Continuation Part 14: Amanda Knox/Raffaele Sollecito

[anglolawyer]

Sounds right to me. As I understand it, he's developed a statistical model that tells you the likelihood of this set of data implying the presence of that set of data, taking into account the height of the peaks and everything else. I can very easily see how a statistician could do such a thing without actually knowing that much about DNA forensics.

Much like a not terribly thoughtful mathematician might write a whole chapter which just dismisses out of hand the idea there could be anything wrong with the DNA results (in spite of the evidence which she preferred to ignore …).

 

[Kauffer]

In other words, if you leave an object in a room and it's contaminated by partial profiles of lots of people's DNA, according to Balding it's still unlikely that by chance you'd get a full profile for Sollecito. Therefore it's much more likely Sollecito's DNA is actually there (regardless of how it might have got there).

But hasn't that been discussed here on this forum - the idea that if partial profiles appear then that is a greater indication of contamination than a full profile? I thought that the consensus was that it is not possible to argue that the appearance of a full profile within a mixed LCN sample is better evidence in itself that the contributor of that DNA did not arrive by means of contamination.

 

[LondonJohn]

In other words, if you leave an object in a room and it's contaminated by partial profiles of lots of people's DNA, according to Balding it's still unlikely that by chance you'd get a full profile for Sollecito. Therefore it's much more likely Sollecito's DNA is actually there (regardless of how it might have got there).

No - again, he's saying that Sollecito's DNA is definitely there, but then, over and above this, he's asserting that the Sollecito DNA profile he can see on that clasp can only reasonably have come from direct primary contact by Sollecito.

In other words, he's saying that in essence, his analysis is showing that Sollecito almost certainly touched the bra clasp directly. That's the important thing, and why he (and the prosecutors, and most pro-guilt commentators) think it's very strong evidence against Sollecito. But he's wrong and misguided, for reasons previously outlined.

 

[Kauffer]

katydid,

I interpret his statement to be about the DNA, not about Raffaele's being present in the murder room at the time of the murder. Clearly the latter would be a fallacy in the sense of taking sub-source level DNA evidence and drawing conclusions at the activity level. With respect to environmental DNA, it is possible that some PG commenters, including Thoughtful, are trying to draw a distinction between one or two drop-in alleles and Sollecito's putative profile.

At the risk of getting hopelessly out of my depth on the science, I too did not interpret his statement to mean that he was effectively saying Raffaele was in the room at the time of the murder, but simply that the appearance of his full profile cannot be explained by environmental contamination - He doesn't rule out all innocent explanations for the appearance of the profile. But in saying what he said, he was, according to my understanding of the science, going further than the evidence permits - or is he being more nuanced than I am giving him credit for?

I would still contend however, that most laymen and women listening to that interview would be left with the distinct impression that there was at the very least, something necessarily fishy about the evidence, which does not cast Raffaele in a very good light.

 

[LondonJohn]

Ah well I guess this all hinges on the interpretation of Balding's very poorly chosen phrase:

"how likely is the evidence under the prosecution assertion that DNA is there from Raffaele Sollecito and again how likely it is without him being present"

The two ambiguous terms here are "it" and "him being present". Is "it" Sollecito's DNA, or a set of peaks corresponding with Sollecito's DNA profile? And does "him being present" mean "his DNA being on the clasp" or "him being present at the time of depositing of the DNA".

At the very least, you'd hope a mathematician might choose his words a LOT more carefully........................

 

[lonepinealex]

Looks like there's a general problem with the science communication with respect to DNA evidence. Which is why Peter Gill felt compelled to write his book.

As with confessions, DNA is generally understood by the public to be some kind of gold standard of evidence, with the problems and nuances being complicated to understand. Similar to the way polygraphs used to be thought of as solid evidence (and still are to an extent by the general public).

Balding is evidently not an effective science communicator, as even people on this thread who understand this stuff are able to interpret his statements in very different ways.

We need more people like Gill explaining the issues with DNA, especially LCN DNA, so that the public understanding of this type of science is vastly improved, for the sake of defendents and victims everywhere.

 

[Chris_Halkides]

PNAS quotes from Dr. Balding

From Dr. Balding's PNAS paper:
"I first compare these prosecution (Hp) and defence (Hd) hypotheses for the contributors
of DNA:
Hp: Kercher, Knox, Sollecito, and one unknown individual
Hd: Kercher, Knox, and two unknown individuals"

"I reran the analysis excluding Knox from both Hp and Hd, and obtained an LR in favor of Hp of 42 million (WoE = 7.6 bans). Thus, although the additional alleles have, by providing evidence for an additional contributor, weakened the evidence implicating Sollecito
by a massive 8 bans, this evidence nevertheless remains strong."

"After reclassifying as uncertain all peaks below 15% of the height at one extra repeat unit, a common stutter guideline (4), there remain four alleles not attributable to either Sollecito or Kercher and the WoE is increased to 10.7 bans.
Note that I cannot address here issues of how the DNA came to be in the exhibit: Possible contamination was an issue in the trial and appeal. I only consider whether there is DNA from Sollecito for which the evidence remains very strong after allowing for the additional alleles identified by Vecchiotti and Conti (2) and the possibility that apparent stutters are allelic."

 

[Kauffer]

Ah well I guess this all hinges on the interpretation of Balding's very poorly chosen phrase:

"how likely is the evidence under the prosecution assertion that DNA is there from Raffaele Sollecito and again how likely it is without him being present"

The two ambiguous terms here are "it" and "him being present". Is "it" Sollecito's DNA, or a set of peaks corresponding with Sollecito's DNA profile? And does "him being present" mean "his DNA being on the clasp" or "him being present at the time of depositing of the DNA".

At the very least, you'd hope a mathematician might choose his words a LOT more carefully........................

Oh I think the "him being present" definitely refers to the sample not to his being in the room. But I have a big objection to his assertion at the end of "extremely strong evidence". He cannot be referring there simply to the sample; he must be referring to a belief he has about relevance. So, "extremely strong evidence" of what, exactly?

 

[LondonJohn]

And if Balding is indeed merely talking about concluding that Sollecito's DNA profile is found within the data he was given to analyse because Sollecito's DNA is actually present on the clasp (as opposed to environmental contamination just happening to match all of Sollecito's profile peaks), then all he can conclude is this:

According to the data with which I was provided, it's indicated that it's highly likely that Sollecito's DNA was physically present on that item.

But again, given that he admits elsewhere knowing nothing about the fact that the data he was given to analyse had already been subjectively filtered by Stefanoni, he can't even logically come to the conclusion that "it's highly likely that Sollecito's DNA was physically present on that item". In order to be able to conclude that, he would have had to look at the source EDFs, and check that the electropherograms were not misrepresenting the EDFs in any way.

And that's even before one gets to the whole contamination issue (that is to say contamination by Sollecito's DNA, not by random environmental DNA).

Thus Balding is logically a VERY long way from asserting that his results are "extremely strong evidence" against Sollecito.

 

[LondonJohn]

Oh I think the "him being present" definitely refers to the sample not to his being in the room. But I have a big objection to his assertion at the end of "extremely strong evidence". He cannot be referring there simply to the sample; he must be referring to a belief he has about relevance. So, "extremely strong evidence" of what, exactly?

Yes - Chris' supporting info seems to clear this up, and it was my misinterpretation. And I totally agree to the rest of your post (see my previous post, which essentially says the same thing).

 

[Sergei Walankov]

The two ambiguous terms here are "it" and "him being present".

I had a similar thought with "very strong evidence". I suspect he means there is very strong evidence of transfer, be it primary or secondary, as distinct from environmental contamination, which I understand to be a different thing. But the audience would have heard it as "very strong evidence of guilt".

 

[Kauffer]

Looks like there's a general problem with the science communication with respect to DNA evidence. Which is why Peter Gill felt compelled to write his book.

As with confessions, DNA is generally understood by the public to be some kind of gold standard of evidence, with the problems and nuances being complicated to understand. Similar to the way polygraphs used to be thought of as solid evidence (and still are to an extent by the general public).

Balding is evidently not an effective science communicator, as even people on this thread who understand this stuff are able to interpret his statements in very different ways.

We need more people like Gill explaining the issues with DNA, especially LCN DNA, so that the public understanding of this type of science is vastly improved, for the sake of defendents and victims everywhere.

One of the difficulties that persists for lay people is that when experts disagree with other experts they do so usually in a quiet and measured fashion avoiding ad hominems completely. They do not behave like polemicists, rhetoricians and lawyers. Often, a real sense of disagreement appears lost amidst the professional courtesy. It is often not particularly easy to differentiate. These people are not Christopher Hitchens.

 

[anglolawyer]

Yes - Chris' supporting info seems to clear this up, and it was my misinterpretation. And I totally agree to the rest of your post (see my previous post, which essentially says the same thing).

I had a similar thought with "very strong evidence". I suspect he means there is very strong evidence of transfer, be it primary or secondary, as distinct from environmental contamination, which I understand to be a different thing. But the audience would have heard it as "very strong evidence of guilt".

Maybe he only means what he is competent to say: very strong evidence of Raffaele's profile in the mixed sample, assuming nothing about the source and origin of the sample itself.

 

[LondonJohn]

Maybe he only means what he is competent to say: very strong evidence of Raffaele's profile in the mixed sample, assuming nothing about the source and origin of the sample itself.

Maybe. But he should know full well that the term "evidence" has a specific meaning when he's working in the world of criminal trials. It means that the fact has probative value in regard to assessing guilt or non-guilt.

And his findings about the bra clasp can only be considered "strong evidence" under this definition once one has also considered how and why Sollecito's DNA might have come to be present on the clasp. Until and unless that is considered, he should have been using terminology such as this: "...it's an extremely strong indicator that Sollecito's DNA was physically present on that bra clasp".

 

[lonepinealex]

One of the difficulties that persists for lay people is that when experts disagree with other experts they do so usually in a quiet and measured fashion avoiding ad hominens completely. They do not behave like polemicists, rhetoricians and lawyers. Often, a real sense of disagreement appears lost amidst the professional courtesy. It is often not particularly easy to differentiate. These people are not Christopher Hitchens.

Exactly, and I think that is a good thing - the professional courtesy and dignity in the science community should be encouraged in the wider world, along with evidence-based thinking.

It's a world away from the combative legal world where sophistry is often more important than nuance and evidence.

However, scientists need to do more to communicate effectively, even if it is a bit of a Catch 22 since, as Ben Goldacre always has it, "I think you'll find it's all a bit more complicated than that".

 

[Kaosium]

Related to this: Maresca said today that his time defending the Kerchers is formally over. Case closed.

But "defending"!? At most he was **advancing** their interests at trial! To the point of his shameful display of Meredith's lifeless body in court.

He was defending the police and Mignini, he was never truly serving the Kercher's interests. I don't know why that's so hard for them to figure out.

 

[LondonJohn]

By the way, and not a propos our current discussion, I note with interest that there's still a good bit of potential defamation of Knox and Sollecito going on (mostly in cyberspace). Perhaps some people need to be reminded that it's potentially defamatory now to call Knox and/or Sollecito "murderer(s)", "killer(s)" etc. (Though it would not be defamatory to express an opinion that you "believe she/they murdered Meredith Kercher", or that "the evidence supported guilt" etc.)

I hope - as others here and elsewhere have said - that Knox and Sollecito are just getting on with their lives and ignoring the small, weird cadre of nutters who cannot help themselves from continuing their vitriolic attacks. But I hope that someone's at least keeping an eye on it on their behalf, just in case. Perhaps legal letter reminders to the people responsible for running certain websites, plus the administrators of twitter, might be in order before long.

 

[carbonjam72]

Indeed - and this is the nub of the problem here. Balding is in fact only able, both logically and ethically, to state that the data he was given show Sollecito's DNA profile. That's all.

When he starts trying to place his findings into a real-world situation, where he draws (incorrect) inferences regarding the actual evidence against Sollecito, he's acting from a position of ignorance and incomplete information.
To go back to my little example for a moment, it would be as if the statistician tasked with analysing the filtered numbers I'd given him were to say something like "The machine outputs the number 7 more than any other number". In fact, all the statistician could say was that from the data (s)he was given, the number 7 occurred more than any other number. And (s)he would have actually have been factually incorrect to say that the machine outputs the number 7 the most, since the unfiltered data show that 4 is the most common output number. And what would be worse would be if that statistician didn't even know that the numbers had gone through my subjective filtering exercise before (s)he was given them to work with.

Balding's comment on environmental contamination is, as you point out, worse still. It's stunningly ignorant and, as you say, entirely outside his field of competence and his remit. From an academic, this is an extremely worrying situation.

I'm finding this conversation very interesting. When I first saw this documentary, I thought Baldong had over-reached when he said something to the effect "strong evidence against Raf".

At the time I thought he had over-reached, but listening to this discussion, I'm getting the sense that it may have to do with creative editing by Vogt to give an inaccurate sense of what Balding actually said.

It's hard to believe a professional would be competent in one minute, and a dumbbell in the next, but there it is.

In Dr Gill's BBC radio 4 interview, he talked about the "CSI effect", the public's bias from TV shows that the mere finding of DNA automatically infers guilt. He said (IIRC) something like, "even some scientists are susceptible". I wonder if Dr Gill was referring to Dr Balding?

Supposedly, 'thoughtful' will be co-teaching a class with Balding in the near future on DNA and forensic stats, something. Birds of a feather...

 

[carbonjam72]

I would guess that the one she was paid for was this "huffpouk" article:

http://www.huffingtonpost.co.uk/selene-nelson/amanda-knox-pr_b_5694432.html

Interestingly, Nelson works as a "content executive" for a digital marketing company called "Jellyfish" in Brighton and Nigel Scott has suggested that her connections there may well have helped her in placing the article.

See:

"Amanda Knox, Raffaele Sollecito and the troll factor"
http://injustice-anywhere.org/wp-content/uploads/2015/02/HolidayNewsletter2014PDF.pdf

http://www.jellyfish.co.uk

I thought HuffPO never pays anyone? Maybe different in UK?

 

[Sergei Walankov]

I thought HuffPO never pays anyone? Maybe different in UK?

My understanding as well. Didn't she have something in Salon?