Your Honor, if I could, I would like to use exhibit 410 again. And I believe that's currently on the elmo, if I may.
In terms of this chart--and you've testified in answer to questions by Mr. Thompson that you believe that this is an appropriate method of describing frequencies in a mixture such as this mixture?
In your opinion, are the numbers, and we'll just look at the three numbers that are at the bottom of the three major ratio groups listed there, they're approximately 50 percent; is that fair?
Okay. I would like you to assume that in that stain--and are you familiar with the actual alleles that were detected in that stain?
I'd like you to assume that that was donated, that is those results from a single contributor. What would the frequency be of that happening?
I would like you to assume that there's two contributors to this particular result.
Is it your opinion that approximately 50 percent of the time, from two unknown people those results would be obtained?
No. If you assume 50--that there are two contributors, then I believe that Dr. Weir's method will give you the appropriate result.
And in fact, if there were two contributors, Dr. Weir's method is in fact an accurate method to describe the approximate frequency of those results?
As a corollary or connected with that, if you assume that there were three contributors to that stain, isn't it true that his method of calculating an approximate frequency as he described to this Court earlier with three assumed contributors is absolutely correct?
Yes. And what's interesting is the comparison between the table that you produced for this, which is the other one, that shows an assumption of two when the theory is that there may be three. It shows why just assuming two, as if you somehow know that from a mixture is wrong because three alleles let's you assume that there are two individuals. But in fact, the 29 stain, some people have suggested that it includes three individuals.
Okay. And would it be a fair--under the assumption that there are three, would it be a fair approach, making that assumption, to use the calculation method by Dr. Weir for three people to describe that situation?
If you use that assumption, yes. But it's a fairer assumption to make no assumption.
Okay. You talked a little bit about juries, the trier of fact in a case during your examination. Do you recall that?
If a jury were satisfied that there were two and only two contributors to a stain, would the model proposed by Dr. Weir in fact describe that situation?
And the same would be true as far as your testimony if the jury found there were three contributors to a stain using his model for three contributors?
Now, Dr. Weir described--and I use the term "His model." Are you familiar with the article that he described earlier today by Dr. Evett and David Stoney?
I've read it a while back. I have not read it recently and I spent some time on a--at the California association of criminalists meeting with David Stoney discussing these issues.
Are you aware of any scientific, that is peer review scientific publication criticizing that method?
Yes. There's a series of articles and commentaries that go from--that are based on a discussion of an article by Kathryn Roeder or Raeder. I'm not sure how she pronounces her last name, but it's r-o-e-d-e-r. And she published a paper exploring many of the issues, some of which included the problems associated with likelihood ratios, which means the problems associated with taking the frequency, adding assumptions and turning it into a probability, which is what Dr. Weir's done in my opinion.
Do you recall when Dr. Weir was using--one of the examples he used was--and this involved a series of databases. In other words, assuming if one person was Caucasian and another one was Hispanic, for instance, or Caucasian, African American--I'm sorry--African American and so forth. Do you recall him coming up with a frequency under one of those databases or two of those databases rather?
Okay. It's actually not a likelihood ratio. What it is is, it's a conditional probability. It's not a frequency. They are conditional probabilities. They are probabilities conditioned on a set of assumptions, for example, ethnic group, number of contributors and so forth and so on.
But that's the same thing. I'm sorry. The same thing happens. In the context of the descriptions of what's good and bad about doing likelihood ratios also applies to discussions of conditional probabilities.
Sure. If you--if you--for example, the one that you have in your--in the other exhibit, the number that comes out of that, which I believe was 1 in 71, is the conditional probability of those two genotypes given those two databases. The given makes it a conditional probability.
Okay. So when, for instance, numbers have--which have already been reported on the board--and let's use the Bronco automobile board to your left there and let's take the example of the center console, item no. 30, found to be--that is not to be excluded was Mr. Simpson. And there are frequencies written in there of from 1 in 520 to in 1400. Do you see that?
And did you hear the testimony about that item, for instance, when this portion of the chart was filled in?
Did you see any of those reported frequencies or have you read any transcripts of how that came about; that is that those frequencies were testified to?
Do you recall that in fact those frequencies were presented because of the differences in frequencies for particular databases?
In other words, 1 in 520 on that example might be from African Americans, Caucasians or Hispanics, correct?
And I simply don't recall which in particular produced one result. Is it your view then that that constitutes a conditional probability, the answer, that is the frequency to item 30?
Yeah. The 1 in 520, you can turn it into a conditional probability, but it's also a frequency. And as a frequency, you don't have to assume that. You can say that the frequency is 1 in 520 in that database.
Then when Dr. Weir uses the number 1 out of 71, he's making the same types of assumptions about databases, correct?
What he's adding to it is an assumption that there is, for instance, in his first assumption that there are two contributors?
So is that then the only thing that's been added on in terms of these conditional probabilities to what's already been reported in court?
It would depend on which one you're looking at. But you can assume that there's two and exactly two, then you do it for different ethnic groups and different combinations of ethnic groups. That adds another layer for example; that you have to do all the pair-wise comparisons among ethnic groups. And if you add the possibility of doing three, you could then say that you should do the probability of two or three, if you think either of those are reasonable. And what I would say is that if you're going to do the probability that it's two or exactly--exactly two or three, exactly three or four or exactly four, you can also do it the way that the NRC recommends, which is just to say, we're not going to worry about whether it's two, three or four exactly. We're going to say what happens if we don't make any assumptions about the numbers.
Incidentally, is it your testimony that the NRC's comment, that one last sentence in the paragraph, is clear and unambiguous?
Is it your testimony that--and you were here for Dr. Weir's testimony earlier this morning?
Is it your belief that the opinion he offered about that particular phrase "Being ambiguous" is wrong?
You also described, Dr. Shields, your opinion that in Dr. Weir's calculations, not likelihood ratios, but the calculation such as 1 in 71, that those were making assumptions about who the contributors were. Do you recall that?
In reality, don't those calculations make no assumptions about any of the known people in this case?
And what I'm referring to now are the calculations he made of frequencies, not likelihood ratios, but using as an example the 1 out of 71?
And in fact, we see the number 1 out of 71 twice in the African American, Caucasian combination for two contributors as well as in the African American, southeast Hispanic combination as well; is that right?
Is it your testimony that that number has anything to do with the known types of the two victims and Mr. Simpson in this case?
All right. It makes no assumptions whatsoever and in no manner uses the known types of Mr. Simpson or the two victims, does it?
No. But that--but see, Dr.--what I heard earlier today was Dr. Weir talking about known genotypes in the numerator. He was talking about likelihood ratios as well as frequencies today.
And in fact, it uses in no manner the known types of these three individuals, does it?
You described seeing mixtures in previous cases that you've worked on; is that correct?
No. The most recent one was in Minnesota in Federal Court, and those frequencies were withdrawn.
No. There was finally--there was a hair left that was not a mixture, and that was the only one that was presented.
Okay. What I'm asking though is, were any numbers presented in any of the cases you've been involved in--
They were originally presented and they were withdrawn the day of the Dahmer hearing.
The kinds of mixtures that are not mixtures associated with sexual assaults. Maybe three, four.
In three, there were attempts to present numbers with those mixtures. They usually were associated with presenting numbers for subsets of the mixture under the assumption that they were particular individuals.
In other words, were those frequencies calculated based on, as you used the term, a "Conditional probability"?
If you know there are two, yes.
It would be fair to say it's nearly zero. Never zero.
And what I heard earlier today was Dr. Weir talking about known genotypes in the numerator. He was talking about likelihood ratios as well as frequencies today.
The court reporter is about to kill both of you.