Dr. Weir, when other well-respected scientists in your field express a disagreement with you, does that ever affect your position on the subject?
Well, Dr. Weir, is it your position that you have read all the critical comments that have been written either in letters or in publications which disagree with your position on these issues?
I'm sorry. I don't understand that question at all. I have read a lot of material involved in the forensic uses of DNA, but you're asking me to say--make statements about people disagreeing with what I do, and I would need you to be more specific than that.
You said on direct examination, Dr. Weir, that you have only consulted or testified on behalf of the Prosecution; is that right?
And that the only time that you've been invited to testify for the Defense, you declined; is that correct?
Now, wouldn't it be also true, sir, that you don't agree with all the methods that are used by forensic laboratories to calculate the statistical significance of a match?
Well, for instance, Dr. Weir, you told us that it is your approach to always put an upper confidence limit on these numbers; is that correct?
I say that--as I said, that goes along with the product estimate. And, of course, I agree with the product estimate reported by these laboratories and others. I would go an additional step.
And the additional step of a confidence limit will actually make the frequency more common, will it not, as an upper limit?
No. It has no effect on the frequency, of course, but the estimate remains as it is and then we have an upper bound on the estimate.
And the upper bound on what that estimate is is a more common number, okay, is not as rare a frequency as is the estimate itself provided by some of these forensic laboratories; isn't that correct?
That's true. We keep both things together. But you're true. The outer limit of this interval is more common than the estimate itself.
KEY QUOTEAnd you also testified, sir, that you believe that there is substructure within the--within the broader American ratio groups; did you not?
Yes. I think I said that. We certainly expect there to be genetic differences amongst groups. Yes. Uh-huh.
And that when you make your calculations in a case, that you take this further step to compensate for that substructure; isn't that correct?
And if you take that additional step to compensate for that substructure, that also will make the frequency a more common number, not as rare an event; is that correct?
Yes, it will. But we should say what effect it has. Reporting in a realistic amount of substructuring has an effect which is almost negligible.
Well, let me ask you this though. But both those approaches, compensating for your approach to compensating for substructure and your providing the upper confidence limit, depending on the data in a given case, that could be of some assistance to a Defendant who is confronted by a laboratory that has chosen not to take either of those two steps; isn't that correct?
That was long. I think--the estimate doesn't change. The estimate remains. The profile has a frequency estimate. That does not change. These additional steps give us some measure of how far apart we might be from the true answer. And, of course, we could give--I could give a limit the other way. We're only interested in the more frequent. Now, you know as well as I do that the estimate may be a lot more rare than we report. I believe it's a lot more rare than what we report. We are giving the upper confidence limit, not because we are saying the estimate is truly bigger than reported. It could possibly be. It could possibly be a lot less.
All I'm saying, sir, is that depending on the data in a given case, by providing the upper confidence limit, which you do in each case, and by taking that next step to compensate for substructure, those two steps can make the evidence less rare or have the impression of being less rare. Would you agree with that?
Not quite. The estimate remains the same. We are attaching a level of precision to it and we're only giving it one direction. That direction makes the estimate look more frequent. It's not changing the estimate.
Fine. But you said that direction makes the evidence look more frequent; is that correct?
That's the direction we've chosen to report. We could have chosen the other direction, but that would be foolish I agree. But this we are giving--we're giving half an answer. We're giving the upper limit.
But some of the testing laboratories that testify in court to their results don't do either those two things that you just described to give a higher frequency; isn't that correct?
Cellmark--to your knowledge, did Cellmark give a confidence limit as testimony in this case?
I'm not a hundred percent sure, but I think in some of those reports, there are confidence limits reported. I would need to go back and look.
To your knowledge, did the Department of Justice offer confidence limits when it gave its numbers in this case?
Okay. So in those situations, depending on what the data is, if a laboratory chose--laboratory used by the Prosecution chose not to take those two additional steps which can create, as you just said, going in one direction a more common frequency, then your assistance to a Defendant or a Defense attorney could have some importance; could it not?
Yet, you have never in all the years that you have been involved in this issue and testified in cases even offered to consult without testifying to a single Defense attorney on any case; is that correct, sir?
Well, that's correct, because the Defense attorneys don't want me to do that.
KEY QUOTEThat's true. We keep both things together. But you're true. The outer limit of this interval is more common than the estimate itself.
I take that point... I certainly do.
Well, that's correct, because the Defense attorneys don't want me to do that.
No, they don't.