« Supreme Court grants cert to (finally!?!) resolve whether Miller applies retroactively | Main | Detailing the dysfunction of Pennsylvania's death penalty system »

December 13, 2014

"The Misleading Math of ‘Recidivism’"

The title of this post is the headline of this effective recent piece of reporting and analysis by Dana Goldstein for The Marshall Project.  Here are excerpts:

Recidivism, the rate at which former inmates run afoul of the law again, is one of the most commonly accepted measures of success in criminal justice.... [But] recidivism, though constantly discussed, can be widely interpreted — and misinterpreted....

In some studies, violating parole, breaking the law, getting arrested, being convicted of a crime, and returning to prison are all considered examples of recidivism. Other studies count just one or two of these events as recidivism, such as convictions or re-incarceration.

When the federal government calculates a state’s recidivism rate, it uses sample prisoner populations to tally three separate categories: rearrests, reconvictions, and returns to prison, all over a one- to five-year period from the date of release. In contrast, a widely cited 2011 survey from the Pew Center on the States relied on states’ own reporting of just one of those measures: the total number of individuals who returned to prison within three years.

Both the federal and Pew statistics leave out an entire group of former prisoners: those who break the law but don’t get caught. That’s why some recidivism research ... relies on subjects’ self-reports of illegal activity.

Another inconsistency across recidivism studies is the period of time they cover. Though three to five years is considered the gold standard, many studies examine a much smaller time frame. One recent study claimed that a parenting program for prisoners in Oregon reduced recidivism by 59 percent for women and 27 percent for men. But the study tracked program participants for only a single year after they left prison. The likelihood of reoffending does decrease after one year. But according to the Bureau of Justice Statistics, an additional 13 percent of people will be rearrested four years after their release....

In its 2011 Brown v. Plata decision, the U.S. Supreme Court cited California’s stratospherically high recidivism rates (according to the Bureau of Justice Statistics, close to 70 percent of former inmates in the state return to jail or prison within three years of release) as evidence that California prisons do not rehabilitate, but instead “produce additional criminal behavior.” The justices blamed recidivism on overcrowding and the lack of adequate medical services behind bars, and ruled those conditions unconstitutional. The ruling required California to decrease its prison population.

But what if the court’s take on the causes of California’s high recidivism rate is wrong? What if it isn’t primarily prison overcrowding that causes reoffending, but an overly punitive parole system — the same trend that drives the majority of recidivism in New York? That’s what the data shows. Parolees in California are actually less likely than parolees in New York or Illinois to commit a new crime. Yet they are exponentially more likely to be arrested and sent back behind bars for violating the conditions of their parole, according to an analysis of BJS data from researcher Ryan G. Fischer. California law punishes technical parole violations with a few days to four months in a county jail or state prison....

[U]sing federal recidivism data for inmates who left state prisons in 1994, parole violations accounted for the entirety of the gap between California’s recidivism rate and the recidivism rates of other large states. In other words: Because of the differences in how states and localities enforce parole, recidivism rates tell us little about the reoccurrence of the types of crimes with which the public is most concerned: crimes that have a victim.

December 13, 2014 at 11:58 PM | Permalink


TrackBack URL for this entry:

Listed below are links to weblogs that reference "The Misleading Math of ‘Recidivism’":


The sole valid measure of recidivism is to obtain a certificate of absolute immunity from the DOJ, and to interview a sample of criminals about their recent criminal activity.

In this study of 100 sex offenders, they committed 20,000 non-sex crimes in the prior year or 200 major crimes a year. Add their dozens of sex crimes, and you have an idea.


One should calculate the value of not being victimized, such as having a child raped, not having a knife held to one's throat and wallet taken. That would estimate the direct cost. Then add the justice system cost. And do not forget the value of real estate that is dropped if buyers know about the crime.

The question then remains if the $50,000 it costs to incarcerate the offender is not among the biggest bargains and most fruitful of government costs and services.

Posted by: Supremacy Claus | Dec 14, 2014 1:09:49 AM

While I think the overall point of the article is valid this one sentence is frustratingly dumb, "Both the federal and Pew statistics leave out an entire group of former prisoners: those who break the law but don’t get caught." Well, of course, in the same way that crime statistics in general leave out those who break the law and don't get caught. If we exclude the latter, there is no good reason to include the former.

Posted by: Daniel | Dec 14, 2014 2:15:11 AM

Crime statistics have no validity, sae those using the gold standard of household surveys of crime victimizaiton. Otherwise, they are politically motivated lies. Anyone making policy using those crimse statistics is just making up stuff, or believes in Santa Claus, my famous cousin.

There were 23 million crimes in 2013, 6 million being violent. Crimes are the FBI Index felonies. There were 525,000 arrests for violent crime.There were 3 million arrest for drug possession and Prof. Berman will pleased, 1.2 million were for drunk driving.

Posted by: Supremacy Claus | Dec 14, 2014 5:50:06 AM

The lawyer run criminal justice system is in utter failure, answering only 10% of violent crimes. It must be replaced by legally mandated self help. Law abiding citizens should be armed, and should have a duty to kill the violent criminal at the scene, in social self defense. If any citizen fails to reach for a weapon, they should be issued a ticket for $100. The neighbors should be encouraged to kill the illegal aliens that have moved in and are destroying the real estate values and peacefulness of the neighborhood, not just with immunity, but with a duty to do so. Label the gang members, and pay for their killing, $10,000 cash for each, starting at age 14. Instead, the filthy lawyer traitor is prosecuting anyone defending our neighborhoods from these criminal gangs, and protecting them to preserve a high crime rate for a few lousy lawyer government make work jobs. If crime is to ever stop, the public will have to eradicate the lawyer hierarchy, a group of 20,000 traitors and rent seekers.

And when the lawyer dumbass has a guy, 10% of the time he has the wrong guy, an innocent person because he is so stupid, and has very stupid Rules of Evidence, from hundreds of years ago, when people didn't know much. To help the lawyer dumbass, all self dealt immunities must end, and the careless lawyer dumbass must compensate the victims of his tortfeasing, both the crime victim he failed to protect, and to whom he has total duty, and the innocent victim of his false, and idiotic prosecution. To deter.

At the gut level, we know the falling crime rates are false.

Posted by: Supremacy Claus | Dec 14, 2014 9:50:49 AM

Crime statistics don't leave out cases where the offender is not caught, at least for violent/index crimes. They do leave out cases where the victim does not report. But those statistics include a lot of robberies, rapes, murders, thefts, etc. where the offender was never known/caught. I suppose they do tend to omit law-breaking where the offender is never caught when you are dealing with drug possession or dealing, etc. (i.e., the "victimless" crimes).

Posted by: anon | Dec 15, 2014 10:54:06 AM

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