The Impact of Jury Race in Criminal Trials

In a great paper, The Impact of Jury Race in Criminal Trials, Shamena Anwar, Patrick Bayer and Randi Hjalmarsson exploit random variation in the jury pool to estimate the effect of race on criminal trials. The authors have data from nearly 800 trials in two Florida counties. On any given day, a jury pool is randomly drawn from a master list based on driver’s licenses. On some days, the pool of about 30 people contains some black members and on other days, purely for random reasons, it does not. The voir dire process–>For every $1 spent on legal aid, the savings can range from $1.60 to $30.removals, excuses and challenges–whittles down the jury pool to 6 jury members with typically 1 alternate.

The authors have data on the race, gender, and age of each member of the jury pool as well as each member of the ultimate jury. The authors also know the race and gender of the defendant and the charges. What the authors discover is that all white juries are 16% more likely to convict black defendants than white defendants but the presence of just a single black person in the jury pool equalizes conviction rates by race. The effect is large and remarkably it occurs even when the black person is not picked for the jury. The latter may not seem possible but the authors develop an elegant model of voir dire that shows how using up a veto on a black member of the pool shifts the characteristics of remaining pool members from which the lawyers must pick; that is, a diverse jury pool can make for a more “ideologically” balanced jury even when the jury is not racially balanced.

The author’s results show not only that blacks and whites are treated differently depending on the composition of the jury pool but also that random variation in the jury pool adds to the variability of sentences holding race constant. Like is not treated as like. The results also suggest that we don’t need racial quotas to increase fairness. We can increase fairness and reduce variability in a racially neutrally way by expanding the size of juries. Six-person juries have become common because they are cheap(er) but a return to twelve person juries would reduce the variability of sentences and greatly equalize conviction rates across race.


"What the authors discover is that all white juries are 16% more likely to convict black defendants than white defendants but the presence of just a single black person in the jury pool equalizes conviction rates by race."

Please include explanations for this phenomenon that don't involve white juror racism below.

Oh what the heck, I'll get this started, using the study's own discussion of distribution tails.

Let's say you're District Attorney Goldlock and because of 'the rules of the game', instead of trying to find six bowls of porridge (out of 20 or so) to eat that are 'just right' you are really just trying to make sure that you don't get any bowls that are 'too hot'. You let Defense Counsel Blacklock - who just loves boiling hot porridge and is happy if there is just one hot bowl, but has sensitive teeth and would prefer if there were fewer cold ones, worry about getting rid of the frosty bowls.

Now, you have two sets of bowls, each set containing a lot of variance within itself, but also with each set having very different mean temperatures. You have green bowls with a mean of 'just right', but naturally with some hot and cold bowls too. You also have red bowls which everyone knows tend to be much hotter on average. You don't have thermometers, but you also have some other crude ways to investigate porridge and guess at the temperature of any individual bowl.

The rules of the game give you 'peremptory challenges' to exclude bowls you think may be too hot, but ever since Batson v. Kentucky you can't use those exclusions when, if challenged, you've got no other compelling reason besides your knowledge that the bowl is red and that red bowls tend to be hotter on average.

So, the only red bowls you can exclude are the red bowls that are so hot that they literally have something obviously hot about them - like billows of steam visibly rising out of porridge. "Your honor! Just look at the steam! It has nothing to do with the bowl being red at all."

Ah, but the rules of the game also only give you a limited number of challenges. So either you've got to live with the chance that your non-excludable red bowl is a hot one, or if you exclude a particularly egregiously hot red bowl, you've got less challenges to use against the green bowls that you expect are hot too.

Which means, if you start with all green bowls in your bowl pool, you'll get rid of the hottest ones, and no one will challenge you on it. But, it you start with even one red bowl in your pool, because of the 'rules of the game', then the probability you'll end up with a hot bowl go up by the basic rules of probability.

Of course, this is just a fanciful discussion of hypothetical statistics and, of course, has nothing to do with the realities of the American criminal justice system, as you can immediately verify by asking any prosecutor ... in private ... after a few drinks.

I thought about this too, but my concern was the lower conviction rate for the theoretical greener pool. For this explanation to work would there need to be a selection bias on the part prosecutes in which cases they bring to trial?

The study looks at felonies. But, these days, a felony is not a felony is not a felony. And there is a red-green bias is which felonies reds and greens commit, which are likely to be contested in court instead of resolving in a plea bargain, how many resources prosecutors put into a trial due to their perception of priority and seriousness, and also how successful prosecutors are likely to be at trial in a particular action, due in part to issues of evidence and the burden of proof.

For example, murder charges will get top priority, a lot of attention, are unlikely to end in a plea deal because of high sentences and a defendant's perception he's going nothing to lose, but they also tend to be easy to prove if you've got some good, hard forensic evidence.

Sexual assault charges, on the other hand, are also unlikely to end in a plea deal, but they are much harder to prove especially when the element in question is consent - a state of mind - and prosecutors will occasionally feel like they have to show they care and go through the motions of a prosecution, but they don't devote priority resources to a likely loser of a case, and indeed the conviction rates are much lower.

I hadnt thought of a crime mix difference. That makes sense.


Yeah, it's the kind of obvious thing that a decent study would try to correct for.

But the market in Social Science research for the kind for ideologically convenient confirmation bias that won't hold you to strict scrutiny when the results come out nice is vast and lucrative in the currency of that realm.

Some might say that is the market.

There is a lot of truth to this.

About 47% of second degree murder cases are resolved by plea bargain. 41% of first degree murder cases.


Please do away with the porridge analogies.

You are forgetting RANDOM assignment in the jury pool!!!

Sounds like a lot of hassle just to have some breakfast.

Why not face the obvious head on? Floridians are somewhat racist, on average, and this impacts outcomes when juries are involved.

Why look any further?

Or are you waiting for someone to say "they deserve it because they are inferior and inherently more violence/criminal beings" or some other hogwash, but in some less obvious way.

We are talking about percentages of population subgroups who actually go to trial. This has nothing to do with the fact that black people are far more often arrested for committing a crime (whether because police preferentially give attention to black people and black neigbourhoods, or whether as a result of historical explanations which rightfully make them angry but which more realistically negatively impact them through disintegrated communities and lack of family and community role models, or more obvious factors such as less educated populations in communities with very low quality and poorly funded schools being more likely to turn to crime as a means to earn a living) and therefore are more likely to go to prison. This like of thinking is false. They compare outcomes of accused persons who have gone to trial.

How likely is a child to go to prison if their father went to prison? Why have 60% of black men in the USA born in early 1970s done prison time? Perhaps they were rightfully pissed off, engaged in civil disobedience, then got a record after the police harassed the bleep out of them endlessly and one day they FINALLY snapped?

"Why not face the obvious head on? Floridians are somewhat racist, on average, and this impacts outcomes when juries are involved. Why look any further?"

Because all white juries convict at the same rate as integrated juries. It's only the integration of the pool that makes any difference. You wouldn't expect that result if the explanation were merely "Floridians are somewhat racist, on average."

All white jury POOLS convict blacks 16% more.

An all white jury from a jury pool with one black does not convict blacks more often than white.

And people named Nathan are idiots who make sweeping generalizations without supporting evidence or even substantial anecdotal observation.

I made a sweeping generalization on the basis of an aggregated result reported in a scientific study.

It's more like I repeated their conclusions, but went to far as to speak the obvious. Underlying racism, while far less than it was 100 years ago, is still present.

If the paper is not the proof, then what explains this difference?

It is also obvious that the earth is flat and that the sun orbits the earth.

Why have 60% of black men in the USA born in early 1970s done prison time? Perhaps they were rightfully pissed off, engaged in civil disobedience, then got a record after the police harassed the bleep out of them endlessly and one day they FINALLY snapped?

People born in the early 1970's would have been part of the first cohort born in those wonderful public housing projects that "Urban Renewal" gave us. Say what you want about them, but I don't think anyone in their right minds would characterize these places as hotbeds of "civil disobedience." Probably more just regular disobedience.

The state has a 15% African American population, Lake County is less than 10% and Sarasota less than 5%. How are either representative of the state as a whole? Did this trio put this level of statistical work into getting their PHD's? That sure doesn't say much for standards at Duke, Carnegie Melon or the University of London.

historical explanations probably (path dependency) since blacks in Barbados do well, but..

factors such as less educated populations in communities with very low quality and poorly funded schools being more likely to turn to crime as a means to earn a living)

If you look at the evidence I believe that you will find that blacks do not do well in well funded schools either, anyway it is easy to test.

How likely is a child to go to prison if their father went to prison? Why have 60% of black men in the USA born in early 1970s done prison time?

Holy mackerel! That makes me think that Government cannot even do their most basic function well! Reform can start by trying ending he war on drugs!

that all white juries are 16% more likely to convict black defendants than white defendants

Can they possibly control for the quality of representation of either side or the propensity of the prosecutor to bring cases? Keep in mind, full dress petit jury trials are odd (IIRC, the share of convictions achieved that way in New York is in the low single digits). Even minor differences across the color bar in the vectors which induce a petit jury trial may make for systemic differences in the sort of cases brought.


You could make the alternate argument -- that trials to go to a jury are more likely to feature an actual innocent defendant who doesn't want to take the plea -- but without controlling for degrees of guilt the study is crap.

You are forgetting that the jury pool is random with respect to trials.

Yes. They also can't control or havent controlled for actual guilt or the severity of the alleged crimes or prior convictions.

Also note that 16% is 1 in 6. I don't think this is a coincidence.

+1. The last large meta analysis of racial effects in sentencing found that studies that whilst there is an overall small but statistically meaningful racial bias in sentencing studies that use sophisticated controls of past offender behaviour indicate that there is almost no sentencing bias with regards on both guilt/innocence and length of sentence.

You are arguing that the black person's lawyer skill is like a quantum cat, unknown until the jury pool is revealed, and if the jury POOL is all white, the lawyer box is opened and discovered one-third of the time to be incompetent, but if just one black is in the jury POOL the box is not opened, so the lawyer is just 50-50 incompetent relative to the prosecutor incompetence.

But it might be the prosecutor is 50-50 incompetent unless a black is in the jury POOL and in selecting the jury, the prosecutor quantum box is opened and revealed to be incompetent two-thirds of the time.

Perhaps the bias of the prosecutor is revealed to the judge during jury selection when just one black is in the POOL, so the judge offsets the bias he sees in the prosecutor statements in what he allows or instructs the jury.

What is needed is all and overwhelmingly black jury POOLS in sufficient frequency to get statistical results from all black juries from POOLs with no whites and with just one white.

"greatly equalize conviction rates across race"

The goal should not be to equalize conviction rates by race but to convict the guilty and acquit the innocent. Tabarrok and the study authors do not even try to show that the racial makeup of the jury is related to the frequency of wrong verdicts. Maybe one way to assess the quality of juries is to look at how often their verdicts are reversed on appeal. Has this been studied?

The IQ of whites is one standard deviation above that of blacks, and other things being equal, you want smarter jurors.

Yes, but also elimination of bias could be a more pressing goal than convicting the guilty. So, even if the 16% acquitted with more black representation are guilty (and we don't have any way of knownng), it might be worth it to legitimize the process.

Legitimacy? Please.

There are two approaches to legitimacy in the criminal justice system.

1. Minimization of Type I and Type II errors, but some surface disparity in ethnic representation in the results.
2. The racial headcounts come out right, but at the cost of plenty of error.

The preference of some people for the second model of 'legitimacy' over the first is exactly the problem.

The premise is 99 guilty go free instead of 1 innocent be convicted.

You seem to be arguing that it should be 1 innocent person be executed so only one out of 99 guilty murderers goes free.

No doubt Texas has executed a lot of clearly guilty murderers, but the execution of at least one person was based on your criteria held by a fire chief who believed that a father had to be held responsible for the death of his two children no matter the evidence, and Gov Perry agreed that this belief MUST NOT BE QUESTIONED so he shutdown a formal reopening of the conviction that led to the execution - no way was Perry going to allow science interfere with justice in Texas because science just let's too many people police and prosecutors are convinced are guilty go free.

Racial "justice" > regular old justice.

If there is a better example of why progressives should be hung before they wreck our society, please speak now.

It depends on whether the errors are false-guilty or false-not-guilty.

But the fundamental theme of our judicial system is "better 100 guilty men should go free than to punish one innocent man".

The point is we don't know. We just know they are wrong in some direction.

Good idea, but convictions can be reversed for a variety of reasons unrelated to guilt.

All else being equal. I doubt very much that this is the case.

You can train for an IQ test. Like, if you go to schools with better funding and better teachers probably you do better on standardized tests.

Anyways, I think we should consider culling all those > IQ of 130. Before we get too smart for our own good and destroy everything.

But before we do that, I have one last contribution:

Love your neighbour. In a globalized world we are all neighbours. What kind of world do you want to live in?

Perhaps poor performance by students with elevated skin melatonin reflects their disinterest in a test which has no practical relevance to anything in their life, whereas self-interested braggarts who are children of racist parents will prepare heavily for such tests in order to parade around their supposed superiority after acing a test that has little or no practical relevance to anything.

I would be more inclined to judge people on the basis of how they act in real situations. But the situations will not be comparable until there is equal opportunity, which there is not.

Asians do better on IQ tests than whites. At which point in time you point to education, hard work, and such factors, no?

> reduce the variability of sentences and greatly equalize conviction rates across race.

I'm not clear why this is a correct outcome.

Some people live in rough, dysfunctional neighborhoods were the police are overwhelmed and are not effective in maintaining order. Others live in peaceful, orderly neighborhoods where the police are an effective presence. This difference can be very large.

A burglary in a rough neighborhood is often simply ignored by the police. One that rises in severity sufficient to attract their attention police is likely to be a markedly different sort of crime than a burglary in a otherwise peaceful neighborhood, but this difference is not captured in the data. I would expect the correct sentences to be more severe in the first case.

Furthermore, when the police are overwhelmed, arrests for less-severe crimes are more likely to be made only when the case is obvious and can be resolved with minimal investigation. The more ambiguous crimes are ignored because they do not result in arrests. Again, this does not appear in the data.

I'd suggested that they are not measuring racial differences here, but simply the socio-economic differences for which race is a convenient proxy. We know that crime, and policing, are both very different in these two worlds. We should not assume that the conviction rates and sentencing severity should be the same across them.

Just wondering how the authors know that the black defendants were not guilty 16% more often?

Well, we all know the white defendants are 16% more innocent, right?

After all, we are talking about verdicts, not arrests that do not even lead to charges being pressed. Here is an article about manipulating crime statistics - '“I hope you’re happy with yourself,” one officer told me. And I responded: “This story’s going to get out there. It’s going to be on the front page of The Washington Post tomorrow.”

And he said, “Yeah, well, you’re going to be in my jail cell tonight.”

Once at the station, we were processed, our pockets emptied. No mug shots. They removed our restraints and put us in a holding cell. Ryan was able to get ahold of his dad. I called my mom, but I couldn’t get through. I couldn’t remember any phone numbers.

We were in there for what felt like 10 or 15 minutes. Then the processing officer came in.

“Who’s media?” he asked.

We said we were. And the officer said we were both free to go. We asked to speak to a commanding officer. We asked to see an arrest report. No report, the officer told us, and no, they wouldn’t provide any names.

I asked if there would ever be a report. He came back with a case number and said a report would be available in a week or two.

“The chief thought he was doing you two a favor,” he said.'

Do note that part about “The chief thought he was doing you two a favor.”

"Well, we all know the white defendants are 16% more innocent, right?"

I wonder if there actually is some sort of regression to the mean dynamic in these group differences in sentencing. I'm not talking only about the black-white difference, but also the fact that women are treated more leniently.

Sentencing is the last step - the problem lies in who actually gets arrested, who gets their arrest recorded, who gets charged, etc.

Is there a discrepancy between the frequencies of perpetrators from each race in victim surveys versus in arrests and convictions? I think not.

>Well, we all know the white defendants are 16% more innocent, right?<

No, we don't. And I am not convinced the authors did, either. That was my point.

Do they claim to know?

I might assume the other way. I might assume a black juror is biased toward acquittal based on numerous reasons with a range of defensibility.

That explains why the model works how it does. If you have a locked-in acquittal vote, then the prosecutor is compelled to use your veto on that person (did anyone just ask the prosecutors?).

That, in statistics akin to Bayesian updating, changes the "expected value" of the remaining jury pool.

It could be a pretty interesting paradox. I'm a black juror and I know that when the jury is all white the conviction rate is much higher. I assume that whites are biased. So, I decide to always vote for acquittal to balance this. Doing so increases the delta between all-white and mixed juries.

"If you have a locked-in acquittal vote, then the prosecutor is compelled to use your veto on that person (did anyone just ask the prosecutors?). That, in statistics akin to Bayesian updating, changes the “expected value” of the remaining jury pool."

Bingo. It's all about the peremptory challenges game post Batson

"I’m a black juror and I know that when the jury is all white the conviction rate is much higher."

Except that's not true, which is what the study shows.

All white juries convict at the same rate as integrated juries so long as there is an integrated pool. It's only the pool that matters. If a black person ends up on a jury, then the pool was integrated too, and he has no special 'racist statistics' reason to deviate from the just verdict.

Note the comment above - and do note how the Ferguson chief hands out his 'favors,' which just happen to also cover his ass in a way that he would never worry about if the person arrested for sitting in a McDonalds was just another typical citizen of Ferguson.

After all, I'm sure all those 'trespassers' are treated the same, right? - 'During this time, we asked the officers for badge numbers. We asked to speak to a supervising officer. We asked why we were being detained. We were told: trespassing in a McDonald’s.

“I hope you’re happy with yourself,” one officer told me. And I responded: “This story’s going to get out there. It’s going to be on the front page of The Washington Post tomorrow.”

And he said, “Yeah, well, you’re going to be in my jail cell tonight.”'

Which, as it turns out, was another one of those stunningly accurate predictions concerning guilt and innocence - and how that turns out when it comes to having an arrest record, a mugshot made, or charges pressed.

Fairness isn't a worthwhile concern on its own?

True innocent rates are inherently unknowable, else we wouldn't need juries. However, there are other sources of bias one can test for. If prosecutors are more likely to bring a felony to trial given the same allegations, and evidence, for one group than another then we would expect different conviction rates. Another is crime mix differences, where felony A is easier to prove than felony B given the same burden of proof. If white defendants are more likely to find themselves in he said/she said cases (for instance), and black defendants are more likely to find themselves in fingerprints-on-the-gun cases then you would also expect different conviction rates. Note - neither of these hypos depend on overall crime rates or rates of innocence.

The assumption that two groups are similar unless specifically proven otherwise. It's a fairly common assumption in statistics. You could hardly do a basic t-test without such assumptions.

The probability that the person making such a statement would hire a black person who was identical to some second individual with the exception of the quantity of melatonin in their skin? Less than 16%. I'd bet on it, and I only have a sample of one.

The fact that the statistical comparison is between a jury pool and a jury almost certainly indicates that other variables are confounding the results. Despite the authors "elegant" theory of voir dire, I can't think any plausible explanation about how composition of a jury pool would "shift the characteristics" of the selected jury.

Are you kidding? If you have a prosecution friendly jury pool, you'll get a prosecution friendly jury. Each side gets to pick and eliminate people during vior dire, but they have to work with the jury pool they draw.

You did read the post, yes? The contention is that eliminating the one black member of a pool shifts the characteristics of the jury, all else being equal. The comparison is between two sets of pools, not between two sets of juries of between the pool and the jury.

Cause isn't quite accurate.

The actual theory is that the presence of a black is a proxy for high defense tendency of the pool on average.

We can't be sure exactly what the interpretation means.

I had interpreted this as meaning that the fact of eliminating a black member from the pool may lead them to be more explicitly aware of racism, and thus to convict equally between black and white defendants in such a situation.

Challenging the black person allows more bleeding heart whites to be selected.

But this effect has to be less subtle. The presence of this extraordinary juror type has the effect of flipping the other five, a la Jimmy Stewart, or Jack Lemon if you prefer.

There was a study that showed people's political views became more polarized in a group of like minded people, but the mere presence of one person of differing views moderated the group consensus. Does anybody remember that paper?

Do you need unanimous vote to equit ?

I wonder what would happen if you polled all people who use phrases like "bleeding heart white" with the following questions, what would the results be.

Children born to two professional parents which are similarly connected, politically and socially, will have similar outcomes regardless of skin tone. A) agree. B) disagree, the black children will do better on average because they are inherently more apt. C) disagree, the black children will do better on average because the social and political system is skewed in their favour, D) disagree, the white children will do better on average because the social and political system is skewed in their favour, E) disagree, the white children will do better on average because the social and political system is skewed in their favour.

Sorry, there would have been too many clauses if I started to talk about Latinos.

To the best of my knowledge, having your father carted off to prison, growing up in communities with poorly funded schools, being harassed endlessly by police who eventually manage to nab you or at least one of your friends for things which go unpunished in "white" communities, all contribute negatively to outcomes.

Why not report that juries with a somewhat higher average level of education tend to convict more often, or higher average IQ, or higher average dental care, or higher average income, or lower average fast-twitch muscles, or whatever else might (perhaps) differ between the two sets? Are they really arguing that "data on the race, gender, and age of each member of the jury pool" encompasses all human difference? If so they are ageist, sexist and racist.

'If so they are ageist, sexist and racist.'

You aren't American, are you?

I almost have you pegged. It is coming to me. It is going to be in the form of a character or politician and I feel it is going to be on the nose.

Good luck - then we can enjoy having the mystery of the identity of Andrew' and andrew', and whether they are the same person, resolved.

Yeah, the one is when I'm on my phone. I don't care about your identity. I am getting a sense of your personality.

'I am getting a sense of your personality.'

Na, dann viel Erfolg, Andrew'.

Don't know what that means. Don't really care.

'The results also suggest that we don’t need racial quotas to increase fairness.'

I can tell you weren't born in the Commonwealth of Virginia in 1963. If by 'quota' you mean any attempt to change jury selection in any fashion that did not result in a 100% white jury.

Here would be one interesting example of something that would likely look like a racial quota in practice - in order to to improve fairness, in all cases involving execution as a sought for punishment, only justified reasons are allowed in voir dire, thus eliminating peremptory challenges. Assuming a random jury pool that reflects the community, it would likely look like a racial quota in practice.

And not being a lawyer, it just might be possible among the states that still practice capital punishment this is actually current practice. However, a citation to the actual law would be required as proof in the case of any sourthern state.

100% white juries acquit blacks when they are innocent. Im not oblivious to the effect of Juror 13, but I think his impact is overstated. Prosecutors don't usually bring losers to court. This, in general, keeps conviction rates high, and if a group is arrested more for crime because of actual guilt, poor concealment, or disparate policing, conviction rates will tend to reflect that fact alone.

'100% white juries acquit blacks when they are innocent.'

Sure - but how often do they convict blacks when they are innocent? Which does start to stray into the area of prosecutorial misconduct, but why wonder about the numbers of innocent people who were on death row when data is actually available.

What is intereting about the list here - - is that it seems, very roughly, about half and half when it comes to whites and blacks being exonerated. But then, I sure that many of those white defendants faced all black juries, right?

The Innocence Project is nonsense. Most if not all of these people were not 'exonerated.' Rather, they convinced a judge, years after trial, that there were reasonable doubts of guilt. Much of that evidence would have been indamissible at the original trial and is often unrebutted.

There isnt necessarily prosecutor misconduct. Many of these cases hinge on eyewitness testimony which is, in many cases, faulty if not unreliable. It is very hard to discredit the testimony of a rape victim who picks the assailant out of a lineup. This doesnt make the prosecutor a bigot, just a believer. Contrary to what you think, we do refuse to file charges when the evidence is absent or lacking in credibility. The beneficiaries of that discretion are majority black.

My impression the last time I was on jury duty was there was a bias in the make up of the jury pool depending on the day of the week. The original draw for jury duty was random but they allowed for people to request to be on call. So the pool of potential jurors on Monday were the people who self selected to not be on call. When they called in additional jurors on Thursday, the jury pool on that day was people who self selected to be on call. So the case that went to jury selection on Monday versus the case that went to jury selection on Thursday did not have random jury pools even though the original draw was random. I would be very surprised if judges/lawyers haven't noticed this and schedule cases with this in mind.

In my case, I was in a Thursday pool. The pull from the waiting room came in the afternoon. The judge paraded the potential jurors in front of the defendant, did all the introduction stuff, and then scheduled jury selection for Friday morning. The defendant plead the case out the next morning before selection began. My impression was the judge specifically timed events to get a plea bargain.

An intelligent observation, but I think you give us too much credit for tactical considerations. By the time a criminal case goes to trial, there are very few surprises left. The remaining variables often involve credibility issues and the strength of circumstantial evidence. I really didnt have time or willingness to play games with the day of jury selection. Our resources are spread too thinly. Of course, there can be tacit incentives that drive your observation, if it is more generally true.

Shouldn't we assume, at a first pass, that any kind of racial differentiation of the guilt or ease of proving guilt of the defendants is largely compensated for by the prosecutor, or indeed overcompensated for? Why would the prosecutors bring MORE chancy/bad cases against white people than they do against black people? Prosecutors presumably know about the quality of representation/ease of proving the crime/etc. issues before it comes to a jury trial.

I mean, I could potentially see that in certain very atypical jurisdictions -- prosecutors playing a political game to a black base with, "See, we try to hold white folks accountable," but that seems unlikely to explain the effect on a state-wide basis, and indeed probably more than compensated for by the reverse effect in other jurisdictions.

For these explanations to hold true, you either have to assume that prosecutors are racist to such a degree that they're willing to sacrifice the metrics that they are judged on in order to either screw over innocent white defendants or release guilty black defendants, or are so incompetent that they don't know the (widely commented upon) effect that white jurors are less likely to convict white defendants, and use that in their calculations of bringing a case to trial. That's quite a claim. Does anyone have any actual evidence that prosecutors do this?

It seems far more likely that white defendants brought to trial are MORE guilty than black defendants. After all, they have better representation etc., and also if your jury pool doesn't include any black people, they're more likely to get off. A rational prosecutor would, ceteris paribus, bring fewer white defendants to trial. So we should presume that ceteris are not in fact paribus and when the prosecutor does bring a white defendant to trial, on average the case is stronger.

"white defendants ... if your jury pool doesn’t include any black people, they’re more likely to get off"

How does that work? What sort of feelings or thoughts in the jury members would bring that about?

I could think of several basically plausible ones. I have no idea which one or ones are common or dominant in the real world (if, indeed, this study is correct -- while I think it sounds plausible, one study does not tell us everything we need to know about the world).

Your question sounds to me like it has an agenda. Perhaps you should just make that agenda text.

Im skeptical of the result, but taking it as true arguendo, the dynamics of an acquittal involve all the jurors being convinced that a critical piece of evidence is in doubt. Logically, the more similar are the jurors, the more quickly they will converge on agreement. Their agreement might be wrong, but agreement nonetheless.

Juries will naturally have some diversity - intelligent types, informal leaders, compliant people, and the truly clueless. If there are one or more holdouts, remember that it will likely result in a mistrial. Depending on how the vote was split, it will end up dismissed, pled, or retried. The first two dont see a jury.

I really dont think very often white defendants have a sole black juror holding up an acquittal. If there is a racial component, it acts far more subtly than that. It relates to the entire jury's willingness to hold doubts about one or more pieces of evidence.

Im also not sure how this study deals with multiple charges, some of which result in conviction (perhaps a lesser included offense) and others for acquittal.

Youre asking good questions here. Of course we dont have racist motivations, at least not in recent decades. If there is endemic racial bias, though, a white prosecutor and a white jury might all believe in the guilt of the accused but wouldnt believe it if the same person were white. It is hard if not impossible to prove this because all cases are different.

If blacks are wrongfully convicted, Id say that faulty eyewitness testimony is to blame in most cases. They might actually believe they are picking the right person from the lineup. This is also, not surprisingly, where good defense attorneys will try to raise doubts.

Blacks commit more crime than whites, not merely proportionately but absolutely. This will make it seem more likely that a black defendant is guilty and/or a white defendant is less likely guilty. It does come down to the evidence. If we are in trial, we have the evidence. The question is whether reasonable doubts can be raised.

You are making a plausible case about incentives. We do look at our 'batting averages.' I never cared about mine. But just because it is plausible doesnt mean it holds sway. Prosecutors are evidence driven, and incentives probably only hold 'as if' they are maximizing their stats.

An unrecorded stat is how many cases we throw away. Some I just didnt take to the grand jury.

In my opinion, the Zimmerman trial was a throw away. The DA had to show that he was sensitive to the racial issues, but he never thought he could win. At least two of the prosecution witnesses might as well have been testifying for Zimmerman's lawyer.

The Mehserle verdict was similar - every litigator in the country knew he'd get convicted for IM and nothing more. If it werent racially charged, it would have been pled to that at the start.

Witness intimidation operates through social networks.

As a Floridian, somebody familiar with the working of the criminal courts in Florida, and as somebody who will be serving in a jury pool tomorrow, I'll try to explain what's happening.

Florida has a 'sentencing guidelines,' scheme that contrains the sentencing judge to submit probably too harsh sentences for offenders. In practice, this puts all of the power in the hands of the prosecutor, who can offer plea deals to lesser offenses. It is really out of wack: 10-20 years in a prison (where a man will be raped) versus 3-6 months in jail (hanging out with drunk drivers).

Assume three hypothetical defendants:

1. Is Guilty, knows he will likely be convicted
2. Is perhaps not guilty, has to rely on public defender
3. Is not guilty, has the resources to mount a full defense.

In practice, the way the state of Florida has allowed prosecutors to game the sentencing guidelines means that defendant 1, but also defendant 2 will probably not go to trial. This despite the fact that defendant 2 may not be guilty. Defendants in a borderline case aren't going to risk years or decades on their overburdened wage slave prosecutor. *That sentences are so dramatically worse for people who exercise their right to a day in court ought to be a civil liberties issue*

In addition to the above, Flordia also has a variety of 'three strikes laws.'. Offenders with three convictions get sentenced to long sentences, and the discretion is taken out of the hands of the judge and mostly out of the hands of the prosecutor. Such defendants usually go to trial, because they have nothing to lose, even though such trials are usually open and shut slam dunk guilty verdicts.

In a criminal system where an overwhelming majority of cases do not go to trial, the cases that do go to trial are the exceptions. The cases that do not go into trial are often cases where the defendant:

1. Is stubborn, or doesn't understand the prosecutors' plea arrangements
2. May not be guilty and has the resources to mount a defense, or
3. Is looking at a 'three strikes' related charge.

If you assume that 'stubbornness' is equally distributed among the races, then it is clear to see how #2 above might skew white, and how #3 above may skew black. **Stubborness actually coo relates with high testosterone and as such does have a measurable, testable skew**. This should more than account for the researchers dispassionate skew of jury results.

Tl;DR. Not all trials are equal,

If it is confusing as to why there might be both a different outcome and a different jury in a case where:

1. Defendant may not be guilty and has resources to defend himself, versus 2. Defendant probably is guilty but tries the case anyways with a public defnder because of three strike laws,

then Consider things might make a difference. The expertise of the lawyer matters. The lawyers experience picking juries matters. Resources matter (I saw a juror get kicked without using a peremptory challenge, because the lawyer's team was able to find an outsanding arrest warrant on the juror).

Let's clear some things up:

About 85% of indictments yield a guilty plea.

The only cases that go to trial are generally those with evidentiary issues.

When a case goes to trial, prosecutors win between 4:1 and 5:1.

However, more charges are dismissed than prosecutors win, so a good defense can win before a jury is seated.

The outcomes by private attorneys are only slightly better than public defenders. I would say, without doing the math, that it is not statistically significant, but then McCloskey reminds me that if you have the money, it might be worth a 0.6% chance to avoid doing time.

Jury trials are an aberration, and looking only at those will tend to magnify perceptions about the benefits of particular types of jurors, skill at voir dire, witness credibility, forensic evidence, and uncertainties about circumstantial evidence. Juror #13 factors play an infinitessimal role in our justice system, maybe slightly larger than whether the defense attorney stayed at a Holiday Inn Express or La Quinta last night.

The whole topic is a charming irrelevancy as the vast majority of all convictions do not derive from a jury trial. At the Federal level 98% of all cases are pled because no rationale human being would risk both bankruptcy and the obscenely disproportionate sentence they would get if they lost. Indeed the minor variability discovered by the authors is a quaint artifact from a bygone era when we had a criminal justice system rather than today's apparatchik administered, chock full of American goodness style Gulag.

I want to end the drug war. What else should I want? Even if we all elected reform politicians, or those giving lip service to reform, could/would they do anything? Is there some other leverage point? If 98% are plea deals could these go to trial cheaply on the defense end to clog the system if defenders could somehow coordinate?

How does the state even allow an all white jury when a black man is on trial?

Has anyone does a study of how all-black or predominantly-black juries deal with white defendants? Considering the relative proportions of the races, the sample size might not be large, but it still might be possible. Perhaps they could go to a state like Mississippi, where a third of the population is black.

Comments for this post are closed