a) “Only one person in 1000 would have matching genetic markers”
b) “The test shows 99.9% probability that the sample came from this person”
Those statements are not the same.
Oh, balls. That’s a long way from true. Partly because a good number of people don’t understand percentages in the first place, and partly because the sort of arguments given in court are often not the sort of things that can be quantified in the first place.
For me it would have to be 100%, or so close an alternative is impossible, based on considering reasonable factors in calculating the probability. So you leave out things like aliens and supernatural acts. But based on the evidence, if someone else could have committed the crime, you can’t convict. What is a reasonable factor in determining probability is often going to be difficult to determine, and not agreed upon by everybody.
I remember a famous case in Boston involving an au pair and evidence of shaken baby syndrome. The au pair was accused of deliberately injuring the baby, but a doctor testifying for the prosecution state that there was only a one in a thousand chance of the babies injuries occuring by accident. That should have been enough to acquit right there. But she was convicted anyway (the verdict later set aside on more general grounds).
You should need enough evidence to rule out any alternative based on reasonable factors to convict. I doubt that’s the standard that’s actually applied.
And then the geneticist applies a Bayesian analysis, using some reasonable prior (such as that the perpetrator was equally likely to be any male living in the metropolitan area), and thereby turns a statement of the form of (a) into a statement of the form of (b).
But if it’s a city of a million people, then you’ve got about 1000 people local with matching DNA.
Find a person with matching DNA, there’s a 99.9% chance he’s the wrong man
There’s information missing from the hypo, though, and without it we’re considering the DNA evidence in a vacuum. How did the police come to suspect the person whose DNA it is in the first place? Was he identified, or did other evidence point to him as a suspect? Did he simply live nearby? Let’s say that the police took the DNA sample and cross checked it against a sex offender DNA database, and a mile away there’s a sex offender who is a 99% match. Standing alone that shouldn’t be enough to convict - in a city of a couple million there are still statistically thousands of people who might meet the same criteria, dozens of whom might be adult males, the vast majority of whom won’t be on any sex offender database. For all we know there could be a guy who is a 99.99999999% match right next door who’s not on the database.
However, change the facts a little and say that police found the suspect hiding in a creek nearby after being seen by running away by neighbors. Both the victim and the neighbors picked the suspect out of a lineup, one of his shoes was found in her house, she correctly identified the clothing and several tattoos that the suspect has, bite mark anaylisis showed that her bite marks matched his teeth, but due to contamination the DNA evidence could only be linked to him with 80% certainty. Standing alone the DNA evidence estabishes nothing, but taken as a whole with all the other evidence it’s another brick in the wall establishing the suspect’s guilt. That’s one reason why the law strongly leans away from assigning percentages to the beyond a reasonable doubt standard, or in many jurisdictions even a definition. To establish a percentage is confusing, since it’s not clear if it’s applied to each part of the evidence individually or to all of the evidence as a whole, and either way it’s somewhat meaningless to assign arbitrary percentages to abstract concepts.
While I certainly agree that DNA evidence can be combined with other evidence to produce an even stronger case, in most cases, the DNA evidence by itself is going to provide a far greater degree of certainty than all of the rest of the evidence combined.
Close. More a) than b).
The report would be more like:
I tested the sample and found the following genetic traits… Gene1, Gene2, and Gene3. One in 1000 of a random sample of people would be expected to have this genotype by random chance. The defendant has this genotype and cannot be eliminated from the pool of persons who may have deposited the sample.
The way around this is to test several more genes. Test 13 to 16 loci and if the defendant matches them all then he is included in a very small group (1 in several billion, perhaps) of persons who may have donated the sample (assuming he has no identical twin). But… if he matches 15 and does not match the 16th then he is excluded 100%.
Provided there is not some innocent explanation for how the DNA got there. I cannot tell when a sample was deposited with any great accuracy.
DNA evidence has becomes a blessing and a bit of a curse at the same time for prosecutors. Juries expect to see forensic evidence provide such overwhelming certainty that they have raised the bar of the level of evidence they want to convict. It is a common enough phenomen that is has a name, the CSI effect.
Then again juries and prosecutors have gotten it all wrongin many cases before and DNA evidence has been used to correct these serious injustices.
So, no reason for your oddly precise statement, then?
In an ideal world no innocent person would be imprisoned. (Many of us share your concern)
But the needs of society and the rights of the victim must not be totally eclipsed by the former considerations, a balance must be struck. That balance is struck by the juror. Who am I to tell him/her they are wrong? Democracy is the prescription you argue against.
Sorry you see it that way ACE.
Whilst trying to convert Reasonable Doubt to percentage terms is doomed to failure. There may be other aspects of reasonable doubt where percentages do have relevance?
For instance: on a majority decision of 10 of the jurors, rather than 12.
Reasonable doubt in that instance is established by a maximum figure of a mere 83.33 % of the jury? Maximum because the ten jurors would have to be a 100% (there we go again) sure to achieve 83.33%.This lower figure is not supported in New Zealand (that has a similar common law structure to Britain) they have no such provisions for minority verdicts. (At least that was the case in 2001, and may still be?)
William Blackstone in his Commentaries on the Laws of England,( published in the 1760s)
Blackstone’s formulation
“Better that ten guilty person’s escape than that one innocent suffer” This is a rare occasion indeed where the law has come out on a limb. (Surely a must for defence council to include in the closing speech?) Perhaps mathematics could be used to advantage to support a contention of a reasonable doubt based on that minimum ratio of 10.1.But again this is a matter for the jury.
This. The answer to the question is 100%, but that’s not the important question. What is important is what constitutes reasonable doubt. A very short answer to that is, if there’s any reason to think that any other person could have committed the crime, or reason to think no crime was committed, then the defendant must be acquitted.
TriPolar and Steophan, since you both say the standard should be 100%, perhaps you could give us some examples of any conceivable situation where you’d be sufficiently sure?
This is just restating the question, if you’re saying that reasonable doubt is doubt based on reasonable factors.
It comes back to the nature of reasonable doubt. Sure, DNA evidence isn’t 100% certain, but if can give a probability of it being the suspect orders of magnitude greater than the number of people who could have been at the crime scene, then it is unreasonable to doubt it. For every single element of the crime, it must be unreasonable to doubt that the suspect is guilty. Not 90% of the elements, or 95%, or 99.999999%. If there is any reasonable doubt, he is not guilty.
So, any case where the prosecution have convincing evidence that the suspect is guilty, and the defence fails to introduce reasonable doubt that he is. If they introduce reasonable doubt to any point, however small, if it’s material to the case, he’s no guilty.
Really, though, the percentages don’t help matters, as we are not talking about anything countable or measurable.
You appear to be using a novel definition of “element of a crime”. I’m not aware of any crime that has 10 or 20 elements to it, and I’m certain that there is no crime with a hundred million elements to it.
But suppose that the defendant is caught on high-resolution security cameras, there are multiple eyewitnesses, and the DNA is as close a match as the screening technology is capable of detecting. The defense argues that the defendant might have a hitherto-unknown identical twin. Does this count as reasonable doubt? After all, even though unknown identical twins are quite rare, they’re still more common than 1 in 100 million.
I doubt it would ever be reasonable to believe that someone has a hitherto unknown identical twin, who showed up at the precise time and location we’ve shown someone to be*, without ever having been seen before, and who then disappeared without trace.
*This is important, as it means that the defence hasn’t introduced a credible alibi. If they have, then we already have reasonable doubt without the need for the mysterious twin.
You can get 100% out of that, because you have video evidence, backed up by the eyewitnesses, and no evidence of an identical twin. I’d assume you also had some motive, and additional evidence that the alleged perpetrator also had the opportunity. But based on DNA alone, you don’t have much. How did that DNA evidence get there? Why would the suspect commit the crime? Why is there no other evidence? You get to 100% (not a mathematical calculation) by ruling out the possibility of someone else committing the crime. Unknown identical twins committing crimes for no reason, and leaving nothing but DNA as the only evidence isn’t a reasonable factor.
Given that there are 12 people “independently” making the decision, with that level of confidence, who would all have to be wrong, wouldn’t the rate be .05^12, not 5%?