February 26, 2009
Great new (though still dated) examination of the death penalty and plea bargaining
Regular readers know that I often note the failure of others to note and consider the potential impact of the death penalty on plea practices. Excitingly, Kent Scheidegger of the Criminal Justice Legal Foundation has now focused attention on this issue through this new working paper, titled "The Death Penalty and Plea Bargaining to Life Sentences." Here is the paper's abstract:
This study examines the disposition of murder cases in a sample of large urban counties to determine if there is a connection between the availability of the death penalty and the number of cases that are disposed of by guilty plea with a life sentence or a long term of years. Consistently with expectations, significantly more defendants plea bargain to a life or long sentence in states where the death penalty is available. The average county with the death penalty disposes of 18.9% of murder cases with a plea and a long sentence, compared to 5.0% in counties without the death penalty. The difference is statistically significant at the p<.05 level. Implications for the claim that repeal of the death penalty will save substantial public funds in trials are discussed.
As the last line of this abstract highlights, one interest aspect of this paper is the suggestion that the death penalty may savemoney by prompting more pleas and avoiding costly trials. The CJLF's official press release about this report (available here) stresses this point:
Legislatures expecting a large savings in trial costs from repealing the death penalty may be in for a disappointment, according to a study released today by the Sacramento-based Criminal Justice Legal Foundation. The most widely cited estimates ignore or minimize an important cost-saving effect of having the death penalty available.
In states where the death penalty is the maximum punishment, a larger number of murder defendants are willing to plead guilty and receive a life sentence. The greater cost of trials where the prosecution does seek the death penalty is offset, at least in part, by the savings from avoiding trial altogether in cases where the defendant pleads guilty. Although this effect is well known to people working in the field, there appears to be no prior study to determine the actual size of this effect.
Any and everyone seriously interested in the realstory of the death penalty and its impact can and should read this new CJLF report. Unfortunately, as the report itself acknowledges, the data used for the analysis in the CJLF report are for 1988 murders and prosecutions. There are reasons to suspect (and hope? or fear?) that plea practices are much different in capital cases two decades later. Because of this, we all can and should be wary about using long-ago past performance to predict future realities.
Some related posts on the death penalty and plea bargaining:
Some recent related posts about death penalty costs:
- States considering laying off the death penalty during tough economic times
- The economic case against the death penalty getting more and more attention
- More discussion of cost concerns in debates over the death penalty
- Death penalty abolition becoming more realistic in a few western states
- Interesting time for Time's discussion of death's demise
- DPIC releases year-end report on state of death penalty in 2008
- What might 2009 have in store for . . . the death penalty in the US?
February 26, 2009 at 10:15 AM | Permalink
TrackBack URL for this entry:
Listed below are links to weblogs that reference Great new (though still dated) examination of the death penalty and plea bargaining:
Stands to reason. Criminal has an incentive to plead to LWOP if death is on table. Once LWOP is max punishment, then a lot less pleas for LWOP.
The burden, of course, should be on those advocating abolition, since they're the ones who are claiming such cost savings.
Posted by: federalist | Feb 26, 2009 12:04:20 PM
Why don't you just go ahead and barter directly with the devil while you're at it, Kent.
Posted by: Samuel | Feb 26, 2009 12:10:08 PM
Samuel, I think such comments can be turned around against you. Why do you want to put prison guards, other prisoners and witnesses on the outside at risk because you don't want the death penalty? Why do you want to put upward pressure on the murder rate because of lessened deterrence?
Posted by: federalist | Feb 26, 2009 12:18:15 PM
I would gladly abolish plea bargaining altogether, Samuel, if that could actually be done. Alas, every attempt has failed.
Posted by: Kent Scheidegger | Feb 26, 2009 12:25:00 PM
I can see that minds work in different ways. Upon hearing of evidence that murder defendants are approximately four times more likely to waive their trial rights and plead in death-penalty jurisdictions, my first thought was not of the "[i]mplications for the claim that repeal of the death penalty will save substantial public funds in trials."
Rather, my thoughts jumped to the possible implications for the claim that the death penalty reduces the accuracy of criminal adjudication by giving the government near-irresistable leverage in high-stakes cases. Such leverage would seem to place in an untenable and Kafkaesque position defendants charged with a heinous crime who are in fact innocent (or only partially culpable), but, for one reason or another (lack of alibi/witnesses, credibility problems), will have a hard time proving it. (Particularly considering that any trial will be before a death-qualified jury likely predisposed to convict.) I don't know if the data reviewed in this study tend to confirm that supposition, but I would sure like to know.
(My second thought was also not about the cost-savings of repealing the death penalty, but about what percentage of the plea cases studied involved the plea-taker providing testimony against another person in a capital case, and the "implications" for the reliability of that testimony.)
I'm not suggesting Kent doesn't also recognize these accuracy and leverage concerns (his desire to abolish plea bargaining in general suggests as much), and I hope to review his full paper when I have time. However, these were my initial thoughts based on the abstract, as well as Doug's post, which seem to spin the paper strongly on the cost-savings angle (very timely, as recent news reports appear to show this angle actually making headway with legislatures around the country) at the expense of these other possible concerns.
Disclosure: This snap reaction is colored by my general view that---leaving aside whether the cost-savings abolition angle is a red herring, or political cover---the real problems, and the real unjustifiable costs, of the DP have to do with its distorting effects on the entire system, including the system's ability to reliably determine the truth in difficult cases.
Posted by: Observer | Feb 26, 2009 1:05:18 PM
Of course, Observer, we have the other issue, that plea deals for LWOP remove the possibility that the guilty get away with it or get a lower sentence.
Also, Observer, have you considered that the existence of the DP enhances the accuracy of the system as a whole? No matter what one thinks of the death penalty, it has brought a ton of scrutiny, and I think that scrutiny has helped the justice system.
Posted by: federalist | Feb 26, 2009 1:15:36 PM
oh noes! if we get rid of the death penalty we'd have to prove up a few more of our murder cases.
Posted by: 4:42 | Feb 26, 2009 4:42:46 PM
The idea behind the paper is clever, and the intuition makes sense. But the method is too weak to bear any weight at all.
That the results in New York and Texas differ should come as no surprise, regardless of whether there is a death penalty in one and not the other. With no other controls, and with only a single cross-section of data from a small number of counties, I just don't think you can draw any inferences from the observed correlations.
The author himself acknowledges this on the one hand, but then denies it on the other. On p. 12 he admits to the limits in the data--limits that to me preclude the ability to draw any conclusions. But on p. 13 he continues: "Notwithstanding these limitations, it is probable that the death penalty is an important factor in the differences observed here." I think this is a jump he simply can't make. You simply can't assume away weak data. Given the weaknesses, we have no reason to assume that the correlation would survive a more rigorous model. No reason necessarily to assume that it would not, either, but that's the whole point: we simply don't know, and the "probably" waves away significant epistemic problems. (The "important" is even more troubling, since it gives even more weight to an unreliable result.)
Again, the cost-benefit point is an interesting one; my opposition to the paper does not come any normative axe to grind about the DP. But I am always concerned when I see statistics used to make arguments they simply can't support, and I think this paper makes that mistake.
Posted by: Overburdened Statistics | Feb 26, 2009 5:43:40 PM
Of course, in Texas, the DP is an effective threat.
Bottom line, those who are promising cost savings are the ones who should have to disprove what seems obvious to those who know how the world works. A person facing LWOP or a max term of years sentence has nothing to lose by refusing to plead guilty. With the death penalty out there, the calculus for the accused changes. I don't think it's fair to say that Kent is assuming away weak data, I think he's saying that the conclusion stands to reason and there's some empirical support for it.
Put it like this, if you're a defense attorney when would you ever advise a client to plead guilty to an LWOP offense if LWOP is all you can get?
Posted by: federalist | Feb 26, 2009 5:49:57 PM
Speaking of plea bargains . . . .
This is an obscenity.
Posted by: federalist | Feb 26, 2009 6:18:04 PM
Federalist, when it comes to statistics, statements like "those who know how the world works" veers dangerously close to Ronald Dworkin's idea of an interpretative fact, a concept that is an anathema to all empiricists of any stripe. An "interpretive fact," according to Dworkin, is one that is so obvious that it need not be proven empirically--it simply must be so. To which any empiricist has an immediate response: (1) if it is so obvious, then it should not be hard to find convincing evidence, and (2) if you cannot find the evidence, maybe it is not so obviously so.
Same here. True, surely there exists at least one offender who accepts a plea more readily than he otherwise would because of the DP. But that is not the issue. The issue is whether the magnitude of that effect is sufficient to offset the other costs of the death penalty. Phrased that way, your burden-of-proof argument loses its meaning. Who has the burden for establishing the magnitude of an effect? Not at all clear.
And that brings us back to the weak evidence. I don't think Kent is aiming to make the claim that there is *some* effect of the DP on pleas. I think everyone in the world would agree with that basic point. He's trying to estimate the magnitude of the cost-benefit effect. And I don't think he has the data to do it. Though I used to feel this way in graduate school, I no longer believe that we should simply do the best we can with the data we have. Sometimes we have to admit that the data are too weak to use; running sophisticated tests on weak data can give weak results a veneer of precision they do not deserve. And unsophisticated consumers of those results cling to the numbers, not the uncertainty about them. Once Kent admitted to the limitations in the model, I think the paper was done--to draw conclusions after that is to simply overlook the full implications of the empirical shortcomings inherent in the data.
Posted by: Overburdened Statistics | Feb 26, 2009 11:12:09 PM
And what is your opinion, Overburdened, of the multiple studies on the cost of the death penalty that implicitly assume the magnitude of the effect is zero?
Because this cost is being touted as the basis of a major change in policy to be made *right now*, I think it is essential to make the best estimate we can with the data we have available.
If someone can find the better data needed to make the more rigorous analysis, great. In the meantime, we need to point out that the assumption that the effect is zero is almost certainly wrong, and it is quite possible that the effect is large enough to wipe out the savings that are being promised.
Posted by: Kent Scheidegger | Feb 28, 2009 8:53:42 AM