« "To change the world, start with prisons" | Main | Florida has first capital case head to jury sentencing after Hurst-required reforms »

March 24, 2016

Pew develops new "punishment rate" metric to provide more nuanced perspective on state incarceration levels

Via email today I learned of this intriguing new report from the folks at Pew Trusts titled "The Punishment Rate: New metric evaluates prison use relative to reported crime." Here is the short data-heavy report starts and ends:

Researchers, policymakers, and the public rely on a variety of statistics to measure how society punishes crime. Among the most common is the imprisonment rate — the number of people in prison per 100,000 residents. This metric allows for comparisons of prison use over time and across jurisdictions and is widely seen as a proxy for punishment. States with high imprisonment rates, for example, are considered more punitive than those with low rates.

A more nuanced assessment of punishment than the ratio of inmates to residents is that of inmates to crime— what The Pew Charitable Trusts calls the “punishment rate.” This new metric gauges the size of the prison population relative to the frequency and severity of crime reported in each jurisdiction, putting the imprisonment rate in a broader context.

Using the punishment rate to examine the U.S. criminal justice system, Pew found that all states became more punitive from 1983 to 2013, even though they varied widely in the amount of punishment they imposed. The analysis also shows that the nation as a whole has become more punitive than the imprisonment rate alone indicates....

The long-term rise in U.S. imprisonment is a familiar story. Although the imprisonment rate is an essential tool in understanding correctional trends, it paints an incomplete picture of the nation’s and individual states’ punitiveness because it does not take crime rates into account. The punishment rate provides a more nuanced assessment by placing each jurisdiction’s imprisonment rate in the context of the severity and frequency of its crime.

Analysis of punishment rates over time and across jurisdictions makes clear that the nation has become more punitive. What’s more, many states punish crime significantly more—or less—than their imprisonment rates alone indicate. States with particularly high or low punishment rates and those that experienced significant increases in their punishment rates over time may benefit from identifying and examining the policy choices responsible for their rankings and trends.

Helpfully, the folks at The Marshall Project have this interesting piece discussing what the new Pew metric does and does not tell us. That piece is headlined "The Tricky Business of Measuring Crime and Punishment: Pew researchers release a new prison scorecard, but it ain’t perfect," and here are excerpts:

We’ve grown accustomed to a quantified world of ever more complicated data available at our fingertips, on everything from how we sleep and eat to how often left-handed pinch hitters hit ground rule doubles on rainy days.  “The incredible databases of what we have for sports just blow away anything there is in criminal justice.  It's kind of crazy,” said Adam Gelb, director of Pew’s Public Safety Performance Project, adding, “We can't answer some of the most basic questions about one of the most important functions of a society.”

Nearly five years ago, Gelb and Pew started by looking at recidivism — how often people released from prisons are arrested again for new offenses. But using recidivism alone to compare how states are doing at rehabilitating prisoners fell short.  One state could have a lower recidivism rate simply because it tended to have more low risk offenders in its prisons.  So then, Gelb said he began thinking about how to assess whether the “right” people are in prison, that is the serious, violent and repeat offenders most likely to commit new crimes.

Pew’s punishment rate focuses on the most serious felony offenses that lead to a year or more in state prison.  The calculation divides each state’s imprisonment rate in a given year by the rate of crimes reported there, using the FBI’s Uniform Crime Reporting system.  To account for some crimes being more serious and more likely to lead to longer prison sentences, Pew weights the annual crime rates by calculating the average time served for those crimes each year.  After all of these calculations, Pew found that as America's imprisonment rate has gone up in the past three decades and as crime has dropped, the “punishment rate” rose by 165 percent.

While the methodology makes sense and is probably the best available considering the shortcomings of federal crime data, the punishment rate is not yet the magic metric. Unpacking the components of Pew’s punishment rate illustrates how tricky measuring criminal justice progress can be.  The punishment rate depends on the number of crimes reported by the FBI.  But the Uniform Crime Report, created in the 1920s, tracks only seven key crimes: murder, assault, rape, robbery, arson, burglary, larceny, motor vehicle theft.  It excludes dozens of offenses — most notably drug crimes, which have been a major factor in the growth of prison populations.  Pew’s report readily acknowledges that the Uniform Crime Report omits crimes for which roughly one-fifth of state prisoners are serving time.

“What that means is not to say that drug trafficking is not a serious crime, just that it's not reported and tracked in a way that you can support adding it to this formula,” Gelb said.  “It does mean that — other things being equal — a state that has a lot of drug enforcement activity and stiff sentencing for drug offenses will have a higher punishment rate.”

The other trouble with the punishment rate is in the lag between crime and judgment.  Pew is comparing the crime rate each year to the current prison population at that moment. It doesn't account for the people being sentenced each year or the prison intakes.  It also doesn't look at what crimes those in prison were convicted of.  So there is an inherent lag between when crimes happen and when someone might go to prison for them. Despite plummeting crime since the 1990s, the growth in the punishment rate didn’t overtake the rise in imprisonment until 2011.  That may be partially explained by the gap in time between crime and incarceration, though Gelb contends that effect is ameliorated by calculating rolling averages for offense severity (but not the crimes themselves or the imprisonment rate). He said the adjustment is meant to be a barometer of the seriousness of crimes in a year rather than a “fine-tuned calculation.”  But that lack of precision could undercut Pew’s implicit argument that in some states we are “punishing for punishment’s sake.”

I find especially important and notable Gelb's astute comment that the "incredible databases of what we have for sports just blow away anything there is in criminal justice." Especially as I am starting to prepare for my upcoming fantasy baseball draft, it is more than a bit disconcerting that I can easily find dozens of statistical projections for the Cleveland Indians' battery but no on-line sources to help predict how many batteries might be committed in Cleveland.

March 24, 2016 at 07:01 PM | Permalink


To use the sports reference, what makes sports stats useful is their ability to break data into discrete issues. In football, stats include not just the percentage of field goals that a kicker makes (a key but inadequate stat) but also the percentage at certain distances (avoiding the distorting effect of distance).

While these new stats are an improvement, they still distort what those of us on the ground "know." The rate of incarceration per crime is more useful than the rate of incarceration per person. Getting down to details would be more useful. How many people are incarcerated for first time drug possession offenses? (My experience is that -- even in rural counties with reputations for harsh sentencing -- the answer is very, very law.)

When data mixes the apples of first time offenders committing "minor" felonies with the oranges of repeat offenders committing serious felonies, it is hard to have a rational debate about whether we are "overincarcerating" people or to devise a proper fix if we are.

In my state, we are about to implement a reform of our criminal statues that took about eleven to twelve years. That reform has somewhat reduced the sentences for drug and other "non-violent" offenses while increasing the penalty for some "violent" offenses. My prediction is that the result of this reform will be an increase in the incarceration rate because prosecutors and judges were already giving low sentences (if any) on the non-violent offenses and were limited by the maximum penalties on the violent offenses. People looking at just the incarceration rate will say that this change is bad. Looking at it from a more holistic approach, however, the increase will be because we will be using prison space on those who most need to be in prison.

Posted by: TMM | Mar 25, 2016 10:05:24 AM


Hide that study. We are on a anti-crime rampage in this state and the last thing we need is a study showing that we rank on the near the bottom of a punishment index. It will make all the crime and punishment folks wet their pants.

Posted by: Daniel | Mar 25, 2016 12:07:11 PM

I, too, appreciate the need for new and better metrics, but I am a little disappointed that the Pew Trusts are claiming that this approach is novel, given that I published a paper in 2010 making exactly the same argument: that prison usage should be measured in terms of the Part I reported crime rate. My approach was to look at new felon admissions for a given year relative to reported crime, and this was used by the California legislature when it reauthorized funding for criminal justice realignment. I think this also points out the local/intra-state issues involved in crime-to-prison-usage: that within states, there is considerable variation due to policing and prosecution (a point John Pfaff has also made repeatedly). For those of you who might be interested in this intra-state focused approach in California, which, again, uses reported crime to new felon admissions as the ratio, the paper is called "Tough on Crime (on the State's Dime): How Violent Crime Does Not Drive California Counties' Incarceration Rates - And Why it Should". You can find it here, with all the data I used: http://digitalcommons.law.scu.edu/facpubs/162/

Posted by: W. David Ball | Mar 25, 2016 12:28:58 PM

One thing that Pew doesn't do, but probably should, is to estimate the magnitude of potential systemic errors (e.g. in weighting of serious crimes) on the result to demonstrate how robust its results are or are not.

Posted by: ohwilleke | Mar 25, 2016 2:38:10 PM

Post a comment

In the body of your email, please indicate if you are a professor, student, prosecutor, defense attorney, etc. so I can gain a sense of who is reading my blog. Thank you, DAB