Update: Prison/Jail Population Allegations of Racism

At the beginning of the month, I ran a correction: I had used data from a Leonard Pitts column in Growth Rate for Jail, Prison Exceeds That For Population Growth. Pitts said "a black drug defendant is 48 times more likely to be jailed than a white one with the same record." However, Pitts dropped a key qualifier: the original factoid dealt with juveniles. But even the original "fact" is misleading.

I explained in the correction that I had been unable to find the "48" factoid in the original report, And Justice For Some, identified by an alert reader. I have in hand an e-mail from Mark Soler, the Executive Director of the Center for Children's Law and Policy -- the publisher of the report (emphasis added):

The statistic does, in fact, appear in the report, on page 20, in Figure 12, as part of the graphic on Drug offenses. As correctly stated in the Building Blocks Advocacy Guide (which I wrote), "African-American youth adjudicated for drug offenses with no prior admissions were committed to state institutions 48 times as often as White youth adjudicated for drug offenses with no prior admissions." The source of the information is clearly stated at the bottom of Figure 12 as a report by the National Council on Crime and Delinquency, and the states surveyed are listed at the bottom of Figure 12.

This stick of dynamite was extracted from a data chart by a person writing a media-oriented summary. Ask yourself: why would the authors fail to mention such a startling "fact" in the text of that section?

I will leave it to reader to determine just how easy or difficult it is to deduce any factoid from this chart. If the lack of legibility gives you an eye-ache, read on (my eyes suffered so that yours wouldn't have to!):

• The factoid implies that all 50 states were surveyed. That's not the case: data reflect only 37 states.
• The report was released in 2000; the data in this chart are from a report dated 1993! As a reminder, Pitts misrepresented the factoid in 2006, implying it was current. We know next to nothing about the data: we don't if this is for one month ... or one year ... or perhaps an average of five years. Nor do we know the year/s. And what's the age range? It varies by state: "Rates are calculated per 100,000 youth age 10 to the upper age of juvenile court jurisdiction in each state."
• The rates for youth adjudicated for drug offenses with no prior admissions follow: whites, 1/100,000; blacks, 48/100,000; Latino, 13/100,000; other, 2/100,000.
• The rates for youth adjudicated for all offenses with no prior admissions follow: whites, 59/100,000; blacks, 373/100,000; latino, 166/100,000; other 80/100,000.

For the calculator challenged: that means black juveniles are slightly more that six times as likely as white juveniles to be incarcerated in state institutions in 37 states, time-frame unknown. This means that in some category (check out DUI), whites must be more likely than blacks to be incarcerated. Based on the logic in this report, that means that the judges must be biased against whites in those cases.

These data may or may not reflect racial bias. They might reflect socio-economic bias. The difference might be explained by the age of the juvenile (since the range has a variable upper end point) or the nature of the alleged crime ("drugs" is a very wide category). Are blacks disproportionately likely to live in inner cities, where crime rates are statistically higher? We Simply Don't Know. The chart raises more questions than it answsers.

Another reader commented on the correction:

This paper is methodologically flawed. The youth in question are neither matched for offense nor are they matched for criminal or sociological background.

The only form of "matching" for background that is performed between white and minority youth is the number of incarcerations; thus a kid with an arrest record "as long as your arm" (but no incarcerations) could be compared to a true first offender. An honor student could be compared to a dropout.

The "matching" for offense (in the case of the infamous factor of 48) consists only of the crude categorization "drugs". Thus, a kid caught with two hits of Ecstasy in his high school locker could be compared to a kid picked up on a street corner with twenty vials of crack, a pager, and $1200.00 in cash.

Given this lack of matching of subjects, the paper simply cannot say anything about racism or bias in the justice system.

The politics of drugs and the US "war on drugs" is a topic for another day. Of course, were drugs legalized and taxed, we wouldn't be having this discussion. [When was the last time you heard of someone trying to sell bootleg vodka on a US street corner? Apologies to Sam Harris.]

So there are two things to take away from this: don't trust anyone who doesn't give you a data source and remember to read the fine print. Oh. One more: correlation does not prove causation.

Technorati Profile
Technorati tags: Crime, Race, Politics
gada.be tags: Crime, Race, Politics

^{Category: Election Issues}