<< News Scan | Main | Prop 34 16% Behind in CBS5/SurveyUSA Poll >>


The Errors of Alarcon & Mitchell, Part 2 -- The Plea Bargain Effect

| 0 Comments
We continue our series on the errors in two articles on the death penalty costs by Judge Arthur Alarcon and his law clerk, Paula Mitchell (mostly the latter, I suspect).  See the intro, part 1, and a post on Mitchell's partisan letter to the editor.

How often do you read in the news that a murderer has pleaded guilty and accepted a life sentence in order to avoid the death penalty?  It seems like it happens every week.  In any event, it is certainly not a rare occurrence.  How often does that happen in states that do not have the death penalty?  Prosecutors in those states tell me anecdotally that it is rare there.

If a murder case ends in a plea bargain, the cost of trial is zero.  The cost of appeals and habeas corpus will probably be zero, as the murderer will generally not want to overturn the favorable deal and risk a more severe sentence.  If he does challenge the bargain, the cost will typically be much less than a challenge to a full trial, as the grounds of challenge are sharply restricted.

Any reasonable, good faith estimate of the net cost of the death penalty would therefore necessarily include an offset for the savings resulting from these plea bargains.  How much offset do Alarcon and Mitchell chalk up for this factor?

Zero.
How do they justify such a glaring omission?  The first article simply made no mention of the plea bargain effect.  It just discussed the costs of death penalty cases, many of which were notorious, complex cases that would have been hugely expensive without the death penalty (the subject of another post), and made no attempt whatever to adjust for the offsetting savings from plea bargains.  See 44 Loyola LA L. Rev. S41, S69-S79.

In the second article, A&M justify the omission by citing a study that they say "support[s] the proposition that the threat of the death penalty does not increase plea bargain rates."  46 Loyola LA L. Rev. S1, S21-S22 (emphasis in original).  They write, "The study showed that, nationally, counties with the death penalty have a higher rate of people pleading guilty to longer sentences but not a higher rate of people pleading guilty."  (Emphasis added.)

I know exactly what that study showed.  It's my study.

In fact, the study found an 11% difference between the overall rate of plea bargains in counties with the death penalty over those without.  I found exactly the opposite of what A&M say I found.

It is true, as my paper acknowledges, that the result does not meet the rule of thumb known by the confusing term of "statistical significance."  But as the paper explains, that simply means that the result is not proved to the 95% standard demanded in contexts where a very high burden of proof is required.  In a debate such as this, where preponderance of the evidence is the question, my result establishes by considerably more than a preponderance that the plea bargain effect does exist.  While there is considerable uncertainty of the magnitude, 11% is the best available estimate.

A&M take a statement that a result is not "statistically significant" and turn it around into a claim that the result has been affirmatively disproved.  (In statistics jargon, this conflates "the null hypothesis is not rejected" with "the null hypothesis has been proved.")  This is completely wrong, but it is a rookie mistake that is often made by people who do not understand statistics.  Is this a good faith mistake on their part?

I don't think so.  The very page of my paper that they cite says, "The result tells us that more likely than not there is a real difference in total plea bargain rates between states with the death penalty and those without it."  It explains briefly that "statistical significance" is only a rule of thumb.  More importantly, the matter is explained in detail in terms understandable to non-statisticians in my letter to the Legislative Analyst's Office on Proposition 34 costs, and we know A&M have that letter because they cite it in footnote 155.

I think we have clear and convincing evidence here that A&M have intentionally misrepresented the findings of my study, and they used that misrepresentation to intentionally inflate their estimate of the cost of the death penalty.

Leave a comment

Monthly Archives