Bryan Caplan  

From the Cutting Room Floor: The Rationality of Juries

What Would Robin Hanson Say?... From the Cutting Room Floor: T...

The earliest draft of "Rational Ignorance vs. Rational Irrationality," (2001. Kyklos 54, pp. 3-26) was unpublishably long.  Here's one of my favorite "deleted scenes":

e.  Juries

Anglo-American rules of evidence [almost] explicitly assume that jurors are not rational.  Judges weigh information's "probative value" against its "prejudicial effect," and only allow a jury to see it if the former outweighs the latter.  But if jurors had rational expectations, seeing graphic photos of the murder victim would not provoke them to convict an innocent person.  Similarly, juries are rarely able to see evidence about a defendant's prior criminal record; the assumption being that juries would systematically over-weight its significance.

A juror's vote, unlike a citizen's vote in an election, often has a high probability of being decisive for the case at hand.  But it is the overall legal climate - not the outcome of one case - that affects a juror's wealth; and as a rule the outcome of one case has no appreciable impact on the overall legal climate.  A juror's private cost of irrationality is therefore as trivial as the voter's; the juror's personal wealth will not vary no matter how ill-founded his opinion is. 

Jurors facing this incentive structure tend to consume a lot of irrational beliefs about the probability that a defendant is guilty.  Nothing deters them from believing that handsome defendants are less likely to be guilty, or that DNA evidence is unreliable.  As a second-best solution, the legal system hides some kinds of evidence from juries when they are likely to respond irrationally to it.

Question: To what extent are the rules of evidence reasonable ways to compensate for juror bias?  To what extent do they instead reflect the irrationality of legislators and judges?

Comments and Sharing

COMMENTS (17 to date)
Les writes:

I do not think the rules of evidence - or anything else - can compensate for juror bias. It seems to me that jurors are asked to decide legal and other issues that are way too complex for laypeople.

I think that juries should be eliminated and replaced by two experienced experts to advise and assist the judge with whatever issues are most important to the case at hand.

ThomasL writes:


I believe that you are joking, but you have done it so well I cannot be sure. Bravo!

Kurbla writes:

I agree that professionals are better choice if advantages (expertize) prevail over disadvantages (corruption.)

Freude Bud writes:

@ Les:

Why do you think that experts are more likely to apprehend the truth of a matter (or thing in itself) than laypeople?

For example--Do you think that a statistically significant higher percentage of financial professionals (or economists) than laypeople (of similar incomes) profited from anticipating today's financial crisis?

I, for one, doubt it.

Les writes:

Dear ThomasL I was not joking. I have testified as an expert witness in well over 100 trials, and many were jury trials.

Dear Freude Bud, I do not believe anyone has a crystal ball, so predictions are inherently suspect. But trials involve analysis of past events, in which I think experts are far superior to laypeople.

Dear Karbla, of course you are right about corruption - but jurors are just as subject to corruption as are experts.

Lise writes:

The only jury I would ever trust to preside over my trial is one composed of Ph.D. level social psychologists who specialized in research on the social psychology of legal issues/trials.

Mercutio.Mont writes:

I was under the impression that DNA evidence is not, in fact, nearly so reliable as prosecutor's typically say. With the advent of DNA databanks, it will become considerably less so.

Prosecutor's fallacy, Bayes' theorem, all that.

Prakhar Goel writes:


Agreed. The jury system while great on paper (judged by your peers) fails in practice: somebody trained in judging would be a much better choice than a group of irrational individuals.


The "not as reliable" for DNA evidence is on the order of 10 million to 1 odds. Normal people have trouble comprehending such odds (and most of the time do not even get a halfway decent intuitive understanding if such odds). For evidence, just look at the number of people who try at the state lotteries even when the expected value is negative. DNA banks may change the odds to 8 million to 1 but those are still astronomical odds which will not get the appropriate weights when considered by the average juror.

Radford Neal writes:

Jurors shouldn't credit odds like 10 million to 1 for DNA evidence, since the probability that the test was incompetently or corruptly performed is far greater than 1 in 10 million.

Freude Bud writes:

@ Les:

No crystal ball, fine. Charles Beard's history of the intentions of the founders of the American Revolution are out of fashion. It has been in fashion and out of fashion before. It will be in fashion again.

I understand your bias (in favor of yourself, of course), but I'm not sure why it is a rational bias.

@ Caplan:

The rules of evidence in use in the US are a result of competition between legal authorities with overlapping jurisdictions in the late Middle Ages. I recommend Harold Berman, Law and Revolution, Harvard, published sometime in the late 1960s early 1970s.

Kurbla writes:

Amateur jurist are less suspectible for corruption because they are new in the system. Beginners tend to follow the rules until they gradually develop methods and personal relations necessary for cheating the system.

Hypothesis: corruption is development of the market economy parasite in the planned economy host.

RobbL writes:

I was on a jury where the admitted evidence was obviously not enough for a "guilty beyond a reasonable doubt" verdict. But the "expert" (the prosecutor) brought the case and the "expert" (the judge) allowed the case to go forward. Fortunately, half the jury saw through these experts and after many days beat down those jurors who accepted guilt based on the "he must be guilty or he wouldn't be on trial" theory.

Sharper writes:

The purpose of juries is not to be better than experts at deciding what happened. That is their secondary role in the system, made necessary because of their primary role.

The primary point of a jury is so that someone isn't convicted of a crime that a dozen regular people don't actually think is a crime, or doesn't actually think they've committed.

If the government were unjustly prosecuting people, or prosecuting under bad laws, where would the incentive be to rock the boat for a couple of government employed "experts" that are "jurying" the case? Especially if the defendant was technically guilty of violating an unjust law?

Millions of people in the former Soviet Union experienced being convicted of nebulous laws with vague requirements for guilt. Some will respond "That can't happen here", because juries haven't had to nullify a bunch of those cases in the U.S., but that's precisely the point. The knowledge that you have to get 12 somewhat regular people to buy into a criminal charge has prevented anyone from seriously trying blatantly bad criminal law since the Star Chamber in England (in which the judge could fine or punish the jury for making the wrong finding).

RL writes:

Without focusing on the details of either juries or charitable giving, I think it was a mistake to leave this material out of your book.

You developed a novel, original theory that expands economic thinking in the area of voting. It is controversial. Showing that the theory explains phenomena that economists study in a wide variety of areas strengthens the legitimacy of the theory. The cutting room floor, we now see, discussed two such areas, areas that one wouldn't have expected ahead of time a theory designed to explain voting behavior to have much to say about. I think it a shame to have left them out of the book.


Chuck N writes:

Prakhar: You write, "The 'not as reliable' for DNA evidence is on the order of 10 million to 1 odds. "

Radford rebutted that it's much less dramatic than that since there's a decent chance of sloppy or corrupt lab work.

He's right of course, and there's also the huge other problem that Mercutio alluded to: The Prosecutor's Fallacy. There's good information on-line about this, but the short version is that if you have a database of 20 million DNA samples and you run your test sample through it, then (using the 10 million to 1 number) you're probably going to get a hit even if the perp isn't in the database.

Now it's a different story if the state scooped up the obvious suspect and his DNA happened to match. But that's not always how these investigations work.

Experts have the uncanny ability to advise based on who's paying them. Experts for the FBI, various coroner offices, police, etc are all routinely making up evidence (see Rodney Balko for more: Ask a government expert on the stand what a drug courier profile looks like and its going to be whatever the defendent was wearing, doing and saying. That's a bias with 1 result.

I would prefer to be tried by the first 12 names in the phonebook (and their biases) than be judged by all the experts on government payroll.

David Fish writes:

It is my belief that the system anticipates both rationality AND bias from jurists and this design is not durable due to the accuracy it provides, rather the fact that the decisions are accepted and trusted by "their peers."

I am thinking of the principle in negotiation of getting to agreement (verdict) with which both side will abide.

Comments for this entry have been closed
Return to top