Yes, DNA science does advance and yes,of couse it would be a good thing to finally clear up the murder of Rachel Nickell. Just one small quibble?
The tests, part of a 13-year reinvestigation into Nickell’s murder, are said to show a one-in-five-million chance that the DNA did not belong to Napper.
Is that what they actually mean? For I\’m not sure how that can actually be calculated. Don\’t they mean that only one in 5 million have that DNA? Thus that there are in fact 10 people in the country with it?
“Is that what they actually mean? For I’m not sure how that can actually be calculated. Don’t they mean that only one in 5 million have that DNA? Thus that there are in fact 10 people in the country with it?”
Well quite. Years ago I attended an excellent lecture by a forensic scientist who beautifully illustrated the nonsense of these quotes. In fact, these kind of quoted odds are so misleading and yet used so often there is a special legal term for it: “The Prosecutor’s Fallacy.”
Positive DNA testing is considered a “slam dunk” by lay people, but this is very far from true. There have been several “matches” for completely innocent people. Raymond Easton was one (Google his case). As the DNA database gets ever larger, the chances of these errors will go up and up. The Home Office is currently stalling FOI requests on the number of cross-matches in the database (i.e. those people in the database who match other people in the database), which would give us a clue to the rate of false matches (the Home Office asserts that all the cross matches are the same people because of the way they don’t check identity properly before putting people in the database, but this is palpable nonsense).
The way I understand it, they only get bits of the perp’s DNA, not all of it.
A bit like, somebody’s record collection might overlap 10% or 20% with mine (because there are plenty of people with all the Stones and Led Zeppelin albums, let’s say). If we only took random samples of his and my collections, on the basis of a small sample, you might assume that they were identical.
The bigger the sample, of course, if everything matches up, the bigger the likelihood that you are looking at the same collection. Or the bigger the likelihood that there is an LP in the sample from his collection that is not the same as mine, which proves it the other way.
But these “1 in 5 million” quotes are total hokum, of course.
“The way I understand it, they only get bits of the perp’s DNA, not all of it.”
Sort of. They keep the original sample, but of course can’t sequence all of it. They run a bunch of enzymes over the DNA to snip out certain marker fragments (or “loci”). These are then what goes into the online database. The Home Office used to keep 6 loci, but this was too small (e.g. poor old Raymond Easton being swept up), so they’ve started to record 9 loci in the database going forward. Doesn’t help all the other innocent people in the database with 6 loci who can get fingered at any time (by any police force in the world: in another case a poor bastard nearly got pulled to Italy on a false match).
“But these “1 in 5 million” quotes are total hokum, of course.”
Actually they are right IF the odds of guilt from OTHER evidence are 1:1. In the absence of any other evidence (e.g. suspect picked up from DNA at the scene but nothing else points to guilt) then the reverse applies: there’s a 5/6 chance of innocence (60 million x 1 in 5 million). Thus it is not possible (should not be possible) to convict based on DNA samples alone.
Prosecutors rely on the fact that attempting to explain Bayes’s Theorem to a jury is doomed to failure.
You prats. Don’t you think it exonerates Stagg?
“You prats. Don’t you think it exonerates Stagg?”
Not sure I follow you.
I wonder how many jurors who think that 1-in-5-million is completely impossible regularly play the National Lottery?
“I wonder how many jurors who think that 1-in-5-million is completely impossible regularly play the National Lottery?”
Pretty much the same ones who smoke and don’t wear seatbelts.
The prosecutor’s fallacy is an excellent point to raise here. It should be considered in conjunction with the opposite fallacy, however. A defense lawyer would argue exactly as has been argued above, that with a match occurring randomly once in every 5million people, ten people will become suspects in a population of 50million, nine of them (presumably) falsely.
Whereas the prosecutor argues that the likelihood of innocence is 1 in 5m, the defense argues that the likelihood of innocence is 9 in 10!
But the opposite argument is that among those ten, it is likely that only one could be a plausible suspect. Probably only one of them lives in the right place, probably only one of them doesn’t have an alibi etc. Assuming there is at least SOME evidence OTHER than the DNA, it should lead you to one person above the others. So your chances of innocence are substantially less than 9 in 10.
In effect, in this example DNA functions to shrink the pool of eligible suspects from 50million to 10. Pretty helpful.
Tim adds: Well, no, because we don’t know who the other 9 are, do we?