavior. Doing so likely inflates the informal probabilities we’re using to justify our profiling.

Aggregate statistics don’t predict inter-”racial” interactions. Again, people tend to commit crime where they live. People also tend to commit crimes against people they know. That’s especially true with assaults and violent crimes where roughly two-thirds to three-quarters of such crimes are committed by people known to the victim. Most of us are more likely to suffer at the hands of someone who looks like us and is known to us than we are by the stranger walking down the street. So, if we’re white, taking aggregate crime statistics about African Americans and applying it in some informal profile is less rational than doing the same with other whites you encounter. It turns out that African Americans are more rational if they profile other African Americans from their neighborhood. Aggregate crime statistics simply fail to justify the profiling that’s happening.

Better Data for Profiling

I think it’s impossible to avoid all profiling. And I think it’s irresponsible to try to avoid all profiling. We all categorize information and draw conclusions. The human mind seems to do that efficiently and automatically.

The issue is whether or not we’re using the appropriate data when we draw our profiles and calculate risks in any given situation. A blanket and imprecise appeal to “statistics” seems to me an improper approach to either protecting ourselves or treating others with respect and dignity.

We might be better off using a simpler set of data: our own personal experience. In your last ten interactions with someone of another “race,” how many times have you actually been verbally accosted or physically threatened? How about the last 50 interactions or 100? It’s not that it never happens. It does. When it does, we understand why persons would be afraid and more vigilant. But for most of us it hasn’t happened–at least not to the level that we can justify our fear and our profiling.

And that’s the heart of the issue for most people–fear. What do I do with my fear? Do I justify it? Do I reject it? Do I act on it? And what must I think of the “other” in a way that responsibly addresses my fear?

The first step would be to admit just how powerfully our fear may be acting upon us. We need to disabuse ourselves of the notion that we’re being “rational” with these statistics when deep down inside we know we’re trying to soothe our fear and justify its existence. Better to face our fear and conquer it with a legitimate use of statistics and information. We don’t want our fear to make a lot of bad decisions for us. We’ve not been given that spirit.

Categories: race

View Comments (51) Post Comment