peterbirks (![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png)
peterbirks) wrote2005-11-01 02:35 pm
peterbirks) wrote2005-11-01 02:35 pm
They're at it again
From today's trial of the alleged murdered of Peter Falconio:
Carmen Eckhoff, a senior forensic biologist based in Berrimah, Darwin, said a DNA profile taken from Miss Lees' bloodstained T-shirt was compared with a profile obtained from a swab of Murdoch's mouth on New Year's Eve 2003 and was "found to be an exact match". Ms Eckhoff explained to the jury that the blood was 150 quadrillion - 150 million billion - times more likely to be his than any other caucasian male's in Australia's Northern Territory.
Now, without going too far into the statistics of forensic probability, I reckon that anyone who can write their own name will know that this number just has to be wrong. In fact, it's so gagaingly mind-bogglingly wrong that I'm still trying to work out how the biologist came up with the number. My best guess is that the entire seqience was taken. As we know, there are four basic "building blocks". So, if you take the entire genome sequence, and make each one a one-in-four chance, I suppose that you might come up with this kind of number.
Except, as we also know, building blocks don't work like that. DNA doesn't work like that.
Now, I'm not denying that the DNA sample is a powerful piece of evidence. But it isn't an irrefutable piece of evidence. In addition, for mixed populations and inbred populations the product rule is not as accurate. To the extent that the product rule is inaccurate, the error usually works against the suspect, unfairly exaggerating the strength of the evidence.
++++
And, on a different tack: this month's figures.
Carmen Eckhoff, a senior forensic biologist based in Berrimah, Darwin, said a DNA profile taken from Miss Lees' bloodstained T-shirt was compared with a profile obtained from a swab of Murdoch's mouth on New Year's Eve 2003 and was "found to be an exact match". Ms Eckhoff explained to the jury that the blood was 150 quadrillion - 150 million billion - times more likely to be his than any other caucasian male's in Australia's Northern Territory.
Now, without going too far into the statistics of forensic probability, I reckon that anyone who can write their own name will know that this number just has to be wrong. In fact, it's so gagaingly mind-bogglingly wrong that I'm still trying to work out how the biologist came up with the number. My best guess is that the entire seqience was taken. As we know, there are four basic "building blocks". So, if you take the entire genome sequence, and make each one a one-in-four chance, I suppose that you might come up with this kind of number.
Except, as we also know, building blocks don't work like that. DNA doesn't work like that.
Now, I'm not denying that the DNA sample is a powerful piece of evidence. But it isn't an irrefutable piece of evidence. In addition, for mixed populations and inbred populations the product rule is not as accurate. To the extent that the product rule is inaccurate, the error usually works against the suspect, unfairly exaggerating the strength of the evidence.
++++
And, on a different tack: this month's figures.
| SITE | $3-$6 | $1-$2 | 50C-$1 | $25 | $5 | $2-$4 | $5-$10 | Grand Total | ||
| PTY | $559.00 | $559.00 | ||||||||
| UB | $33.00 | $168.50 | $83.90 | $9.00 | $47.00 | $341.40 | ||||
| Betfair | $14.50 | $5.50 | $35.25 | $142.00 | $157.25 | |||||
| SJ | $1.25 | $1.25 | ||||||||
| EMP | $530.75 | $95.50 | $29.10 | $655.35 | ||||||
| PS | $10.50 | $48.75 | $123.50 | $182.75 | ||||||
| PP | $9.00 | $25.00 | $34.00 | |||||||
| Total | $574.25 | $231.75 | $98.40 | $1.25 | $35.25 | $111.00 | $482.90 | $498.50 | ||
| Total Hours | 23.25 | 30.50 | 10.00 | 3.50 | 11.50 | 23.25 | 4.50 | 106.50 | ||
| Avg per Hour | $24.70 | $7.60 | $9.84 | $0.36 | $3.07 | $4.77 | $107.31 | $4.68 |
The defence
As such, your prediction on what line the defence will take is irrelevant to my point. This is a general principle here, not a reference to a specific case and whether or not the defendant is guilty (or indeed whether or not it is his DNA).
My point is, just because something is "very likely" (as it is in this instance), that does not mean that you can throw numbers like "one in xxx quadrillion" around with gay abandon.
Of course the defence is not going to fight the case on the fact that the true chance is (say) one in 7000, rather than one in 150 million billion. How does that deflect my argument? Your point is correct, but a complete non sequitur.
PJ
Re: The defence
(Anonymous) 2005-11-02 12:33 pm (UTC)(link)This is really my point; they wouldn't, logically, gain any ground in knocking off a couple of zeros, since it would only add a trivial amount to reasonable doubt. Whether or not it impacts on the subcosncious of the jury, makes the expert-witness/defence less credible is another matter.