Listen to this story
The advent of computers and the internet—and the declining costs of using both—spurred law enforcement to construct massive databases of information about criminal offenders. Two prominent examples are DNA databases and sex offender registries. Both have been widely used since the 1990s, but they have different features that result in wildly different impacts on crime. DNA databases reduce recidivism and crime rates. Sex offender registries have no meaningful impact on either — and might even increase recidivism by registered offenders. Why the difference, and what does this tell us about how to use technology effectively to fight crime?
DNA databases aim to match known offenders with crime scene evidence. If a convicted offender or arrestee must provide a DNA sample — which depends on state law — authorities use a simple saliva swab to collect a sample, which is analyzed in a lab to generate a DNA profile (basically an identifying string of numbers akin to a Social Security number). Once uploaded to the database, an individual’s DNA profile is frequently compared with DNA profiles from crime scene evidence across the country, looking for matches that could point authorities to suspects in unsolved cases. As of February 2017, there were 12.8 million convicted offender profiles in CODIS — the national network of state DNA databases — and an additional 2.6 million profiles from arrestees.
Whether an individual is added to the database is known only to the individual and law enforcement; the information is not made public. But it appears that’s enough: Individuals know that once their profile is in the database, they’re much more likely to get caught if they commit another crime, and this knowledge deters them from reoffending.
In research published earlier this year, I compare individuals scheduled for release from prison just before and after a database expansion to test the effect of the DNA requirement on subsequent behavior. Those released before are the control group; those released after are the treatment group. These groups are nearly identical except that individuals in the treatment group are added to the DNA database and so are now more likely to be caught if they reoffend. I find that DNA profiling reduces recidivism by at least 17 percent for violent felony convicts and 6 percent for property felony convicts. Using the timing of laws that expand state DNA databases as a natural experiment, I then show that these effects on behavior result in aggregate benefits to public safety. Adding more people to DNA databases reduces crime — violent crimes like murder and rape as well as property crimes like larceny and motor vehicle theft.
Contrast this with sex offender registries. These registries require anyone convicted of a sex-related crime to keep their contact information (including name and address) updated with local authorities. The resulting registries are publicly available, so anyone can see if a registered sex offender lives nearby. In many places, communities are actively notified if a sex offender moves into their neighborhood. As of 2016, there were more than 800,000 registered sex offenders nationwide.
Sex offender registries have two purposes: 1) to help local law enforcement keep tabs on known sex offenders, and 2) to provide information to community members that may help them protect themselves and their children from sexual predators. Both could reduce recidivism by sex offenders and make communities safer. But they don’t.
Two national studies — one by Amanda Agan and another by J.J. Prescott and Jonah Rockoff — used the timing of registry laws to estimate the effects of registries on local crime rates. They found that sex offender registries had no effect on sexual offenses. There is some evidence that the existence of sex offender registries deters sex crimes by people not in the registry, perhaps because the prospect of being added to the registry is viewed as an additional punishment. But there is also some evidence that registries increase recidivism by offenders who are already in the registry. The stigma they face may make crime more appealing, because registered offenders — now less able to find jobs, housing, or social support — have less to lose if they reoffend.
Several other studies have found similar results across an array of states and law changes: There is no empirical evidence that sex offender registries reduce recidivism by registrants or improve public safety. But they do impose severe costs on individuals who are required to register: Many have lost their jobs or homes and struggle from depression due to social isolation. (See this review for more empirical evidence on the effects of sex offender registries.)
A primary difference between these databases is whether the contents are public or private. Providing information about sex offenders’ whereabouts to the community could make local parents feel in control but appears to do little more than scare people (one study finds that home values respond dramatically to news that a sex offender has moved nearby — to a degree that seems completely unjustified by the change in crime risk). Contrary to popular opinion, more information is not always helpful.
Another difference is the target population we’re trying to affect: Sex offense convicts in the case of sex offender registries versus a wide range of convicted felons and arrestees in the case of DNA databases. Recidivism rates for sexual assault convicts are extremely low — only 5.6 percent were rearrested for the same type of crime within five years. (This same-crime recidivism rate has barely changed since 1994, so it’s not due to the registries.) It’s tough to reduce that further, and the low recidivism rate implies that the vast majority of convicted sex offenders aren’t much of a threat. In contrast, same-crime recidivism is 34.4 percent for assault convicts and 41.4 percent for larceny convicts. There’s a lot of room to reduce those rates and to prevent current criminals from escalating to more violent behavior. That’s what DNA databases appear to do.
As we continue to deploy technology in the criminal justice space, it is important to consider whether a particular tool is appropriate in a given context and what the likely impacts are on everyone involved. How do we expect people to respond to new information or incentives that the tools put in place? Are there alternative policies that could achieve the same goals with fewer unintended costs? Since it’s impossible to predict all of a policy’s effects, it’s crucial that we follow the evidence on which tools are worth continued investment and which we should scrap. Research has shown that DNA databases are a useful and cost-effective crime-fighting tool. Sex offender registries, on the other hand, should be abandoned.