Thank you, ladies and gentlemen. Please be seated. All right. Let the record reflect that we have now been rejoined by all the members of our jury panel. Good morning, ladies and gentlemen.
THE JURY: Good morning.
All right. Dr. Weir, would you resume the witness stand, please.
Bruce Weir, the witness on the stand at the time of the evening adjournment, resumed the stand and testified further as follows:
Doctor, you are reminded, sir, that you are still under oath. And Mr. Neufeld, you may continue with your cross-examination.
Good morning, ladies and gentlemen.
THE JURY: Good morning.
CROSS-EXAMINATION (RESUMED) BY MR. NEUFELD
Dr. Weir, is it your position that the appropriate or ideal--let's call it the ideal database in a criminal case for analyzing DNA evidence would come from a census of the population of possible perpetrators?
Yes. The word "Census" means a complete enumeration. I suppose that would be ideal. That would be beyond practicalities.
Well, have you ever in fact stated that in your own articles, in your own writing?
In fact, I think--could you explain, by the way, what that term means, to the jury?
Certainly. I think that is quite a good point. What we are trying to do is to estimate the frequency of one of these now absent profiles. We want to know how often something occurs and the way to answer such a question is to go out and see how often it occurs. I mean the ideal, and I say ideal meaning that is sort of the ultimate extent, is to go out and type everybody in the world and that would be certainly wonderful, because then we would know, when we saw a profile, how many times it occurs. I think I said on Friday, or maybe Thursday, we can't even count everybody in the world, let alone profile them.
Now, I think also in your writing, sir, you used as an example in a criminal case where the police were actually able to go out there and type the entire reference population of possible perpetrators, which was a case in England in Leicestershire?
Yes. That was the first--the first criminal application of--of DNA was an interesting case for many points. But indeed the crime was committed in a village and the police got blood from everyone, I think in actually two adjacent villages, something like 5000 people.
And in that case what they did is they typed all the male members, I believe, of those two villages?
So that would be an example where in fact you have a database comprised of a census of the population of possible perpetrators?
Well, it wasn't a database. They didn't do any frequencies. They just types everybody and found the one--found the person who matched, so there were no calculations done. I wanted to clarify the word database.
Okay. But they at least had data on the entire population of potential perpetrators, at least in this case?
Would you agree, sir, that in each case, each case defines what would be its own appropriate population of potential perpetrators?
That's right. We always--when there is a case and there is a matching profile, there are two possibilities, either we know who the profile came from or we don't know, but this unknown person has got to come from somewhere. And as Mr. Neufeld said, it is convenient to call that group of unknown people the population of potential perpetrators. That is a funny phrase.
So, for instance, in this particular case, where Mr. Goldman and Nicole Brown Simpson are killed, in--in a place somewhere in Los Angeles, in the area called Brentwood, would the population of potential perpetrators include those people with a certain geographical proximity who could potentially have committed this crime? Would it include those people?
It would certainly include those among others. That may be the most natural description.
And when we say "Include," that would mean at least including the people of Los Angeles or the greater Los Angeles area, if you will?
Okay. Now, in this particular case, sir, did either Cellmark or the Department of Justice rely on a database comprised of the population of potential perpetrators?
Not in this case nor in any other, because the population of potential perpetrators is not sufficiently well-defined that it can be sampled.
Well, you've also written, I believe, Dr. Weir, that shy of doing a census of the entire population of potential perpetrators, the next best approach would be to take a representative sample of that population of potential perpetrators; is that correct, sir?
Yes. We can't get everybody. A sample from the--kind of easier if I say the irrelevant population. If we could do that, we would. We--I--it is almost always the case in a crime that that population is not sufficiently delineated. You know, we say that this one could involve Los Angeles, it could involve San Francisco. It could involve someone who has just hopped off a bus, committed a crime and hopped back on. So it is very difficult for us in practice to say who--which exactly--which group of people there is that we should be considering, so because we can't describe it, we can't take a sample from it.
Umm, Dr. Weir, do you have any published evidence that would show that the multilocus frequencies in the Cellmark and DOJ databases are the same as the population of potential perpetrators in this case?
And would you agree, sir, that the databases that were relied upon in this case are not even representative of the population of potential perpetrators for the crime committed in this case?
Well, that is really confusing things a bit, I think. It is not representative in the sense it doesn't contain people just from this area. It is representative, however, in the sense that the frequencies we obtain from the Cellmark or the FBI databases are applicable to the crime in this case.
Well, sir, would you agree that, for instance, not even large segments of the population, of people who live in Los Angeles County, are reflected in any of those databases that you relied upon from either the FBI, the Department of Justice or Cellmark?
Well, I agree in the sense that those specific people, but I disagree in the sense that people with those--with those profiles. We have a lot of experience now from FBI and Cellmark and many other laboratories in this country and around the world showing that whenever we make calculations, each calculation differs according to which database we use, but they don't differ in the sense of importance, in that profile is very rare in this population, which would be very probative or very common in this population, which would not be very interesting. So we find great consistency in the conclusion that these multilocus profiles, these complex DNA profiles, are rare no matter which database we use.
I'm sorry, but Dr. Weir, didn't you say just on Friday, for instance, for the various PCR markers, that you see large swings in certain profile frequencies between different racial populations, be it black, be it white or be it Hispanic?
Well, yes. For the PCRs the numbers are much smaller. The proportional differences are quite large. The actual differences are not and we are going from tens to hundreds, maybe to thousands where you are getting a range. Of course that is a big range of number, but we didn't start out by saying the PCR--PCR profiles are extremely rare, but no one has ever claimed that. The RFLP profiles are extremely rare no matter which database we use.
Now, Dr. Weir, returning to that issue of whether or not the databases that you relied on in this case are a representative of the population of potential perpetrators, are you aware of the fact, for instance, that California has the largest concentration of Asian Americans in the United States?
Okay. And, sir, the databases that you relied upon in this case--well, furthermore, are you aware that there is a very substantial Asian American population residing in the greater Los Angeles area?
So would you agree that the databases that Cellmark relied upon in this case do not include Asian Americans?
Well, of course they don't, but I think I've explained that I don't believe that matters, because when we--when we examine Asian databases, we still find that an RFLP profile is rare and a PCR profile is in the same ballpark as from the Hispanic and African American and Caucasian. Your point is quite valid. There are no specifically designated Asian databases used in this case. I don't think it is going to cause us any misunderstanding of the nature of these profiles.
And sir, not only are there no Asian American databases from Cellmark, but again, you agree there is no Asian American database that you have relied upon here in analyzing the DOJ or FBI work; isn't that correct?
No, I haven't used Asian. I have used the Hispanic, African American and Caucasian, three somewhat different racial groups, and their frequencies are consistent in that all three of them tell us the same story. The RFLP profiles are astonishingly rare and the PCR profiles are rare, but not astonishing.
Well, and that the PCR profiles vary considerably depending upon which database you use; isn't that correct, sir?
Proportionately, but not in absolute terms. They still are in the range tens to hundreds or thousands.
Sir, whether or not you do the PCR databases, wouldn't you agree that there are significant differences between the various ethnic groups?
Umm, well, I'm not sure on the profiles. We will have to examine each one specifically. When--the numbers are different. If they are significantly different, we mean is--is one included in the confidence limit of the other--
No, I hadn't. Now, for the profiles, consisting of PCR typically--well, up to 14 bands where there are seven PCR. Each one of those 14 bands by itself can differ markedly. We've got some numbers here that can be two or three percent in one database and maybe 20 percent in another, an astonishing difference for each individual item, and some--some people have got very upset about this--this difference for each band. It is interesting, but it is misleading in that we are not using individual bands. We are using the whole package. Now, if I have a band which is more frequent in African Americans and less frequent in Caucasians, it almost has to follow that another band must be flipped around. We've got to have some balancing out between the databases. It just couldn't be otherwise. So when we multiply together these varying bands, the whole package, the whole profile doesn't differ as much as the individual bands.
So just to cut to the numbers themselves, if you look, sir, at the table you have for three contributors for the various mixtures--could you look at that for a second?
Well, this comes, if these three unknown people we are supposing--we don't know where this profile came from. If those three were African American, southeast Hispanic and southwest Hispanic, in that case the profile would occur one time out of four.
Okay. That is if the mixture came from an African American, south eastern Hispanic and a southwestern Hispanic, correct?
Well, I thought so. It is actually two African Americans and a Caucasian. I apologize.
So this is going to be for three contributors. Now, would you please tell me what is the most common profile?
Now, according to these databases that you relied upon, which--could you please give me the racial or ethnic mix of that group of people which would have the rarest frequency?
Would you agree, Dr. Weir, that the difference between a frequency of 1 in 2 and a frequency of 1 in 3500 is extremely significant?
Umm, well, I'm going to have to do it in two parts. Of course it is. And I should explain why the difference is so dramatic in this case. I've done--I woke up this morning at 5:00 and thought how can I possibly avoid embarrassing myself again in Court, and I thought I will do every conceivable thing to make these frequencies as conservative as possible. And I noticed that I believe item 29 has an LDLR--has a polymarker component and I noticed that in also African American database it was a suggestion, just a hint of some dependence between components of the polymarker. So I took out some of the information for the African American contributors to item 29, which is--so really we are comparing--this is--I think this is an appropriate number, but we need to keep in mind we are comparing profiles with different numbers of bands in them.
All right. And sir, just as your calculations and your databases did not include any Asian Americans in them, would you agree that in the Los Angeles community there are various people, umm, of different Hispanic origin?
Oh, yes. I would go further. I think when we come down and look at people in particular, if we went round the room, would find that we each have different ethnic background, if we go back far enough.
To your knowledge, Dr. Weir, does either the DOJ databases that you relied upon or the Cellmark databases for their Hispanic database, contain people who originally are from South America as opposed to being from Mexico?
In fact, you have no idea whether there are any people in these databases from, say, Colombia?
Now, Gary Sims said that he relied on the FBI's RFLP database. Are you aware of that, sir?
And are you aware, sir, that a substantial portion of the FBI's Caucasian database is comprised of 220 self-described white FBI agents?
In your opinion, Dr. Weir, would you consider 220 white FBI agents to be within the population of potential perpetrators for the murder of Nicole Brown Simpson and Ronald Goldman?
Because their profiles--actually the identifiable agents is a lovely collection of people because they are from all over the country and although they say, as Mr. Neufeld said, they self-described them as Caucasian, we can be sure they have quite different European background, so this is really a lovely mixing up. We've got people from all over, each contributing their own unique profiles and making sure that the FBI's database represents the whole of the United States. I think it is good.
Dr. Weir, how sure are you as to the fact of whether or not the FBI's 220 white FBI agents represent European background from all over Europe? Do you have any personal understanding on that subject?
You have no personal information as to the ethnic origins or national origins of any of those 220 white FBI agents, do you?
And in fact, Dr. Weir, you said that other people in this white FBI database come from other parts of the country. I think you said all over the country; is that right?
All right. Now, other portions of the white database come from Texas, I believe you said, right?
Well, the FBI Caucasian database has actually four origins. There are FBI recruits, there are people who have given blood, blood donors from I think Miami, somewhere in Texas, I'm not sure if it is Houston or Austin, and somewhere in California here, probably Los Angeles.
Do you know for a fact that there are actually blood donors or from some other source?
Could you show me the source of information that you are relying upon when you say that the sources of subjects from California are derived from a blood bank?
Well, I'm going to have to cite my own paper in 1992 and then when I wrote that I checked with Dr. Budowle and that was my understanding at that time.
In the 1992 paper do you say that they come from actual blood donors in California?
Okay. Now, you mentioned that as to the white database of the FBI, that 220 FBI agents are included in that white database. Are you aware of the fact that the FBI's African American base does not include any FBI agents?
It is parallel to the Caucasian. That is the blood banks from those three areas; Miami, Houston and Los Angeles.
Okay. So if it is just had blood banks from Miami, Houston and Los Angeles, you have not included in that group any African American fib agents; isn't that correct, sir?
And sir, are you aware that the reason there is no African American FBI agents in the FBI's African American database is they don't have sufficient numbers of African American FBI agents working for the FBI?
Well, that makes sense. They just probably didn't choose to separate out that group if it wouldn't be numerous enough.
So you are aware of the fact that these databases are really nothing more than convenience samples? Is that a fair statement?
They are the best things we can do. If they call it convenience, that is a good term. We've got to get blood from people. It is hard to get anything from people, and it is very hard to get blood from people, so it makes sense to go to the place where the blood is kept, the blood banks.
Dr. Weir, as opposed to convenience sample, you would agree that the samples used in these databases are not what statisticians refer to as truly random samples; isn't that correct?
Well, they are not truly random. That would require us to list the population and take a sample from this list and some other scheme. It is unnecessary to do that. And I say that on the basis of what we see in databases collected under a variety of rules within this country and around the world. We just don't get led astray by using these FBI databases.
Dr. Weir, a moment ago you said that taking these blood bank samples is, I believe you said, a wonderful source of samples for analyzing this data; is that correct?
Now, Dr. Cotton testified several weeks ago in this case that the 1992 Cellmark database for African Americans is derived from 150 people who made contributions from a blood bank in Detroit. Are you aware of that fact?
Now, sir, do you have any idea how ethnicity or race was determined at that Detroit blood bank?
No, but what I'm hoping is that it was self-reported. I think what--I think it is very easy to overstate the ethnicity of these databases. Ethnicity is--we now know it is a very vague term genetically. We may look different but underneath we are extremely similar, so although it is interesting and maybe convenient to label these databases African American, Caucasian and Hispanic, it is sort of overstating the differences. For these profiles, particularly the RFLPs, they just don't differ very much in frequency.
But Dr. Weir, you are aware of the fact that the laboratories who did the work have chosen to rely on those racial and ethnic classifications in their databases; isn't that correct?
Well, I don't know that I say "Rely." They keep them separate and this has the advantage that we get different answers, so that it is clear to us what difference it makes. If we go to different sets of people, it is--sort of make my point. It just doesn't make a difference. I would be just as happy if they amalgamated all the databases into one and just gave a single figure. But by giving a range, if you like, they are being very open and saying, well, I have used three different samples and this is how they differ and they all tell me the same thing; these profiles are very rare.
Dr. Weir, a moment ago you said you do not have any knowledge as to how ethnicity or race was determined at that Detroit blood bank; is that correct?
And sir, would it be fair to say that you also have no knowledge how ethnicity or race was determined for anybody of the databases?
Well, if in fact--you mentioned the method a minute ago of self-reporting. Do you remember that?
In other words, where a person comes in and gives blood and the nurse or phlebotomist says, "What race are you" and the person says I am white, I am black, I am brown, I am green, I am whatever, okay, that would be self-reporting, correct?
Would you agree, sir, that self-reporting is not a scientific method of classifying people by race?
Mr. Neufeld, I don't think we understand what race is at a genetic level. We had a very interesting conference in Atlanta last fall where this was discussed extensively and reported in the newspapers. We now know as scientists that the whole idea of race is a very vague one, genetically, so what I'm hoping is that people don't get too detailed and specific about their race, whatever they choose to call themselves. What I'm more concerned is that we have a sample of people from as wide as base as possible.
Dr. Weir, if the Detroit blood bank relies on self-reporting of the individual donors, would you agree that that would not be a scientific method of classifying people by race?
I think the question is confusing in that it doesn't make any difference what the people are called. They can call themselves Martians and we will still get the same conclusions, that these profiles are rare. I think that the reporting of racial groups just is irrelevant.
Dr. Weir, you have reported data in this case based on racial groups, have you not?
I have used the databases as supplied to me and those are described by racial names, yes.
Have you in fact expressed ranges in your testimony that are present indicated on race as well?
Yes, because that makes it very clear what I'm talking about. It shows the effect of using different databases.
So Dr. Weir, I now ask you do you agree or do you disagree that simply relying on self-reporting to determine one's race is not scientific?
I don't understand the question, I'm sorry. Scientific meaning how we should calculate profile frequencies? I think what they've done is most appropriate.
Dr. Weir, if the person who drew the blood simply wrote down race based on the self-reporting, would that be a scientific way of determining someone's race?
For the purposes of these calculations, I don't think it matters what the person writes down.
I didn't ask you whether it matters, sir. I asked you whether it would be scientific. Can you answer that question?
I can answer the question in the context of what I'm doing here, and I think in the context of what I'm doing here it just doesn't make any difference.
I didn't ask you whether it makes any difference, sir. And your Honor, I would ask that the witness be instructed to at least answer the question.
No. The way you have phrased the question he is giving an answer within the context of his expertise.
Do you have an opinion as to whether or not it is scientific to simply rely on someone's self-reporting that I am either African American or Caucasian or Hispanic as a term of determining which database the person fits in?
I think it is doing appropriately. When you say it is scientific in this context, yes, I think it is.
Okay. How about if the person in the blood bank simply relied on his or her observations of the blood donor to determine which race the individual belonged in? Would that be scientific?
Well, given what I've said, that just--it is just as appropriate. It is just as arbitrary as self-reporting. I think it would be fine.
Dr. Weir, if it makes no difference how someone describes themselves and what database they fit into, how do you explain the huge differences on that board between 1 in 2, if you have people who are two African Americans and one Caucasian contributing to the mixture versus a frequency of 1 in 3500 if you have three Caucasians contributing to the mixture?
Well, the difference is in part with the different number of bands, but the more important difference is that they are based on three different--excuse me--four different samples. We could very well have found the same range for four samples of Caucasians.
But you couldn't find it in four different Caucasians in these databases, could you?
Sir, wouldn't you agree that when people simply report their race, if asked, say, at a blood bank, that that is not done in a scientific fashion?
I modified the language thinking that perhaps this will give the witness a different answer.
Do you believe that people who are reporting their race when they go to a blood bank, that those descriptions would be consistent with doing this in a scientific fashion?
Well, I can't answer unless we talk about what it is used for. If it was used to construct a evolutionary tree of the evolution of races of men, then it would be very important. If it is used to calculate forensic profile frequencies, I don't mind what they say.
But the reason you say you don't mind is because you don't think race makes a ditches now; is that correct?
And you don't think it makes a difference, notwithstanding the example that I put on that board which is--I don't think we gave a number to this board, your Honor.
And you don't think it makes a difference, Dr. Weir, notwithstanding the example that is reflected on Defendant's 1206; isn't that correct?
Well, as I have explained, it is a little wrinkle here, but if I had kept the same number of bands, the numbers would still have been different. This is a range for every profile over all these four databases, and I think the range would be similar if we took different databases of the same ethnic background, so the numbers are different. And I think that is very--very informative. It gives us a very concrete feeling of what these numbers are saying. We don't know what the frequency of a three-person pool of DNA is. We are trying to convey some sense. And in this particular case it can be quite common, 1 in 2, it can be moderately common, 1 in 3000, but now we have laid out all the calculations and we've got a very good understanding of the variation that we get by taking different samples. I think it is the best possible way to present this information.
KEY QUOTEDr. Weir, didn't you say, either early Friday morning or late Thursday afternoon during your direct testimony, that there is, in your opinion, substructure within the different racial and ethnic groups? Do you recall that?
Yes. There are groups of people. If we could--if we could identify a group, just put a rope around a group of people and say this is a group of people and measure their frequencies, they will be different from another group and another group.
And so I--sorry. So for instance, even amongst Caucasians you would, if there is this thing called substructure and that there were different groups of Caucasians, who may have different genetic frequencies for particular alleles than other groups of Caucasians, then for some of those groups alleles could be much more common or more common than for other Caucasian subgroups; isn't that correct?
Yes, I think you have captured the essence of it. It is the specific alleles that differ one group from another. It is the profiles, the collection where the differences balance out, that don't differ very markedly.
And so these databases, though, do not break people down into those various subgroups for which there may be different allele frequencies; isn't that correct?
The databases don't, but the calculations I took take that into. I do take that into account.
The calculation itself, that you say you do take that into account; isn't that correct?
Okay. But the calculations being done by Cellmark that Robin Cotton offered in this case do not take that into account, do they?
And the calculations that Gary Sims from the Department of Justice delivered to this Court also don't take that into account, do they--does it?
They don't take it into account. Of course what they have done is give the different racial groups, so they have conveyed a sense of how much variation there is in frequencies. They--instead of taking the subgroups within the Caucasian and looking at those differences, they have taken the differences between Caucasians, African Americans and Hispanics, so they have given a sense of how the answers vary.
Let's talk about that term "Hispanics" for a second, Dr. Weir. Cellmark used a database in this case which they characterize as Hispanic; is that correct, sir?
Wouldn't you agree, sir, that Hispanic isn't even a race, it doesn't define a race?
It doesn't define a race, and just to go to the next point, it doesn't have any relevance whether it does or not in this context.
Okay. And in fact when Cellmark is using the term "Hispanic," what they are simply saying is that these are people who contributed blood who have a Hispanic sounding surname; isn't that correct?
So you have no--no knowledge as to what their criteria was for putting somebody in their Hispanic database, do you?
Oh, I imagine there was a reason for calling--either they or someone calling the brood donor Hispanic, but as I've said, in several different ways now, that is fine.
Well, Dr. Weir, wouldn't you agree, again getting back to the issue of subgroups and substructure, that for instance people who have Hispanic surnames, who have a lot of, let's say, Central American Indian blood in them and they are living in, let's say, Southern California, may have very different frequencies for particular alleles than would people with Hispanic surnames living on the east coast who may be coming from Puerto Rico and have African American and Spanish blood in them?
Well, you can refer to my paper where I show that statistics are different. Sometimes RFLP structures have fifteen different types we can put into fifteen different categories. When I do a test between Miami and Houston, 14 of those 15 frequencies are different at a particular locus, and that doesn't surprise me. As Mr. Neufeld says, people who--someone has called them Hispanic living in Miami, they have a very different background from people living in Houston. The important thing, as I went on to show, to demonstrate in that paper, when we amalgamate all the bands from all the loci, the profiles don't differ to an extent which would be misleading. Sometimes the profile are not even statistically different. Sometimes they are, but they always convey the sense, locus RFLP profile is a rare phenomena.
But nevertheless, would you agree then that the Cellmark database which they characterize as Hispanic, I guess after what you just said, is a rather meaningless term? Would that be fair to say?
No, it is not meaningless. There is some rationale, after all, for someone being called Hispanic. It is a convenient label on the database. It is as good a name. I would have been happy if they called them databases 1, 2 and 3.
But that convenient label of Hispanic has nothing to do necessarily with the race of the individual who inhabits that database, does it?
Well, I don't think it can because we don't know what race means, so I don't--you know, I'm not trying to be evasive. I think we just don't mind what--what--what race a person is. We are concerned with the profile frequencies.
Well, sir, how do you define someone as African American for putting them into an African American database?
Okay. You have relied on the FBI's databases and they, too, have labels on them for different races, don't they?
And you have been consulting to the FBI, I believe you said, since 1989, haven't you?
And you and your university have been consulting to Cellmark for almost as long a period of time, haven't you?
Well, let me ask you this: Given the characteristics and descriptions that Cellmark and the FBI use for putting someone in their white database or their African American database or their Hispanic database, where would you, as the advisor to these people, put, for instance, the children that are the product of Mr. Simpson, O.J. Simpson and Nicole Brown Simpson? Which database would they go into according to that criteria?
I would be happy to flip a coin. I don't think it is going to make any difference to the calculations how a person is classified in the database. It is a convenient way of identifying different samples of people. The samples are all giving you the same kind of an answer, and I don't think it makes any difference.
KEY QUOTEI don't think--when you come down to it, I would be hard pressed to define my own ethnicity precisely. I'm obviously Caucasian, but I have grandparents from different European countries and I differ from other Caucasians in the room. I don't think it matters.
Well, Dr. Weir, if you don't believe in these classifications by race as being relevant, then how in good conscience can you sit here and make racial classification estimates as you did Thursday afternoon, Friday, and apparently today as well?
I will move on. In this case, Dr. Weir, when you provided a statistical estimate of the value of a certain piece of biological evidence, such as, let's say, a blood drop stain at the Bundy walkway or a drop recovered from the rear gate, you made certain assumptions, did you not?
I will go through them. I just want to at least establish that certain assumptions were made.
Okay. And I believe as you stated in your report, one of the critical assumptions that you have made for all your calculations is that the laboratories made no errors in the processing of the evidence; isn't that correct?
Oh, I can't--I can't even address that in the sense I just don't know what the labs did. They give me their reports and I just go from there.
Well, but in your report that you submitted to this Court and to counsel, you have a discussion of error rates, do you not?
Well, as a disclaimer in essence I say whether or not the laboratory made a mistake, it is really beyond my expertise and knowledge, so I'm just going to go with what they said.
Well, you also said that you are assuming that no errors were made; isn't that correct?
Well, my calculations start with the data, so that I said that, but it really means I just--the errors, any errors at any stage don't have any impact on my calculations. My calculations are frequencies of profiles.
Didn't you state that your calculations in this report assume the validity of all the DNA profiling done in this case?
As you sit here today as an expert in this case, sir, are you aware of the testimony of Dennis Fung and Andrea Mazzola and Mr. Yamauchi of the LAPD laboratory?
Did you ever review any of the manuals of the Los Angeles Police Department Scientific Investigation Division prior to your testifying today?
Were you aware that the Los Angeles Police Department has no protocol or manual in use for the collection and preservation of biological evidence?
Now, I believe you said a moment ago, Dr. Weir, that you have sort of chosen to ignore the impact of the probability of laboratory error would have on calculations in this case; is that a fair statement?
Well, have you taken into consideration in your own calculations the probability of laboratory error?
Well, let me ask you this, Dr. Weir: If you could quantify that variable known as laboratory error, would you agree that that would certainly be relevant to your calculations?
And it is not relevant to your calculations even if you could quantify that variable? Is that your opinion, sir?
You are asking me, in essence, to quantify something I don't think exists, so I reject the notion of an error rate. I reject the notion that it should be incorporated.
Well, sir, I believe you said that there are portions of this report by the national research council that you agree with and other portions that you disagree; is that correct?
Dr. Weir, in your report, which you filed with this court, did you not make the following statement and I quote: "Coincidental identity and laboratory error are different phenomena so the two cannot and should not be combined in a single estimate." Did you make that statement?
And you are aware, when you endorse that statement, that that was a statement that you were lifting right out of the national research council's report "DNA technology in forensic science"; isn't that correct, sir?
Sir, and in that very same statement or that very same thought, doesn't the report state the following--
Your Honor, just so the jury can read along, I'm asking to put the quote which I'm about to read from that paragraph up on the board.
All right. Dr. Weir, do you agree with the following quote from the NRC report: "Interpretation of DNA typing results depend not only on population genetics but also on laboratory error. Two samples might show the same DNA pattern for two reasons: Two persons have the same genotype at the loci studied or the laboratory has made an error in sample handling, processing"--I'm sorry--"In sample handling, procedure or interpretation. Coincidental identity and laboratory error are different phenomena so the two cannot and should not be combined in a single estimate; however, both should be considered." Do you agree with that paragraph, sir?
I agree regardless of the fact that it is quoted there. I have those thoughts entirely on my own volition.
Well, by the way, in your report, you include a portion of that paragraph, do you not, on page 3 of your report?
Well, you specifically excluded the last sentence, didn't you, sir? The last sentence of this report--
Hold on. Hold on. Hold on. There is a rule here. Only one persons gets to talk. Both of you are violating that rule. Dr. Weir, let him finish asking the question. Let him finish the answer. Proceed. Thank you.
In fact, the last sentence of that paragraph, Dr. Weir, is: "However, both," referring to laboratory error rate, and those population frequencies that you have given this jury, "Should be considered"; isn't that correct? Isn't that what the sentence says?
KEY QUOTEThe sentence states that correctly and of course that is way beyond anything I'm empowered to do. I'm an expert in the interpretation of coincidental frequencies. I'm not an expert or I have no knowledge, I should not venture into talking about errors.
Sir, in that last sentence, saying that "Both should be considered," was omitted from your report; is that right?
It is omitted from my report because I must omit it. I must omit that consideration in what I do. I am not allowed to say anything about what the laboratory does. I would be way out of bounds. I can only address the science of statistics and population genetics calculating these profile frequencies. To do anything else I would be way out of line.
Dr. Weir, do you agree or disagree with the following statement: "Especially for a technology with high discriminatory power, such as DNA typing, laboratory error rates"--
You need to give us a source, for starters. This is a foundational objection I take that as.
Would you please take it out and look at the top of page 89 and read that first paragraph to yourself.
Sir, I asked as an expert in the field of statistics do you agree with the statement in that paragraph?
Well, the statement has nothing to do with statistics and it is not anything that I have even considered, so I'm not sure--I can read bits and pieces from this document and agree with some and not others.
Sir, did you testify under oath in the case of state of Florida versus James Chassen on May 20, 1995, in a deposition in Raleigh, North Carolina?
And in that deposition, sir, when you were asked whether you agreed or disagreed with that paragraph--
When you were asked whether you agreed or disagreed with the first paragraph on the top of page 89, didn't you state under oath that you agreed with it?
I have no recollection. I talked to the counsel for about eight hours that day. I just don't remember.
While he is looking, if I may just ask one other question of the witness, umm--one second.
All right. The request was to ask him to read it and see if it refreshes his recollection at that point.
I'm asking him to read beginning on page 152 of the deposition through the top of page 153.
All right. Doctor, if you would just read that to yourself and see if that refreshes your recollection.
This is your examination of you that starts with that question and then here is your answer.
Now, sir, referring to that entire paragraph at the top of page 89,--and by the way, is the--
Sir, on this previous occasion on May 20, 1995, when you were under oath in that deposition, didn't you state that you agreed with that top paragraph on page 89?
All right. All right. Dr. Weir, this is a paragraph from page 89 of the NRC report. I'm asking you now whether you agree or disagree with the following, quote: "Especially for a technology with high discriminatory power such as DNA typing, laboratory error rates must be continually estimated in blind proficiency testing and must be disclosed to juries. For example, suppose the chance of a match due to two persons having the same pattern were 1 in one million, but the laboratory had made one error in 500 tests. The jury should be told both results. Both facts are relevant to a jury's determination." Do you agree with that statement, sir?
Well, when you testified--one moment. When you testified on May 20th, just a little more than a month ago, in Raleigh, North Carolina, and you were asked about that paragraph, when you said, "Yes, I agree," you never said, "Only with the second half of the paragraph," did you?
What I said, if you read there, I said yes, I agree, and then I went on to qualify what I meant by combining error rates, so in the context of that conversation, it was clear to me at that time counsel was asking about combining error rates and coincidental identity rates. He had me agree with that paragraph, and then as you see, I immediately came back and quoted this previous sentence that they should not be combined, so that is the sense. I don't see any inconsistency.
Sir, what you come back and you say is that: "Coincidental identity in laboratory error are two different phenomenons, so the two should not and cannot be combined in a single estimate"; isn't that true?
But there is nothing in your answer on May 20th under oath where you call into question your agreement with the first half of this paragraph where it talks about how estimates on blind proficiency testing must be disclosed to juries? There is nothing in there about that, is there, sir?
All right. On May 20, 1995, were you asked this question and did you give this answer--
Sustained. As a matter of courtesy, counsel, you have to give a copy to opposing counsel if you are going to do this.
Dr. Weir, when you testified under oath at that deposition on May 20, 1995, were you asked the following questions and did you give the following answer: Question with the questioner quoting from the NRC report.
Sustained. Rephrase the question. You don't get to editorialize. Just state what was asked and answered.
Were you asked the following question and give the following answer, question, quote: "Before the method can be accepted as valid for forensic use, it must be rigorously"--whoops. I'm sorry. I'm sorry. I was on the wrong page.
Mr. Clarke, it is your preference. I can read to line 6 or continue reading to line 13.
Were you asked the following questions and did you give the following answer: "Especially for a technology with high discriminatory power such as DNA typing, laboratory error rates must be continually estimated in blind proficiency testing and must be disclosed to juries. For example, suppose the chance of a match due to two persons having the same pattern were 1 in one million, but the laboratory had made one error in 500 tests. The jury should be told both results. Both facts are relevant to a jury's determination," unquote. "Answer: Yes, I agree. I've already stated that. I think it is a little misleading for you not to read the next sentence, and that is, quote, `a coincidental identity and laboratory error are different phenomena so the two cannot and should not be combined in a single estimate.' "Question: So you would say then that laboratory error rates are crucial to evaluating DNA evidence? "Answer: No, that's--they are talking about the--the rate of errors in proficiency tests. As I said, those are relevant, but I reject the notion of error rates in case work. I don't think that those are relevant, applicable to each case." Were you asked those questions and did you give those answers under oath on May 20 of 1995?
Sure. To your knowledge, Dr. Weir, have any of the witnesses in this case presented that error rate, which you believe is a good thing to do, as expressed in this quotation?
That question does not reflect what you just read out. The error--the rates refer to the frequency with which mistakes are made on proficiency tests. I reject the notion of an error rate applicable to case work.
Sir, do you agree with the portion of the paragraph that says that laboratory error rates must be continually estimated in blind proficiency testing and must be disclosed to juries?
Sir, do you agree that proficiency tests must be truly representative of case work?
Dr. Weir, do you believe that the results of blind proficiency tests that a laboratory undergoes should be presented to juries?
In 1993, Dr. Weir, you wrote an article entitled "Population genetics in the forensic DNA debate," did you not?
Published in 1992. And in that article you discussed the issue of laboratory error rates, do you not?
Not really. That was a review paper and I reviewed a great many things which a lot of people had been saying. At that point a lawyer by the name of Richard Lempert had discussed error rates and given a formula, which if you knew what the chance of making an error was on a continuing basis, you would combine. I didn't agree with his formula and gave the correct formula in the context of reviewing his work. I reject--I reject the foundation, however, of doing that thing.
In that article you said that Professor Lempert or Dr. Lempert provided an equation and you rejected that equation and gave the correct equation?
Could you please show me Dr. Lempert's equation in this article. And I would ask that it be marked for identification, I think it is 1207.
If you want to show that to the witness while we are waiting, if he wants to refamiliarize himself with the article.
I believe you said moment ago, Dr. Weir, that in that 1992 article one of the people you reviewed in there was Dr. Richard Lempert and his approach or formula to giving weight to error rates and that you disagree with that; is that correct?
Well, I'm trying to remember. I wrote that three years ago and at the time I think I disagreed with his formula. At that time I was writing a review. It was not a matter of much concern in the committee. I think Dr. Lempert may have been the first to even raise the issue. Since this time I have thought a great deal more about the issue, so that if I was writing that now, I wouldn't have even given him the credence I did.
I see. Do you recall on Friday, during the cross-examination, Dr. Weir, being asked these questions and giving these answers, referring to page 109 of the transcript. Question by me on cross-examination. "Question: And Professor Richard Lempert was on that committee as well; is that correct? "Answer: Yes, I know Dr. Lempert quite well."
"Question: And he is an expert in the field of statistical inferences and evidence, is he not? "Answer: Yes. He and I share a common belief in the correct way to interpret this data."
Were you asked those questions and did you give those answers to this jury just this last Friday?
Certainly, but that was not talking about error rates. We share a common belief in the likelihood ratio approach.
Sir, in your statement you made some distinction between how one presents this data in terms of error rate and how one presents this data in terms of frequencies?
In arriving at your own opinion, sir, on the role of error rates, did you take into consideration the published writing of Dr. Hagerman?
Authored by you, do not you refer to an equation's handling error rates in a paper written by Dr. Paul Hagerman? And I show you a paper. Beginning here, (Indicating).
What I said was as Hagerman explains, this ratio can be diminished if the typing laboratory may have falsely declared a match. Nothing else that follows refers to Dr. Hagerman's work.
Three years ago I cited his paper, certainly. I don't recall--at this point I don't even recollect reading the paper.
And sir, in Dr. Hagerman's paper that you are referring to, he provides an equation on this very issue, does he not?
Okay. So at least now that you've looked at your own article that you wrote on this subject in December of 1992, does that refresh your recollection as to whether or not you have read a paper by Dr. Paul Hagerman on the issue of the role in assessing DNA evidence in terms of error rates?
Have you read any of the work that Dr. Laurence Mueller has written which refers or describes the error rate issue?
I don't think so. I have read one of Dr. Mueller's papers. I don't recall error rates being in there.
And have you read the publications of Dr. J. Koehler on the issue of error rates in DNA profiling?
Probably. Dr. Koehler and I were at the same conference a couple of years ago and I read the proceedings, so I probably did.
Is Dr. Richard Lewontin is an extremely highly regarded population geneticist at Harvard University?
Have you read their papers where they describe the phenomena of error rates and its role that should be made in assessing DNA evidence?
I don't think so. I think I have read everything that they have written in this field. I don't recall what they say about error rates. And I would point out to you, in this and another context in science, it doesn't matter who says it, it matters what is said, so whether or not they said anything in itself wouldn't give it special credence.
So in other words, the fact that there may be a half dozen articles addressing this very issue which may disagree with your position, is not as important as the reasons put forward for their position; is that correct?
There are several parts to that question. The scientific literature in the feeling of the forensic community does not subscribe to error rates. I know that on the basis of conversations with forensic scientists. In the more general setting, if something is said by a scientist, that doesn't make it true. We need to see all the context, all the argument, get the facts marshaled. I think I have lost the question. I think the answer was yes, though.
Dr. Weir, you just said a moment ago that some of your opinions are from conversations you have had with people who are in the forensic science community; is that correct?
Yes, some of them are, but we talk about these things in formulating our ideas. We don't work in a vacuum.
Didn't you say at the very beginning of your testimony, Dr. Weir, that one doesn't rely, say, on what is said in cocktail conversations or other kind of conversations, but in science one relies on what is actually published, as opposed to mere conversations? Didn't you make a statement like this early in your testimony?
Yes, I said that, and that is when the finesse of words are said and evaluated in the course of writing papers, I talked to other people.
Well, Dr. Weir, in contrast to these different authors who I just asked you about with regard to their publications, what, if anything, have you personally published on the issue of error rates?
Now, Dr. Weir, getting back to the calculations of frequencies in mixtures, I believe on Friday you admitted to this jury that you had made an error which you repeated in your frequency estimates for various stains on the Bronco console and for a stain on the glove; is that correct?
Well, it was worse than I realized. It was both better and worse. I made the same error on every mixed stain involving DQ-Alpha alleles 1.3 and 4. I know how to do that. I had laid out, which is the annoying thing, I had laid out the correct method in my report. I should have applied the same method to all such stains. I did apply the same method to all the stains. Unfortunately for me my program had a mistake, so I was consistent, but consistently wrong, and I have now corrected that.
Dr. Weir, in your original report where you simply list the different pairs of genotypes that could give rise to the 1.1, 1.3 and 4 alleles for some items of evidence, where there are mixtures, you actually put down all the possible pair combinations, assuming on the one hand there is a 1.2 allele, and assuming on the other hand there is no 1.2 allele; isn't that correct?
That's correct. What I just said, though, was the answers I gave were consistent and they do not have those included alleles.
But what I'm saying, sir, is even in your description of the pairings in your report, you describe the pairings one way for stains which included Mr. Simpson and another way in some of the stains which excluded Mr. Simpson; isn't that correct?
Well, that is--that is correct that it doesn't state what in fact I did, and the numbers that went on the board. The numbers on the board were the--were wrong in all the profiles that included 1.3 and 4. I'm going to have to live with that mistake for a long time.
And as you may recall, also on Friday, I asked you in contrast to the numbers that you put on that board for, let's say, for instance, item 29, the steering wheel, under your method and your approach, I then asked you to calculate the frequency of that percentage of the population that could not be excluded as contributors to that mixture using the NRC approach that you were critical of in your report. Do you recall that?
Yes, and I would like the opportunity to explain the difference of those two approaches.
You will have that opportunity, sir, but I just want to try and get through this part as quickly as possible, all right, but there will be plenty of opportunity for that. Now, after you left court on Friday morning, sir, did you go back and confer with any other people about what had happened and what method you had used?
Not--not on the method I used. The method I used is the correct method. There was no need to confer with anybody.
I have had a conversation with Gary Sims on the detail of item 31. He confirmed his testimony and what was presented on the chart, so my analysis of that item was correct.
All right. Did you have any conversations with anyone other than Gary Sims from the laboratories?
Did you have any conversations with the people who are responsible for manufacturing and designing this DQ-Alpha kit?
Okay. Other than speaking with people from the laboratories, did you confer with any Prosecutors over the weekend?
I have met with them. Not--not on what I should do. Not on the details of my analysis, however.
Did you, after you left the witness stand on Friday, until this morning, speak to any other statisticians any other place about this?
All right. Now that you have had the weekend, sir, have you generated new numbers to replace the old numbers on those boards?
Okay. I would like to go through that now, your Honor. Let's start with--do you want to help me get the Bronco board? Let's start with the Bronco board.
Do we have a rag available so the witness can erase the numbers he put on direct examination and replace them now?
All right. Let the record reflect we have all the jurors present. Mr. Neufeld. And you have People's exhibit which up?
Sir, on Thursday afternoon and Friday morning did you write certain numbers on People's exhibit 260 to express certain frequencies?
Okay. Did you quote certain frequencies so Mr. Clarke could write them on the board?
Have you looked at the frequencies that Mr. Clarke wrote on the board when he was simply writing on the board what you were quoting?
No, I couldn't see the board, so I just quoted and I hope he wrote them down as I called them out.
Now, are the frequencies for items 303, 304 and 305, at the bottom of that board, for those three stains, those--are those the numbers that you quoted under direct examination on Thursday afternoon and Friday morning?
Well, I'm sure they are. I can't answer with certainty because I don't have those numbers with me.
Well, could you please check the report to make sure that the numbers that Mr. Clark wrote down on Thursday afternoon and Friday morning are the numbers that you testified to before this jury at that time.
Is there really any dispute as to that, that these are the numbers that were quoted?
Sir, now that you've had the weekend to reanalyze the data, would you agree that those numbers for items 303, 304 and 305 are wrong?
Those numbers--every number with a 1.3 and a 4 in the DQ-Alpha, that is the only--I guess that is the only ones. Well, no, this--this one is correct, so--
Excuse me. 31 is correct. Mr. Sims confirmed that, so on this board it is 303, 304 and 305 have the 1.3 and the 4 which is the combination that confuses things, that hides the 1.2. We don't know if it is there or not. I know we should either assume it is there and/or assume it is not and add those together. I know that, I didn't do it, so that these numbers need to be changed.
Okay. Your Honor, since Mr. Clarke was simply writing down a written record of this man's testimony, I'm asking that these numbers now be replaced.
It is his exhibit, counsel. If you want to have a separate sheet of paper marked 303, 304 and 305, but it is their exhibit, if they object.
And Dr. Weir, I'm going to entitle this exhibit "Mixture frequencies," okay? All right, sir. Now, let's start with items 303, 304 and 305, since that is where we just were, and what I would like you to do first, I'm going to divide this up into three different columns, sir, and the first column we will refer to as your old results, meaning the results that you testified to last week. Is that okay?
And the second column we will call--we will call it "Weir new," reflecting your new results that you arrived at over the weekend. Okay, sir?
And we will do it for a number of different items. The first item we will do it for is evidence item 303, 304 and 305. Okay?
Do you have the data in front of you, sir, that reflects both the old numbers and the new numbers?
All right. Now, for instance, if we look simply at a two-contributor approach to this data and we want to look at the most common frequency in the data--okay?
I believe you testified that for two contributors the most common frequency under the old data that you gave this jury last week was 1 in approximately 1413. 1 in approximately 1413, would that be correct?
That is the precise number. I try not to give any of these figures to--I try not to give more than two significant figures so I will be more comfortable with 1400.
All right. So that is 1 in 1400 for the rareness of that particular profile in the mixture? Okay?
Now, under the new calculations that you did over the weekend, sir, what is the most common probability, or I'm sorry, common frequency for the mixtures in 303, 304 and 305?
Okay. And so would you agree, sir, that the difference from the old approach to the new approach means that it is approximately two and a half times more common in the new approach than it was in your old testimony last week; is that correct?
Okay. The next one I would like to ask you about, sir, is G10, stain on the glove.
Mr. Neufeld, I have a black--I would really prefer you not use the dry erase markers because I have a feeling there is a lack of permanence in there. I have a regular black. I'm just concerned that it might disappear if you use that dry erase stuff.
Now, sir, looking at the glove stain on G10 which also, according to the Prosecution witnesses, also includes and cannot exclude Mr. Simpson as one of the contributors, did you arrive at a most common frequency for that particular profile using the old data that you testified to before this jury last Thursday and Friday?
Okay. Now, sir, that you have recalculated these numbers, having seen your error, over the weekend, what is the most common probability for this particular profile under your technique, given the new data?
And again, sir, would you agree that given your new calculations, the new calculations make this profile approximately two and a half times more common than in your old calculations?
Sir, now, I would like you to look at item 29, which is the stain on the steering wheel, and under your old calculations for a two-contributor assumption, what was the most common profile?
And am I correct in saying that on the old data, or your old calculations, I should say, that you put down the most common frequency for this mixture is 1 in 59, not 1 in 70?
In fact, under your new calculations what is the most common frequency for item 29?
So sir--and for that one you would agree that it is at least two times more common, given your new calculations, as it was in your old calculations; is that correct?
And these are--these are all stains where the Prosecution says Mr. Simpson cannot be excluded, correct?
I'm thinking of the word. I think that's right, isn't it? I'm just looking at my table. Yes, Mr. Simpson is not excluded, that's correct.
Okay. But each of your new calculations makes those profiles that you observed in the mixture certainly significantly less rare than they were under your initial calculations; isn't that correct, sir?
Now, sir, I asked you, as you may recall, to do some calculations on the frequencies of these mixtures relying instead not on your approach, but on the--that NRC method of which you are critical. Do you recall that?
And sir, when we looked at the Caucasian, African American and Hispanic databases using that NRC approach, which you were critical of--
Well, it is not me that is critical of it. It is wrong and it is contrary to the published literature.
Well, it is not my position. It is wrong and it is not the published way of doing things.
You said a moment ago, sir, that you yourself--well, withdrawn. But under that method, when we calculated the most common frequencies using the three databases for item 29, because that was the only one we did it for last week--do you recall that?
Yes. We did a couple of different things. You better remind me which one you are talking about.
Well, we did a series of calculations to show what percentage of the population could not be excluded as having contributed to the mixture which is a profile described in item 29. Do you recall that?
Yes. I want it to be clear that we are talking--are you talking about genotypes or alleles at this point?
Okay. And when we did that, sir, the most common sum, using that NRC method that you believe is wrong, was approximately 1 in 2 people; isn't that correct?
I wouldn't even dignify that number by putting it on the same piece of paper. It is talking about something so different that when you lay it along side, you are given the impression we are talking about the same thing. It is--it answers a completely different question.
Excuse me. Excuse me. That is not an opinion; that is a different--completely different question. It is irrelevant and I would very much like to take a minute and explain why it is irrelevant.
You will have more than ample opportunity, sir. Right now what I'm asking you is this: Did you some calculations on Friday morning--
You made some calculations on Friday morning on the frequency of that percentage of the population that would not be excluded as having been a contributor to the mixture found on item 29; is that correct?
Were any of the numbers that I reported on that piece of paper inconsistent with the numbers that appeared on the Cellmark databases?
Thank you. Sir, and when you summed those frequencies, was the number--was the most common aggregate of those frequencies for item 29 using that NRC method that you say is wrong--
Okay. But using that NRC method, sir, is the number that was arrived at approximately 1 in 2 people?
Your Honor, I'm going to do some more calculations. Perhaps this would be a good time to break for lunch.
All right. We will take our recess at this point. Ladies and gentlemen, please remember all my admonitions to you. Don't discuss this case among yourselves, form any opinions about the case, conduct any deliberations until the matter has been submitted to you or allow anybody to communicate with you. We will be in recess until 1:00 P.M.
I woke up this morning at 5:00 and thought how can I possibly avoid embarrassing myself again in Court, and I thought I will do every conceivable thing to make these frequencies as conservative as possible.
I would be happy to flip a coin. I don't think it is going to make any difference to the calculations how a person is classified in the database.
I have not, I could not and I should not.
In fact, the last sentence of that paragraph, Dr. Weir, is: 'However, both,' referring to laboratory error rate, and those population frequencies that you have given this jury, 'Should be considered'; isn't that correct?
Well, as I have explained, it is a little wrinkle here, but if I had kept the same number of bands, the numbers would still have been different.