The definition of reasonable doubt

[Bojack Horseman]

As we all know, the prosecution in a criminal case is required to prove the case beyond a reasonable doubt in order to secure a conviction. But the next logical question is what reasonable doubt is.

For example, if you take the O.J. Simpson case. I think the most logical conclusion that can be drawn is that O.J. did in fact stab to death Nicole Brown Simpson and Ron Goldman, and three racist cops decided to plant evidence to guarantee a conviction, which obviously ended up setting him free. Now the question is, does the fact that Mark Fuhrman perjured himself about being a racist and the LAPD planting evidence at the scene rise to the level of reasonable doubt to warrant an acquittal?

Another example is Casey Anthony. Casey lied repeatedly about where the child was for 31 days before she finally admitted that the child was missing and called police. The child's skeletonized remains were found in the woods five months later, and eventually Ms. Anthony was charged with murder. The prosecution simply said, "connect the dots." She was irresponsible, someone who should have never had a child, and a great argument for keeping abortion legal. However we have no evidence to directly tie the defendant to the death of the child. A skeleton doesn't tell you cause of death, and they had absolutely nothing to tie the defendant to the actual act that killed the child. Did she kill the child? Probably. But probably isn't good enough.

A final example was Scott Peterson. The prosecution simply showed the jury graphic photos of Laci Peterson's dismembered body and said "He did it!" and Mr. Peterson was convicted and sentenced to death based on nothing but inflamed emotion. There was, again, no evidence to tie the defendant to the murder. Now the prosecution alleged that because Laci's hair was in a rowboat they allege was used to dump the bodies, that proves he was the killer. You and I both know that there are a million plausible explanations for how a hair or two from the victim could end up in a boat stored in the victim's own home. If I'd been in that jury room, the vote would have been 11-1 forever.

[Dereich]

Whole books can and have been written on the definition of reasonable doubt. If it helps, here's the standard instruction on reasonable doubt given to jurors here in Florida:

You must be logged in to read this quote.

[Antonio the Sixth]

The scope of reasonable doubt is directly proportional to the price of your lawyer.

[Del Tachi]

A final example was Scott Peterson. The prosecution simply showed the jury graphic photos of Laci Peterson's dismembered body and said "He did it!" and Mr. Peterson was convicted and sentenced to death based on nothing but inflamed emotion. There was, again, no evidence to tie the defendant to the murder. Now the prosecution alleged that because Laci's hair was in a rowboat they allege was used to dump the bodies, that proves he was the killer. You and I both know that there are a million plausible explanations for how a hair or two from the victim could end up in a boat stored in the victim's own home. If I'd been in that jury room, the vote would have been 11-1 forever.

Eh, Scott Peterson was also discovered to have been having multiple affairs in which he described himself as a "widower" weeks before Laci had disappeared, had searches on his computer for tidal patterns in the San Francisco Bay (where Laci's body was discovered), and was arrested carrying $15,000 in cash, four cell phones, a fake driver's license, and having recently cut/died his hair. 

So yeah he killed her beyond a reasonable doubt.

[Associate Justice PiT]

A final example was Scott Peterson. The prosecution simply showed the jury graphic photos of Laci Peterson's dismembered body and said "He did it!" and Mr. Peterson was convicted and sentenced to death based on nothing but inflamed emotion. There was, again, no evidence to tie the defendant to the murder. Now the prosecution alleged that because Laci's hair was in a rowboat they allege was used to dump the bodies, that proves he was the killer. You and I both know that there are a million plausible explanations for how a hair or two from the victim could end up in a boat stored in the victim's own home. If I'd been in that jury room, the vote would have been 11-1 forever.

Eh, Scott Peterson was also discovered to have been having multiple affairs in which he described himself as a "widower" weeks before Laci had disappeared, had searches on his computer for tidal patterns in the San Francisco Bay (where Laci's body was discovered), and was arrested carrying $15,000 in cash, four cell phones, a fake driver's license, and having recently cut/died his hair. 

So yeah he killed her beyond a reasonable doubt.

     Jurors do not suspend their faculty of reason. Failing Occam's razor in a sufficiently spectacular fashion is good grounds for dismissing a doubt as unreasonable. In Scott Peterson's case, his behavior around the time of Laci Peterson's death was quite damning.

[Torie]

This particular lawyer is so weird, that he thinks of these evidentiary standards in statistical terms, to wit, standard deviations from the mean. So we have:

1. Beyond a reasonable doubt - two SD's from the mean  (95.45% odds)
2. Clear and convincing evidence - one SD from the mean (68.27% odds)
3. Preponderance of the evidence (above 50% odds)
4. Reasonably likely (the grand jury standard which I just learned) - one standard deviation from the mean the other way  (31.73% (1- .6827) odds)

I want the odds damn it! The rest is noise.

[MASHED POTATOES. VOTE!]

Hard question. Usually we see such a term as something self-evident.

[Antonio the Sixth]

This particular lawyer is so weird, that he thinks of these evidentiary standards in statistical terms, to wit, standard deviations from the mean. So we have:

1. Beyond a reasonable doubt - two SD's from the mean  (95.45% odds)
2. Clear and convincing evidence - one SD from the mean (68.27% odds)
3. Preponderance of the evidence (above 50% odds)
4. Reasonably likely (the grand jury standard which I just learned) - one standard deviation from the mean the other way  (31.73% (1- .6827) odds)

I want the odds damn it! The rest is noise.

Quantitativism run amok.

(and this is coming from a quantitativist)

[Torie]

This particular lawyer is so weird, that he thinks of these evidentiary standards in statistical terms, to wit, standard deviations from the mean. So we have:

1. Beyond a reasonable doubt - two SD's from the mean  (95.45% odds)
2. Clear and convincing evidence - one SD from the mean (68.27% odds)
3. Preponderance of the evidence (above 50% odds)
4. Reasonably likely (the grand jury standard which I just learned) - one standard deviation from the mean the other way  (31.73% (1- .6827) odds)

I want the odds damn it! The rest is noise.

Quantitativism run amok.

(and this is coming from a quantitativist)

Why?  I might add that the case law in some instances gets into odds in discussing these standards, and in particular "reasonably likely." One court yammered about a 20% to 40% test, and several opined that it was less than 50%.  I need this kind of guidance myself. The mere words are just too loosey goosey for me.

[Antonio the Sixth]

Well, for starters, the p-value has never been intended as a measure of "reasonableness", and its meaning has been so thoroughly distorted (especially in academia!) that poor Gosset must be rolling in his grave. Besides, it is very simply not applicable in the context of a trial.

[Beet]

It's deliberately ambiguous on purpose, which is the only possible answer.

[muon2]

I would have a hard time using 2 SD (95.45%) as the standard for beyond a reasonable doubt. 95% is the usual definition of the margin of error (1.96 SD) and as we know from polling, actual results routinely fall outside the polls MOE.

In papers submitted in the hard sciences we usually look to a minimum of 3 SD (99.73%) to treat the result as significant. For some instances of discovery of an unpredicted new phenomena, 5 SD (about 1 in 2 million) might be used as confirmation.

I suspect that it comes down to what is one's tolerance for the false conviction rate compared to the failure rate for legitimate prosecutions due to lack of evidence. The current US prison population is just over 2 million, so a 5 SD standard would be expected to lead to only 1 wrongly held prisoner in the entire country. A 3D standard would be expected to lead to about 6,000 wrongly held prisoners throughout the US.

[Torie]

Yes, 95% percent is the usually understood odds for a beyond a reasonable doubt, but I view that as just rounding. The masses won't be comfortable with decimal points. And given that we have odds for that, why not for the other standards? As I said, some courts talk about odds. I really needed to know, given that I am on a grand jury. I live by the odds. I'm a Bayesian. I find it works for me, when making decisions.

[muon2]

Yes, 95% percent is the usually understood odds for a beyond a reasonable doubt, but I view that as just rounding. The masses won't be comfortable with decimal points. And given that we have odds for that, why not for the other standards? As I said, some courts talk about odds. I really needed to know, given that I am on a grand jury. I live by the odds. I'm a Bayesian. I find it works for me, when making decisions.

Fascinating. Do those same masses know that 95% probability of being correct translates to 1 out of 20 convictions being in error? I suspect that the actual error rate for convictions is much lower than 1 in 20, so somewhere along the way a threshold much tighter than 95% is being applied.

[muon2]

As I think more about the actual false conviction rate I think there may be other factors at work reducing it below 1 in 20. Specifically I think the prosecutor's decision on whether to proceed may matter.

For example, suppose the prosecutor uses the same 95% threshold for reasonable doubt to decide whether to go to trial. If the jury saw the facts exactly the same way as the prosecutor (as of course they wish) then a 95% threshold applied by the jury would exactly correlate with the prosecutor and there would still be a false conviction rate of 1 in 20. However, if the jury saw the facts completely independently from the prosecutor, but still used a 95% threshold, they would be statistically independent. An innocent suspect would expect to be wrongly convicted in 5% * 5% of the time. That is 0.25% or 1 in 400. Real situations would have some correlation so the compound probability would be somewhere between 1 in 20 and 1 in 400.

This still seems like a higher false conviction rate than happens in reality. If juries were known to apply a 95% threshold, then I would surmise that prosecutors were applying a higher threshold in their decision to go to trial with a case.

[Torie]

As I think more about the actual false conviction rate I think there may be other factors at work reducing it below 1 in 20. Specifically I think the prosecutor's decision on whether to proceed may matter.

For example, suppose the prosecutor uses the same 95% threshold for reasonable doubt to decide whether to go to trial. If the jury saw the facts exactly the same way as the prosecutor (as of course they wish) then a 95% threshold applied by the jury would exactly correlate with the prosecutor and there would still be a false conviction rate of 1 in 20. However, if the jury saw the facts completely independently from the prosecutor, but still used a 95% threshold, they would be statistically independent. An innocent suspect would expect to be wrongly convicted in 5% * 5% of the time. That is 0.25% or 1 in 400. Real situations would have some correlation so the compound probability would be somewhere between 1 in 20 and 1 in 400.

This still seems like a higher false conviction rate than happens in reality. If juries were known to apply a 95% threshold, then I would surmise that prosecutors were applying a higher threshold in their decision to go to trial with a case.

A lot of factors go into the decision as to whether to indict by a DA, and some DA's are corrupt. In Columbia County, the DA will not plea bargain, after securing an indictment. A plea bargain needs to be done before an indictment. In federal court, the conviction rate is incredibly high, because federal judges get mad if a criminal case is brought before them, that is not airtight, and will punish DA's who bring marginal cases down the line. You are right, that the independence factor, reduces the error rate in general though. And juries sometimes do nullification, which I think is appropriate actually. And some DA's are ethical, and know, even if the case is not beyond a reasonable doubt, if they prosecute, they will get a conviction. That happens with alleged cop killers, drug dealers, and child sexual molesters (the latter two are the most common kind of cases in Columbia County - drugs are everywhere, and there seems to be a shocking number of child molesters). An ethical DA will be very careful to not bring such cases that will result in a conviction, but should not so result.

I know the local DA well having interned for him. We discussed these issues at length. Now I will be assessing his cases for six weeks. I suspect to the extent he does not already, he will be bringing well prepared cases. He knows that I am very suspicious of authority, and am the type that tries to level the playing field, to protect the powerless, and will not cut corners. Not when somebody's freedom is at stake. I am acutely aware that even though I have the skill set to avoid being screwed, most do not. That is most sobering to me. And the criminal defense lawyers around here, are not very good, which is also sobering.