The Death Penalty Deterrence Myth: No Solid Evidence That Killing Stops The Killing

Posted June 18, 2007

Among the many factors in the debate about the death penalty is whether capital punishment deters violent crime. Although solid research indicates that there is no valid evidence of such deterrence, recent attention has been given to a few flawed studies concluding that the death penalty does deter murder.

A June 10 Associated Press article pointed to statistical studies that claimed to directly link numbers of executions with numbers of murders prevented, including a 2003 study from the University of Colorado at Denver and studies from 2003 and 2006 by researchers at Emory University. But follow-up studies by top social scientists soundly reject those conclusions as well as the flawed methodology used to reach them. Jeffrey Fagan, a professor at Columbia Law School and an expert on statistics, testified to Congress that the Emory and Denver studies were "fraught with numerous technical and conceptual errors," and "fail[ed] to reach the demanding standards of social science."

The truth is that it might be impossible to determine a true statistical relationship between homicides and executions because the number of executions is so small compared to the number of homicides. But what we can say with certainty is that there is no legitimate statistical evidence of deterrence.

John Donohue, Yale Law School professor and Research Associate at the National Bureau of Economic Research, and Justin Wolfers, Wharton School of Business professor and Research Affiliate at the NBER, analyzed the same data used in the Emory and Denver studies, as well as other studies by the same researchers and many other nationwide reports. They found that if anything, executions increase homicides, concluding: "The view that the death penalty deters is still the product of belief, not evidence ... On balance, the evidence suggests that the death penalty may increase the murder rate."

Donohue and Wolfers analyzed data from the 2006 study by the Emory researchers using non-death penalty states as a control group, a basic statistical tool used to study causation not used in the Emory study. When they compared death penalty states with non-death penalty states, they found no evidence of any effect of executions on murder rates, either up or down. Donohue and Wolfers also analyzed the data from the 2003 Emory study that concluded that each execution prevented 18 murders and found that the reduction or increase in murders was actually more dependent on other factors used in the study than whether or not the states had the death penalty. For example, when Donohue and Wolfers slightly redefined just one of the factors included by the Emory researchers, they found that each execution caused 18 murders.
Donohue and Wolfers also recomputed data from the Denver study of select states to account for overall crime trends, a factor not included in the Denver study, and reached inconclusive results. For two states included in the Denver study that had abolished the death penalty, Massachusetts and Rhode Island, Donohue and Wolfers found that the homicides rates actually fell after capital punishment was ended.

Other studies also refute the deterrence theory. For example, researchers Lawrence Katz, Steven Levitte and Ellen Shustorovich analyzed state data between 1950 and 1990 and did not find a correlation between the death penalty and crime rates. Moreover, one of the Emory researchers, Joanna Shepherd, published a state study of her own and found that while the death penalty deterred murder in six states, it actually increased murder in 13 states, and had no effect on the murder rate in eight states.

Other statistical analyses show that states with the death penalty do not have the lowest murder rates in the country. In fact, according to the Death Penalty Information Center, states without the death penalty have consistently lower murder rates than states with the death penalty, even when comparing neighboring states. In addition, while southern states account for over 80 percent of the executions in this country, they have consistently had the highest murder rate of the nation's four regions.

Comparing American and Canadian statistics is also telling. While Canada has not had a single execution since 1972 and the United States has executed over 1,000 people in that time, the homicide rates in the United States and Canada have closely tracked each other. If anything, Canada's experience suggests that ending executions leads to a drop in the murder rate.
As the death penalty debate continues, it will inevitably be filled with the emotion and passion that have historically and rightly characterized it. But when it comes to analyzing data and reaching statistical conclusions that are used to affect our nation's policy and legislation on a matter as dire as capital punishment, it is critical that the research use statistically valid methodology. When we come across studies that are as specific as to tie a number of executions to a number of prevented murders, a healthy skepticism is in order, especially in the face of substantial countervailing evidence. This is, after all, a matter of life and death.