« The latest numbers on executions around the world | Main | State judges in New York asking for greater drug sentencing authority »

March 25, 2009

Yet another look at the death penalty deterrence question

I just saw on SSRN this paper, titled "Another Look at the Deterrent Effect of Death Penalty." Here is the abstract:

There is a question whether the execution rate is appropriate to examine the deterrent effect of death penalty.  Instead of using execution rate, this paper uses dummy variables to categorize states into different groups and to compare the group mean homicide rates.  With U.S. state-level panel data for the period 1995-2006, this paper fails to find a significant homicide-reducing effect of death penalty.

Because I lack a deep understanding of statistical methods and social science techniques, I cannot readily assess whether this new paper is a significant contribution to the long-raging debate over the deterrent impact of the death penalty.  I would be grateful to hear from informed commentors about whether and how this latest paper contributes to what we know (and don't know) about the real-world impact of capital punishment.

March 25, 2009 at 07:44 AM | Permalink


TrackBack URL for this entry:

Listed below are links to weblogs that reference Yet another look at the death penalty deterrence question:


Without reading it, they failed to address the dose-response curve applying to all remedies.


None of these studies has any validity until that is addressed.

About half the violent criminals are murdered or commit suicide. They see their friends come to these ends. They are not deterred. The value of the death penalty is in attrition, getting rid of the other half. The main principal is that the deceased have a low recidivism rate.

Posted by: Supremacy Claus | Mar 25, 2009 1:09:37 PM

It’s an interesting article and an interesting idea, but there’s not really enough detail on what data was used to comment on how useful it is.

I think that the issue of the efficacy or otherwise of capital punishment will probably never be amenable to statistical analysis. Although Supremacy_Claus’ comment is (presumably) tongue in cheek, it highlights some of the problems that arise.

Other obvious problems are trying to disentangle all the other socio-economic factors from the sentencing policy in determining what impacts on a crime rate (let alone the problem of defining a ‘crime rate’). Also, given the low numbers of executions (and relatively low rate of murder), it will be difficult to track what is a genuine cause and effect and what is just a statistical fluctuation. And then the question of whether if someone is executed for two murders, how do you factor that in? And conversely, one murder may have two or more perpetrators.

These are just examples of potential difficulties…

Posted by: Legal Geek | Mar 26, 2009 12:57:19 PM

Here is empirical support for the proper use of the death penalty in One-Two-Three Dead.


Most murderers and manslaughterers committed prior offenses. Had they been executed, a homicide would not have happened.

There are no real specialists in crime. So executing a burglar prevents a murder.

Each arrest stands in for 1000 unreported and unprosecuted crimes. The earlier the execution, the more thousands of crimes it prevents by each criminal.

If the third conviction is mistaken, one has executed a bad guy, and its wrongfulness is of no importance. Today, about 20% of people on death row are likely innocent. This fraction represents the fraction of uncertainty in the beyond a reasonable doubt standard. So the error rate in inherent and permanent under current rules of evidence. Under One-Two-Three Dead, the uncertainty drops to 0.8% likely innocent (20% cubed).

This is not satire, nor irony. This is utilitarianism.

There is a question about this scheme. If one is executing 10,000 people a year, and one has reduced crime to about 10% its rate now, by attrition of criminals only, will this rambunctious country be too pacified and lose its spirit? That is why, there should be pilot experiments in small jurisdictions, with outcome studies, then retesting in larger jurisdictions.

The lawyer loves the criminal, as a source of jobs. The victim may rot. The lawyer hierarchy must be removed by force before the defense of the public from the lawyer vicious client can begin to get tested.

Posted by: Supremacy Claus | Mar 26, 2009 6:37:46 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