Dylann Roof appears via video link at a bond hearing at a North Charleston, S.C., court on June 19, 2015. Roof was sentenced to death for killing nine African Americans on June 17, 2015, at a historically black church in Charleston. (Grace Beahm/European Pressphoto Agency)

A little more than two weeks ago, Dylann Roof was sentenced to death for a racially motivated church slaughter that shocked the nation. The sentence came after a year in which the United States hosted just 20 executions, a dramatic drop from the 98 carried out in 1999. Death sentences also dropped dramatically, to just 30, the lowest number in modern history, down from 315 in 1996.

Those death sentences and executions are extremely concentrated in just a few places. A recent two-part report from the Harvard Law School Fair Punishment Project shows that some district attorneys seek death at extraordinary rates. This confirms a previous report showing that just 2 percent of all U.S. counties account for more than 50 percent of the death sentences and executions.

Those executions are not mathematically correlated to the number of homicides in that jurisdiction. Nor are they correlated with local public opinion support for the death penalty.

The maps below show where you can find concentrations of executions and of homicides. Homicides are measured using the most complete county-level database available from the Bureau of Justice Statistics, which covers the period of 1984 through 2012. Executions include each one from 1977 through 2015. States that did not have the death penalty over the entire period are shaded in dark gray. Small and large black dots distinguish those counties with low and medium levels, and the highest values in each map are marked by proportionately sized red circles. (For executions, blank means no executions at all; for homicides, blank indicates fewer than 100 homicides over the entire period.) Executions are listed by the county of the crime which led to the execution; each state that carries out executions has a central execution chamber generally within a state prison. The two maps are formatted identically to facilitate comparison.



Executions by county, 1977 to 2015

Homicides by county, 1984 to 2012

A glance reveals little connection between the two. Executions are concentrated in Texas and Oklahoma; homicides are in Los Angeles, Chicago, and the Philadelphia to New York corridor. The only city high on the homicide list in a non-death state is Detroit. (Click here to see two other maps, showing homicides per capita, and executions per homicide.) Even huge numbers of homicides may lead to very few executions, as New Orleans, Chicago, Philadelphia and Los Angeles show.

If we look at states rather than counties, we see a very slight correlation between homicides and executions, but many states with very high numbers of homicides have few executions: California, Illinois and Pennsylvania all have more than 20,000 homicides in the period but no more than about a dozen executions. Oklahoma, Virginia and Missouri are three of the top five execution states, but they rank relatively low on the homicide list.


We exclude California and Texas from this figure because they are such outliers that they skew the rest of the graph: California has about 80,000 homicides but just 13 executions, so it appears far out to the lower-right. Texas has more than 500 executions and about 50,000 homicides, so it is at the far upper-right. (Click here to see the figure with those two states included.)

The high-execution states, at the top of the graph, are Texas (not shown), Oklahoma, Virginia, Florida and Missouri. But they are not all toward the right; in fact, they cover almost the full range in terms of homicides. Only a few small states have fewer homicides than Oklahoma, but Oklahoma follows only Texas in executions. Do executions keep the murder rate low? The National Academy of Sciences 2012 review of deterrence suggests that a more likely explanation is just random chance. For instance, Florida has high executions but high homicide numbers, as the graph shows.

What if we look at executions as a proportion of homicides, and homicides as a proportion of population? Are the dangerous states more likely to execute? The figure below shows they are not. But neither does it show the opposite; it shows no correlation whatsoever.


Executions don’t track local public opinion about the death penalty

Neither are executions correlated with local public support for the death penalty.

With the cooperation of the Gallup organization, we compiled data from their national survey from 1978 through 2015. This large compilation of data enables us to estimate public opinion for almost every state, over time. Opinion has shifted dramatically over the years.

The figure below shows the number of executions compared to state-level public opinion, leaving out those states that did not have the death penalty during most of this period, and leaving out states in which we had fewer than 100 respondents. (Click here to see a table listing our results state by state, including the number of respondents.)


You can see that the number of executions in a given state is unrelated to its residents’ level of support for capital punishment. Virginia, the state with the third most executions during this time, has the fifth lowest level of public support for the death penalty among the states listed. Texas, which has nearly five times as many executions as any other state, is closer to the middle of public opinion: 17 states have more public support for the death penalty than Texas, and 17 have less. Among the big users of the death penalty — Virginia, Georgia, Alabama, Texas Florida, Ohio, Missouri and Oklahoma — public opinion ranges from the lowest quartile to the highest. The overall correlation, 0.07, shows no relation between opinion and use of the death penalty.

That’s even more striking if you look at Texas in more detail. Harris County has executed more individuals than any state other than the rest of Texas. But the Houston Area Survey reveals that residents in and around Harris County actually don’t support the death penalty as much as the rest of the state.


Over 17 Texas polls, support averages out to 76 percent. Out of nine Houston Area Surveys, the average is 69 percent support.

Why is there no or very little statistical connection between homicides, or public opinion, and executions? First, a very small number of district attorneys in just a few areas of the country aggressively pursue the death penalty. A few are in states where appellate courts later allow the executions to go forward. Some are not.

Second, when a district attorney or U.S. attorney is going to seek the death penalty in a case, citizens who oppose the death penalty are excluded from the jury. Those who do serve tend to be more likely to convict and to impose that sentence.

Whether a particular murder is punished with death is related, in other words, to the preferences of the local district attorney.

Frank R. Baumgartner, who holds the Richard J. Richardson Professorship in Political Science at UNC-Chapel Hill, is the author most recently, with several student co-authors, of “Deadly Justice: A Statistical Portrait of the Death Penalty” (forthcoming, Oxford University Press, 2017). This article draws substantially from the new book.

Arvind Krishnamurthy is a senior at the University of North Carolina at Chapel Hill and is applying to PhD programs in political science. He is co-author of “Deadly Justice.”

Emily Williams is a senior at the University of North Carolina at Chapel Hill.