« Is Wood v. Moss generalizable? | Main | Schemas, Shortcuts, and Software »

Wednesday, May 28, 2014

The Problematic National Research Council's Report on Incarceration: Some Initial Thoughts

The National Research Council, the well-respected research arm of the National Academy of Sciences, recently released a putatively authoritative report on the causes and implications of US incarceration growth. Sadly, it appears to be a deeply, profoundly flawed report. It is, in short, a rehashing of the Standard Story that I have argued time and again lacks real empirical support.

Dangerously, this report gives the Standard Story the NRC’s seal of approval, which will only increase its hold on policy-makers’ perceptions. The New York Times has already written an editorial pushing the NRC’s Standard-Story arguments, and no doubt it will be cited widely in the months to come.

So in the posts ahead, I want to dig into the report more deeply. I will certainly acknowledge what it gets right, but my sense so far is that it is one rife with errors. Many of these points will be ones I have made before, but they are clearly points that need to be repeatedly. In this post, I want to point out (once again) the troubling ways in which drugs crimes are given too much weight.

To the report’s credit, its conclusion does not focus solely on the war drugs, but rather claims that drug-related admissions drove growth in the 1980s and longer sentences for violent offenders drove growth in the 1990s and 2000s. However, the body of the chapter often takes a less nuanced tack.

Consider Figure 2-7:

Screen Shot 2014-05-28 at 10.10.57 AM

As the report explains:

Most striking, however, is the dramatic increase in the incarceration rate for drug-related crimes. In 1980, imprisonment for drug offenses was rare, with a combined state incarceration rate of 15 per 100,000 population. By 2010, the drug incarceration rate had increased nearly 10-fold to 143 per 100,000. Indeed, the rate of incarceration for the single category of drug-related offenses, excluding local jails and federal prisons, by itself exceeds by 50 percent the average incarceration rate for all crimes of Western European countries and is twice the average incarceration rate for all crimes, including pretrial detainees, of a significant number of European countries.

This is an argument that puts a lot of weight on drugs over the entire period. But Figure 2-7 is a very strange figure. Sure, drugs is the fastest growing category, because it is compared to all the index violent crimes separately. “Drug offenses” is no less a heterogenous category than “Violent crimes,” so why is it the only homogenous category?

It’s like saying the United Kingdom has more people than New York State, California, Ohio, etc. Sure, that’s true, but the United States has more people than the United Kingdom, and the big US states have more people than the big UK counties.

Take a closer look at Figure 2-7. Just eyeballing it, if we aggregate up all the violent offenses (which are all the other crimes except burglary, which is an index property offense—and strangely not the only one: where is larceny?), then the incarceration rate for “Violent Index Crimes” is about 115 per 100,000 in 1980 and about 280 per 100,000 in 2010. So violent index crimes are incarcerated at twice the rate of drug crimes in 2010. Drugs aren't as central as Figure 2-7 suggests.

But we can attack this argument even more directly. The NRC report is correct to argue that we shouldn’t view prison growth as a single monolithic process since 1975: surely the factors that mattered in 1980 differ from those in 1990 or 2010. And a lot of work on prison growth—including some of my own—does merge multiple decades together. But even then the report overstates the importance of drugs.

The report’s conclusion clearly puts the onus of growth in the 1980s on drug incarcerations:

The growth in imprisonment—most rapid in the 1980s, then slower in the 1990s and 2000s—is attributable largely to increases in prison admission rates and time served. Increased admission rates are closely associated with increased incarceration for drug crimes and explain much of the growth of incarceration in the 1980s, while increased time served is closely associated with incarceration for violent crimes and explains much of the growth since the 1980s. These trends are, in turn, attributable largely to changes in sentencing policy over the period….

But let’s look at the components of incarceration over the whole period. As I’ve argued before, between 1980 and 2009, over 50% of prison growth is due to increases in violent inmates, and only about 22% due to increases in drug offenders. But that number makes the error of treating the whole period as monolithic. Let’s break the results up into 1980 - 1990, 1991 - 2009:

Screen Shot 2014-05-28 at 11.49.34 AM

What are the percentages? Between 1980 and 1990, state prisons grew by 387,400 inmates, and 36% of those additional inmates were incarcerated for violent crimes. (The math is below if anyone wants to see it.*) Two things stand out here:

  1. The NRC is right that drugs mattered more during the 1980s than after, and that violent crimes played the dominant role in the 1990s and beyond.
  2. But even in the 1980s violent crimes mattered more. Drugs were important, but (by a slight edge) violent crimes even more so. US incarceration rates have always been a story about violence.

Actually, the “violence” point can be made even more strongly. Drug incarcerations were most important during a period of acute violence (I have shown elsewhere that in New York, the rise in drug incarcerations has nothing to do with the Rockefeller Drug Laws and everything to do with the onset of crack-related violence). To what extent are these drug incarcerations—perhaps unlike the drug incarcerations taking place in less-violent times—pretextual cases aimed at violent offenders whose violence is harder to prove?

Perhaps such pretextual use drug offenses is troubling—perhaps it is seen as undermining the presumption of innocence or of having disturbing racial effects—but it nonetheless cautious about much importance we put on drug sanctions even in the period when drug sanctions were most important.

That’s enough for now. I next want to consider the NRC's major flaw looking at admissions, and then turn to the problems with its discussion of time served.


* 387,400 is just 681,400 - 294,000; the number of new violent inmates is 140,300, which is just 313,600 - 173,300. 36% is just 140,300/387,400. All the other percentages are calculated the same way.

Posted by John Pfaff on May 28, 2014 at 11:52 AM | Permalink


Hello Professor Pfaff. I am curious as to how you are factoring in sentencing in your analysis of the contribution of violent criminals to the prison population. I think it's important to remember that violent crimes are generally prosecuted more harshly than non-violent criminals, and that includes prosecutions in regards to the 1980s and 90s. Therefore, even if less violent criminals are arrested, they will always be predisposed to make up a large concentration of the prison population. And according to the Bureau of Justice Statistics, drug criminals have accounted for a higher percentage of new prison admissions than violent criminals; To give an example, in 1991, according to Bureau of Justice Statistics, drug offenders constituted 28.3% of all state prison admissions, while violent crime accounted for only 26.9% of new prison admissions. In 2001, the difference is even more stark, with drug offenders accounting for 32.1%; almost a third, of state prison admissions, while violent criminals only accounted for 27.3% of prison admissions. Only in 2011 did violent criminals make up a greater percentage of state prison admission (29.1% vs 25.4% for drug offenders) but on a national level, that’s not even including federal prison admissions, which tend to have higher concentrations of drug offenders in their general population. How could violent crime have been the primary focus of the tough-on-crime era if more people were being admitted into prison for drugs than for violent crime in general?

(http://www.bjs.gov/content/pub/pdf/p12tar9112.pdf, See Table 4)

Posted by: James Richardson | Sep 1, 2016 4:18:23 PM

The comments to this entry are closed.