In the Spring of 2009, the FBI started warehousing DNA samples of Americans who had been arrested — but not necessarily convicted — of a crime. People declared not guilty of a crime, or who were never even charged with a crime, can have their DNA records expunged from the FBI database if they go to court, obtain a court order, have the court order certified and submit it to the appropriate criminal records laboratory in a state and/or the Federal Bureau of Investigation Laboratory Division in Quantico, with…
… you get the idea. Very few of the DNA records of innocent people are going to be expunged. Most of them are going to continue to aggregate in the records of the FBI. Indeed, since I last checked in November of 2009 the number of arrestee DNA samples (the FBI insists on assuming guilt by calling them “offender profiles”) has shot up from about 6 million to over 9 million.
What’s interesting to me as time passes and arrestee DNA data continues to be gathered by the FBI is that the rate of increase in DNA samples varies quite a lot according to the state where the DNA was gathered: the load of samples from Rhode Island is more than four times now what it was in 2009, while in Utah the number of “offender profiles” has actually declined by a third. Meanwhile, changes in the FBI’s Uniform Crime Reports from 2009 to 2010 (2011 state data isn’t yet available) allow us to see whether the change in DNA gathering in a state is associated with a decline in reported crime in a state. If the FBI DNA databanks are really keeping us safer, then in states where DNA collection has increased we should be seeing a decline in crime as nefarious evildoers are taken off the streets.
The following is a scatterplot of 50 dots for the 50 states, with each state placed on the graph according to two values: in the X-axis is the change in the number of “offender” DNA profiles coming from a state between the fall of 2009 and the fall of 2011, and in the Y-axis is the change in the Uniform Crime Reports Violent Crime Rate from 2009 to 2010:
There’s a very, very, very small rise in crime rate, not the predicted fall in crime rate, for states that collected greater amounts of DNA. The R-squared statistic, which when multiplied by 100 measures the % of variation in crime rate explained by variation in DNA collection, is very low as well: a mere .01 (explaining just 1% of the variation). There appears to be essentially no relationship between changes in DNA collection of a state and changes in the violent crime rate of a state. This may be why FBI statistics on DNA warehousing refer vaguely to the number of “investigations aided” thanks to the scraping together people’s DNA, not to arrests or convictions following DNA matches.
You might say the lack of an impact of the DNA program on crime is because there just hasn’t been enough time for super-duper federal government DNA warehousing to have its full impact on crime. Well, let’s see about that. Wait until this time next year, when we’ll look at the issue again. As of right now, the justifications for mass DNA collections don’t wash.
P.S. For a more systematic look at the connection between crime rates, consider reading Mark Rothstein and Meghan Talbott’s article, “The Expanding Use of DNA in Law Enforcement: What Role for Privacy?,” in the Journal of Law, Medicine & Ethics. Rothstein and Talbott write:
It is also not clear how many of the ‘investigations aided’ actually result in conviction. Based on the claims of increased efficacy in solving crimes, one might expect that the percentage of crimes being solved or ‘cleared’ would have increased as DNA databases expanded. However, while the crime rate has dropped over the past ten years, clearance rates have changed very little during that period. Moreover, the clearance rates for crimes typically associated with the availability of perpetrator DNA, homicide and forcible rape were actually lower in 2004 than in 1995, meaning that law enforcement actually solved fewer of the reported crimes than it did when DNA databases were still in their infancy.