« Are most 1/3 downward variances reasonable, while most larger ones aren't? | Main | More dialoging about Danforth »

November 2, 2007

A potent (and pernicious?) new claim in the death penalty deterrence debate

Thanks to this post at C&C, I see that Roy Adler and Michael Summers of Pepperdine University have this Wall Street Journal commentary discussing a potent new deterrence claim.  The piece is entitled "Capital Punishment Works," and here are excerpts:

Most commentators who oppose capital punishment assert that an execution has no deterrent effect on future crimes. Recent evidence, however, suggests that the death penalty, when carried out, has an enormous deterrent effect on the number of murders.  More precisely, our recent research shows that each execution carried out is correlated with about 74 fewer murders the following year.

For any society concerned about human life, that type of evidence is something that should be taken very seriously....  The conclusion that each execution carried out is associated with the saving of dozens of innocent lives creates an extraordinarily difficult moral dilemma for those who campaign against the death penalty.  Until now, those activists could look into the eyes of a convicted killer, hear his or her sad story, work tirelessly to set aside the execution and, with that goal accomplished, feel good about themselves for having "saved a life."  These data suggest that the moral equation is not nearly that simplistic.

It now seems that the proper question to ask goes far beyond the obvious one of "do we save the life of this convicted criminal?" The more proper question seems to be "do we save this particular life, at a cost of the lives of dozens of future murder victims?"  That is a much more difficult moral dilemma, which deserves wide discussion in a free society.

The remarkable assertion that each execution could be saving 74 lives creates a stunning new math as the current de facto moratorium unfolds.  Specifically, since cert was granted in Baze, roughly 15 executions have been stayed.  Are the authors of this commentary prepared to assert that the Supreme Court's handiwork may be costing the lives of over 1000 future murder victims?  Indeed, given that de facto moratorium seems likely to delay or prevent many dozens of executions, are the authors of this commentary prepared to suggest that the moratorium could end up costing more lives than the 9/11 tragedies?

As detailed here, I tend to be agnostic about the deterrence debate. But the suggestion in this commentary that every execution could be saving 74 lives seems bold to the point of recklessness.  (Before this commentary, the boldest assertion I saw in the literature was that each execution might save 18 lives.)   I don't know much about Professors Adler and Summers, but I would like to hear more about their study and how far they might be willing to push their conclusions.

Some related death penalty deterrence posts:

November 2, 2007 at 03:54 PM | Permalink


TrackBack URL for this entry:

Listed below are links to weblogs that reference A potent (and pernicious?) new claim in the death penalty deterrence debate:


As far as I can tell, they're claiming that each execution saves 74 lives. They're not making any claim about the effect of temporary stays of execution. Obviously, if their findings are true, you would expect each stay to produce a marginal (if immeasurable) in increase in the murder rate.

But the truth of these assertions is truly doubtful. The folks who commit death-eligible crimes are not the same folks who are watching scotusblog every day to parse the latest orders of the Court. I'm trying to imagine an interview with one of these depraved murderers:

"Well, I was not planning to kill the guy. But then I checked on scotusblog, and I saw that no one's being executed these days. I figured, 'Life without parole ain't so bad.' Lethal injection really sacres me, but prison is fine. So I went ahead and slit the guy's throat."


Posted by: Marc Shepherd | Nov 2, 2007 4:15:16 PM

Does death penalty work?

In 2002 Americans were very happy because they had only 16.638 criminal homicides: and they were right because from 1984 to 1993 criminal homicides were 22.000 per year. Au contraire, in the same 2002, in Italy we were very afraid because, with a population that is grosso modo one fifth of the American one, we had 638 criminal homicides, and we were very concerned about it, even if those 638 were less than one third the homicides we had in 1991. Americans love to think the drop is a benefit of the death penalty. We cannot agree because we are a death penalty free country. (In Europe this punishment is strictly forbidden and the majority of the world is abolitionist).
Actually Italy ended capital punishment in 1888 and had it again only under fascism. In those sad years the homicide rate was five times bigger that we have now, and, in the twenty years following the definitive end of the death penalty (1948-1968), the homicide rate dropped from 5 to 1,4. Something like this happened in Canada in the years that followed the end of capital punishment in 1976. Curiously, in the same year, the Supreme Court gave green light to the “new and improved” American death penalty and, with the shooting of Gary Gilmore (17th January 1977): the hangman was back in business and the experiment begun. Now, after more than 1.000 human sacrifices, we can say with Justice Blackmun: “the death penalty experiment has failed”.
Death penalty is an enormous waste of lives, money, time and resources. This cancer is destroying the American justice. It is not a deterrent and kills the poor, the weak, the mad, the illiterate, and the black. In the thousand killed some were innocent, many mad and much many not guilty of a capital crime, but quite all will be alive, and some free, if they have had a competent counsel. Hangman states are not in a better situation of states without death penalty. Sooner or later Americans will realise that death penalty is and immoral, indecent, illegal, expensive, stupid, cruel, dangerous, racist, classist, not working violation of human rights.
Best regards.

Dott. Claudio Giusti
Via Don Minzoni 40, 47100 Forlì, Italia
Tel. 39/0543/401562 39/340/4872522
e-mail giusticlaudio@aliceposta.it
Claudio Giusti had the privilege and the honour to participate in the first congress of the Italian section of Amnesty International; later he was one of the founders of the World Coalition Against The Death Penalty.

Posted by: claudio giusti | Nov 2, 2007 4:23:08 PM

May I just point out that Roy Adler is a professor of marketing? He also has some great photos on his personal website: http://www.royadler.com/photogallery.htm

Posted by: Well | Nov 2, 2007 4:34:34 PM

a month or two ago Kent used a phrase that
I think is apt here. "correlation does
not necessarily mean causation."


Posted by: bruce cunningham | Nov 2, 2007 4:44:53 PM

Maybe Murdock hired Adler to market the Wall Street Journal. It was only a matter of time before the WSJ was as bad as FoxNews.

Posted by: George | Nov 2, 2007 4:50:33 PM

In response to Claudio Giusti,

In America we have a culture of violence and murder. Sadly, thanks to this culture we were able to expand out territory and become the world's most powerful nation.
I don't believe that the people's view of the death penalty will change anytime soon, nor do I believe that if there is a halt in executions, it will last enough for the effects to begin to notice. Republicans don't seem to have any regard for human life, they are willing to use the state to kill and put away non-violent people for stupid offences, and the Democrats are not much better, because they are afraid of being portrayed as weak on crime. America already has 1/4 of the world prison population, and so far every person i've spoken to believes that the death penalty saves them money, because they wont have to support these people for the rest of their lives.
It's sad that people dont bother to see the statistics, and just believe the government proganda to continue to murder people. As for this study, I dont find it very credible. I did a speech on the death penalty (I'm a college student, so it was to my class.) and I found many more reliable studies that say the exact opposite.

Posted by: EJ | Nov 2, 2007 5:10:39 PM

Looking just at the WSJ article and not the underlying research, the argument seems to suffer from the flaws associated with making bold statistical observations from small numbers without parsing critical distinctions.

E.g., since 26 of 42 executions this year took place in Texas, our murder rate should by far be lowest of any state by this analysis, but that's not remotely true. We have a relatively high murder rate compared to states that have no death penalty at all. In fact, the states with the highest murder rates (in the South) are also the places where the death penalty is most frequently carried out.

I too would like to see the underlying research. If it's accurate, then it would mean Texas' death penalty prevented nearly 2,000 murders in 2007, or about as many as have died over the same period in the Mexican drug cartel wars. Could that possibly be true? Really?

I do not believe there's that much latent, unrequited violence lurking in the hearts of my fellow Texans, and find these conclusions highly suspect.

Posted by: Gritsforbreakfast | Nov 2, 2007 5:37:24 PM

In response to what I think are misleading statements in the article:
1) anybody have the data on murders before, during, and after 1972-76 period? (I know somebody does...) That would quite obviously be extremely important in evaluating these claims.

2) In re: "Occam's Razor" - it doesn't work that way in social stats. The statistical version of occam's razor is Bayes' theory, and there's no shortage of methods to use in conjunction w/ it to make claims about whether something is relevant. Even if his claim held as true in stats as it did in the natural sciences or religion (god is, after all, the simplest explanation by many criteria!), it's not clear to me that his position (deterrence works) is the most direct explanation given what Marc and many others have observed about crimes of passion and the existence of other credible arguments (e.g. that the sudden decrease in murders in the 90's is due to the long term impact of Roe v. Wade - likely murderers from bad families just weren't born).

3. This is particularly bad: (quoting)
Die-hard campaigners against capital punishment could argue that there might be yet a third variable, such as a stronger police presence or a population shift to urban areas, related to each of the other two variables. Such a variable might exist, but until it can be identified, Occam's razor suggests the simplest solution is probably the actual solution. We know that, for whatever reason, there is a simple but dramatic relationship between the number of executions carried out and a corresponding reduction in the number of murders.

There MIGHT be a third variable? Can anyone claim that exectutions are the only thing that affects murder rates and keep a straight face? (I admit I don't know what he means by 'related to both variables;' given the article's complete failure to treat any other variables at all, I think it's setting up a straw man as part of the distracting argument re: causation).

The study doesn't seem to be publicly available, so I can't really tell, but it seems like they've done a univariate regression - or bivariate, if a lag variable was used, but that's hardly better b/c it does not truly accound for causation from different sources.

If I only have to use one variable to explain behavior, I can prove a HELL of a lot of things that are coincidence. The various r measurements of correlation mean little except in a sophisticated multivariate model, and there's no evidence here that they've developed one.

And I think it's telling that they spent most of the article dealing with the non-issue of causation. Who's arguing that fewer murders lead to more executions?

And where do they get off suggesting they are the first researchers to do a study like this, and the issue is now closed?! Infuriatingly irresponsible pseudoscience.

Posted by: Gray Proctor | Nov 2, 2007 5:57:08 PM

I represent two or three accused murderers each year. I cannot imagine any of these people taking into account whether they are going to face the death penalty or "only" life in prison or 20-something years. I don't believe they are thinking about consequences at all. In my experience this seems to be the hallmark of criminal behavior. And let me comment on the reality of death penalty defense. Do you imagine defendants begging their lawyers to get them a life sentence instead of the death penalty? Many times I have harried a prosecutor into a non-death offer only to face the bigger task of getting a defendant to accept it. Lawyers who actually do death penalty defense are well acquainted with "the Patrick Henry syndrome:" give me liberty or give me death.

Posted by: TJC | Nov 2, 2007 11:03:35 PM

I represent two or three accused murderers each year. I cannot imagine any of these people taking into account whether they are going to face the death penalty or "only" life in prison or 20-something years. I don't believe they are thinking about consequences at all. In my experience this seems to be the hallmark of criminal behavior. And let me comment on the reality of death penalty defense. Do you imagine defendants begging their lawyers to get them a life sentence instead of the death penalty? Many times I have harried a prosecutor into a non-death offer only to face the bigger task of getting a defendant to accept it. Lawyers who actually do death penalty defense are well acquainted with "the Patrick Henry syndrome:" give me liberty or give me death.

Posted by: TJC | Nov 2, 2007 11:04:44 PM

Just as the NY Times can be relied upon to take the liberal position on any issue, regardless of the facts, the WSJ editorial page can be relied on to do the exact opposite. It would appear that they found the deterrence study that makes the most extravagant claims, and put it on their editorial page.

I believe the deterrence effect of the DP is real. And for me, that's reason enough to support the DP. But that doesn't mean that the most extreme study is correct.

Posted by: William Jockusch | Nov 3, 2007 12:13:39 AM

If a person who is considering murder or other capital were to use a logical process they would check to see if there was a death penalty in the state where they planned to commit the murder. The numbers for murders in the figure shown in the WSJ article are for non-negligent homicides for all states including those with no DP. If you plot executions vs murders the resulting curve is multivalued because murders both increased and decreased during the selected time interval while the execution rate increased at an irregular rate.

Not all murders are solved or result in a conviction. The DP can also be the sentence for kidnapping and rape in some states. A DP is not the only possible sentence for non-negligent homicides and the probability that someone on death row will be executed in a particular year is about 2% If the DP is a deterrent then it could apply to all capital crimes in the states with a DP and would not apply to capital crimes in states with no DP.

Posted by: JSN | Nov 3, 2007 7:02:04 AM

There are "lies, damn lies and statistics."
Articles and statistics about the "murder rate" are meaningless unless we factor in who is killing who. I've done capital defense for three decades and over the past few years most cases I've had are drug dealers killing drug dealers motivated by revenge or robbery or turf protection. I suspect the average person envisions a totally different scenario when they hear about the murder rate. I suspect they project themselves into the class of murder victims and think that the number of innocent victims killed by depraved, serial killers has increased. When I started practicing law the weapon of choice was a box cutter. Now it is a 9 mm. and Glocks are tools of the drug dealer's trade.

Posted by: bruce cunningham | Nov 4, 2007 1:13:08 AM

What we have is yet another example of mathamatical modeling to predict an extreme and rare behavior, a method that has been discredited in criminology research for years. Homicides are not calculated. Marketing researchers ought to know that their rational models fail to explain non economic behavior.
Economists and others using a business model are literally the only academics to claim research showing a deterrent effect of extreme punishment. Criminologists, of which I am one, don't buy it.

Posted by: Michael Israel | Nov 4, 2007 10:45:06 AM

Someone asked about how many murders there were in the mid 70s. I found data on this by going to the Fedstats web page. So I made my own page with the number of murders and exectuions over the years. This is at


I teach economics at San Antonio College. I found this blog by googling the title of the article

Posted by: Cyril Morong | Nov 4, 2007 11:19:49 AM

If the death penalty is not a deterrent, is there anything that is a deterrent?

A long prison sentence seems like a deterrent. If the chance of getting caught and serving a long prison sentence were also rising at the same time the number of executions was going up that could account for the decline in murders. Some research suggests that lower unemployment rates mean less crime. That might have been a factor as well. Maybe someone can comment on that.

But I did run the regression that these guys did and got 74. It is, as someone above suggested, a univariate regression. Total number of murders as the dependent variable and number of executions as the independent variable.

Posted by: Cyril Morong | Nov 4, 2007 11:46:18 AM

JSN - did you find the underlying study? Have the authors released it?
Thank you for confirming that the regression was merely murders on executions,Cyril. Did you get the underlying work or just use their numbers?Disheartening to see that such important work was done so amateurly - and it got in the WSJ!

Posted by: Gray Proctor | Nov 4, 2007 7:16:03 PM

I got the numbers myself and ran the regression for the years 1979-2005 (since there was one more year at the Fedstats page, I used it). I think I actually got 73.57. So I did not the underlying work or use their numbers, but the numbers I found must be the ones they used. I posted them at the link I have above. Yes, they need to include other variables.

Posted by: Cyril Morong | Nov 4, 2007 7:43:57 PM

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