JUSTICE ABRAHAMSON: Thanks. So let's do the public comment because that's what we agreed in the agenda, and then we'll come back to this report, and we had someone who wanted to make a public comment.
DR. PROSNITZ: My name is Don Prosnitz. I work for the Department of Justice. My question is assuming that -- and I'm not -- but assuming I were director of a forensics lab, where should I
put my investment? What should I do for my people so that I'll be ready for some of the things you see coming along? You listed all sorts of things that might happen. How would you prioritize them in terms of where I should put my effort now so I'm ready?
DR. CROW: Well, the report does a little better than random with respect to this part becausewe do try and make predictions two years from now, five years from now, and ten years from now as to what things are likely to happen. I don't think we ought to say which are most important. I think that's a value judgment.
JUSTICE ABRAHAMSON: Let me -- is there any other public comment, questions?
(No response.)
JUSTICE ABRAHAMSON: Hearing none, we will proceed with the commission, but if somebody from the public would like to make a comment, I will recognize them, so just stand or raise your hand. But we'll go back to the commission. Jeff?
MR. THOMA: Just very briefly. First of all, you never cease to amaze me. This is from my reading a remarkable document, and your humility in not mentioning that some of your work in the past has got to be referenced --
DR. CROW: Oh, I mentioned all of my work that's relevant.
MR. THOMA: Okay. But you didn't speak of it. But I just had one question that you may not have the answer to, which would be remarkable in and of itself. What is the likelihood of identical twins?
DR. CROW: They're about 1 in 70 births, something like that. One in 70 births is a twin birth, which means 1 in half of 70 persons is a twin.
MR. THOMA: But identical twins?
DR. CROW: No. Identical twins are about 25 percent of all twins.
MR. THOMA: Thank you.
JUSTICE ABRAHAMSON: We were commenting, Jim, that we didn't think it was that high. One out of 70?
DR. CROW: It's higher now than it used to be, incidentally.
JUSTICE ABRAHAMSON: Because of fertility drugs?
DR. CROW: Yes. Because of fertility drugs, and I'm quoting, naturally, at my age, old data.
MR. SMITH: That doesn't mean that Jeff has an identical twin maybe part of the time or anything?
JUDGE REINSTEIN: I'd like to find out how Jan and Paul feel and then also you, Dr. Crow, about Dr. Smith and Dr. Budowle's comments in regard to the draft on page 9, the second paragraph in f, that particular area about whether it would be relatively easier in the near future to have 20 or more loci and whether it will eventually be like fingerprinting; that type of comment.
MS. BASHINSKI: Well, I think that he's -- what he's saying is accurate, but there may be another way tophrase some of these things so that they can't be misinterpreted as meaning a mandate to do additional things. I don't believe the approach that the FBI is taking -- and correct me if I'm wrong, Paul -- has been widely accepted as yet among forensic laboratories, although it may well come to be.
So I would recommend, for example, the day when DNA profiles are regarded as fingerprints is not here yet. You may say may soon arrive, or some -- say that in a way that doesn't appear to denigrate --
DR. CROW: I guess if there are people who think that 13 loci is just as good as fingerprints we'd better at least accept that as a possible viewpoint and --
MS. BASHINSKI: Yes. Rather than just saying it isn't. And the same thing with the number of --
DR. CROW: I'd be very happy to make that kind of a change.
MS. BASHINSKI: -- loci going to 20. Again, I agree with you. I think there will be things added from different laboratories, but maybe the implication here is that you need to go to 20 andmaybe that could be adjusted.The other thing that I picked up on was on page --
JUDGE REINSTEIN: where you say some such criterion -- this is the FBI procedure -- be accepted by the legal and forensic community not as scientifically proper. I know what you mean by that, but someone like Jeff might well turn that around and say, This is not --
DR. CROW: Not scientific.
MS. BASHINSKI: -- not scientifically proper. So I think a little more attention to the way in which the phrases are used would probably be in order. But I don't -- I know our organization does not use their particular approach as yet. It may at some point in time. So I think there is not universal agreement or acceptance that we're there yet with regard to DNA --
DR. CROW: Let me give a partial answer to that. In writing the 1996 report we were very conscious of the fact that the 1992 report led itself to creative interpretations, highly creative, I might add, and we tried to -- and we tried very hard to make the statements in there such that no matter how you tried, you couldn't misinterpret them. Well, I don't think anybody ever succeeded in that, but it's worth the effort. And I think we should take the same view here, and that's one reason for vetting this with you people now. You're more conscious of possible misinterpretations than I am likely to be.
JUSTICE ABRAHAMSON: Would you say not as scientifically precise rather than proper?
DR. CROW: Yes, or something like that.
MS. BASHINSKI: I might even just remove that phrase. Some such criterion will be accepted as a practicaldefinition for forensic purposes, and just not even deal with whether it is or isn't going to be acceptable by the statisticians.
DR. CROW: Yes.
MS. BASHINSKI: That's just --
DR. CROW: That's self-evident?
JUSTICE ABRAHAMSON: Those are good comments. This is on the new handout at 8, and it's line 23.
MS. BASHINSKI: I'd be interested in whatever, Stan, your agency is taking with regard to the --
JUSTICE ABRAHAMSON: Paul?
DR. FERRARA: To follow up on your question, Judge, and follow up on something Jan said, I'm uncomfortable with a comparison of DNA and fingerprints, because they're so different in so many respects, and I don't know really how to articulate it. I mean I'm not telling anybody anything that they don't know already. For example, we have no elegant formulae for the quantification of a fingerprint match.
With respect to the use of 13 loci degree of specificity or precision and accuracy of an individual, it's hard to compare because are we talking about a full comparison on ink print to ink print with all of the resultant points of identification?
One thing I did -- so in a way I'd like this to suggest that even today, the DNA technology in many respects is far superior to fingerprints, even up to and including the point of identification with the difference of the identical twins, of course, whose fingerprints are distinguishable but not obviously the DNA. Another thing that almost needs to be pointed out is when we talk about
database searches with DNA profiles, you search a profile of 13 loci, any good size data bank you come up with one match.
Latent fingerprint -- the search algorithm of automated fingerprint identification systems is such that it's only 60 percent accurate, so again, there are so many aspects to the DNA, which is much more precise -- I don't know how to get that message across. But I'm uncomfortable with thatincreasing the number of loci eventually will become like fingerprinting, because I think it's --
DR. CROW: Yes. I take that. I think that was --
DR. FERRARA: Do you know what I mean?
MS. BASHINSKI: And yet really, what you're trying to say is that we expect it will be accepted in the same way that fingerprints are.
DR. CROW: Yes.
DR. FERRARA: Absent the 100 years' history that we have with fingerprints that we don't with DNA. Now --
DR. CROW: It's interesting to go back into the history of fingerprints, which I did a little bit. When Francis Galton and his successors were talking about fingerprints, they did a lot of statistics. They don't look too good by present standards --
DR. FERRARA: Right.
DR. CROW: -- but they were very inventive for the times. And then finally it just became so convincing that people forgot the statistics. Well, I think that population geneticists are going to be out of a job in ten years on this issue too, because we won't be needed any longer. The technology will make it --
DR. FERRARA: With respect to the searching of the databases -- and here -- one thing I'm sensitive to. Right now in Virginia we've got a population database of about 120,000 convicted felons. Now the vast majority of them are only done at 8 of the 13 loci. They've got to go back and pick them up.
What we're doing in Virginia when we find a profile at the scene of a crime, no suspect or we've
eliminated the suspect, we report to the law enforcement agency as, quote, an investigative lead, a search of our database of 120,000 individuals resulted in a single match to a single individual, and that information is provided as an investigative lead, no statistics are reported at all except indirectly with respect to the size of the database --
DR. CROW: Uh-huh.
DR. FERRARA: -- and the number of loci search. Now, because it is an investigative lead the law enforcement agency is then armed with that information -- would that combined with other information come up with what is probable cause to get an arrest warrant and get a sample from an individual, and also look at other factors that may be developed during that investigation.
Now, once that individual is identified and a sample taken and a direct comparison to the crime scene evidence, that 13 loci, then you report a random match probability or a likelihood ratio if we're dealing with a suspect. What I'd like this recommendation to -- well, if appropriate, to demonstrate that I don't have to necessarily use different loci in my search of a database and not use those loci when doing a direct comparison of the crime scene material where I use the entire gamut. Does that --
DR. CROW: I don't agree, Paul.
DR. FERRARA: You don't?
DR. CROW: No.
DR. FERRARA: Darn.
DR. CROW: It seems to me that --
DR. FERRARA: Investigative purposes now.
DR. CROW: Yes. Well, you use a 13. You found this guy, and then you test the same 13 again.
DR. FERRARA: Actually, I just search on eight --
DR. CROW: Yes.
DR. FERRARA: -- but I theoretically it could work with 13.
DR. CROW: Well, one practical consideration, with13 loci the probability of a match -- just random match is about ten to the minus 15. Well, suppose you have a database of 100 -- I mean a million, that's still ten to the minus nine or something like this.
DR. FERRARA: Uh-huh.
DR. CROW: So using this multiplication principle that we advocated in '96 would lead you to the same conclusion here, but I just think it's illogical to say that you do a second test which isn't independent of the first test and treat it as independent. They teach you in elementary logic course you don't test a hypothesis through a division that suggested it.
DR. FERRARA: But if the purpose of the data banks is an investigative tool, even if it doesn't come up with one name -- let's supposed we come up with five names. Actually, automated finger identification searches can come up with 100 names.
DR. CROW: Uh-huh.
DR. FERRARA: It depends on where you want to cut it off. But the point is from an investigative standpoint, not a statistical, we're trying to provide useful leads to the police to develop an investigative suspect. That's why I --
DR. CROW: I think what you wanted to advise them to do -- I'll tell you how to run your show -- is not to use all 13 for investigative purposes --
DR. FERRARA: Which we don't.
DR. CROW: -- and then save some others.
DR. FERRARA: But not use any of the --
DR. CROW: Not use any of the six or the eight that you use. Yes.
DR. FERRARA: So when we get to a point where we have a single amplification at all 16 loci, then I'm really --
DR. CROW: Yes.
MR. ASPLEN: -- hamstringing myself.
DR. CROW: Well, I don't think this is easy, but I'm also assuming that this is the database for the random sample of the population and that a person you found there is no more likely to be a suspect than a person taken randomly off the street.
DR. FERRARA: Well, these are convicted --
DR. CROW: I think -- I tried to say a while ago, if these are convicted felons --
DR. FERRARA: Right.
DR. CROW: -- they have a greater -- because of recidivism, they have a considerably greater chance than a random person off the street, and I think that's the way to look at this.
DR. FERRARA: So that if we articulate in your report that a search of a database of, for example, convicted felons --
DR. CROW: Yes.
DR. FERRARA: -- for investigative purposes it might be sufficient to not even specify some statistical approach --
DR. CROW: Yes.
DR. FERRARA: -- but that upon -- ultimately when a random match probability or a likelihood ratio is reported in a certificate of analysis at trial as a result of a comparison, then of course we use the NRC II formulae.
DR. CROW: Uh-huh.
DR. FERRARA: That's what I guess I'm trying to --
DR. CROW: Did you -- you know, it's hard for me at least to answer this without being a Bazian [phonetic], and yet we don't do Bazian in the statistics, but let me say it that way anyhow.
DR. FERRARA: Yes.
DR. CROW: Suppose you have a likelihood ratio of ten to the minus 14th? The prior probability of convicted felons is maybe 2 or 3 percent. It's whatever the recidivism rate is. Maybe it's 20 percent. Well, that's a very high prior, and that means this is quite dispositive. On the other hand I'd feel differently about it if this were random people from the population.
So I guess we're talking at cross-purposes. You're talking about a database from convicted felons --
DR. FERRARA: Which is presumably --
DR. CROW: At the moment what we have. Yes.
DR. FERRARA: Unless we were to develop forensic DNA databases that extended beyond all convicted felons, misdemeanants --
DR. CROW: Or even records from the Army or something like that.
DR. FERRARA: Let me get into -- now you're talking about more or less the population at random --
DR. CROW: Yes.
DR. FERRARA: -- in which case, you wouldn't object.
DR. CROW: Right. And I ought to emphasize that both the 1992 report and the 1996 report were not considering convicted felons. Nobody thought of that at that time. We didn't, anyhow. We were treating these as if they were a random sample of the whole population. That's really for the future to worry about I think now.
DR. FERRARA: But would it be informative for your report to address convicted felon databases --
DR. CROW: Yes. And I think we should. Right now it has one sentence, but it could do better than that.
DR. FERRARA: Okay.
MR. CLARKE: I need to stop you for a second, because this is going over my head faster than an F-16. I don't understand, in the context that Paul has put it in, which is as an investigative tool, because that's how it works.
DR. CROW: No problem there.
MR. CLARKE: Let's assume his database consists of ten people and he's done all 13 markers on the unsolved evidence and he runs it through his database and sure enough there's a match, and the reality is there's a match because the person who left the sample is in the database.
Paul goes back and does a redraw from this person, because there's obviously probable cause -- abundant probable cause, tests the same 13 markers, comes up with exactly the same results. The evidence matches the profile of the man or woman whose sample has been taken, and then he performs the statistical analysis as usual and comes up with some vastly small probability. I don't understand why that is the -- why that isn't the only relevant statistic.
DR. CROW: Well, there are people who say that. I disagree. These are just not independent. With a database of only ten, it doesn't really matter. Or the database of a million or database of 5 million, then it becomes important. If -- I think the main thing we ought to say -- just to punt and say this is a problem that's going to have to be faced in the future, as we finally begin to have databases based upon the whole population, not just convicted felons.
MS. BASHINSKI: But if source attribution is possible and you can do source attribution with 16 or 20, now you have in fact identified that individual, and therefore the statistical analysis becomes irrelevant.
DR. CROW: The source attribution assumption really is -- it is not a database assumption. When the FBI worked that out, they're thinking of this person as being found by police investigation or something like that.
MS. BASHINSKI: But at some point in your report you say, I know I can now distinguish this person from everyone in the world, because I have taken care of all the possible substructure and everything else, so I've got the sibs I can -- with 20 loci I can -- now you know for a fact, pretty close as you can know anything for a fact, that you aren't going to find another person with that type, and you're still saying you can't use that?
DR. CROW: I specifically did not say database search in the list of things they were considering for assumptions of this.
MS. BASHINSKI: But -- okay.
DR. DAVIS: Let me ask just a quick question to Paul here. Paul, from the investigative standpoint, you're dealing with all sorts of samples which may be attenuated, partially destroyed, what have you, and you come up with something that for investigative purposes might be this person. On the other hand, maybe the sample you've got isn't that good and it turns out that this guy at the time the crime was committed was definitely giving the commencement address atHarvard and it ain't him.
So there may be other elements here that explain why it still must be approached as an investigative tool rather than a definitive tool.
DR. CROW: Amen. I agree --
DR. FERRARA: That's one of the -- I mentioned this before in respect to retention of samples. So as to avoid even the remote likelihood -- and I know this isn't necessary to send the police looking at somebody closelybased on a mixed sample, and as they did in the U.K., they went to the wrong guy and found out everything doesn't add up. Let's run some more loci, boom, eliminate the guy.
In an effort to minimize that, that's why the first thing we do is double check our data bank sample against the original convicted felon to make sure that it's not --
DR. CROW: You want to make sure you didn't make a lab mistake. Sure.
DR. FERRARA: Right. And then I scratched myself, and when I talked to my folks about this and say, Look; how many times do law enforcement agencies, based on any kind of investigative information end up looking closely at somebody who's innocent but in good faith are operating on some principle?
I might add one other complicating matter, and correct me if I'm wrong, CODIS folks. But don't we have to search at a minimum of ten of the 13 loci in order to search on a national level? That means if I have a crime scene DNA profile and I've developed nine of 13 loci, which is pretty significant, I can't search a national system on those nine? I have to search at ten.
Now, that of course then, again -- now I've got all ten loci if I've got them, I've got to use.
DR. CROW: Yes. Well, I don't think the FBI wants my advice on this, but it would be that ten's too many for pure search purposes.
DR. FERRARA: Two or three --
DR. CROW: Or half a dozen maybe, something like -- with a half a dozen, the probability is around one in a few tens of millions.
DR. FERRARA: As it turns out right now we're doing two amplifications typically for each of eight different -- two different multiplexes but eventually will come a time when we're going to -- it will be simultaneously all 16, so you're not going to have much to hold back.
DR. CROW: Well, I think whether the FBI likes it or not, it wouldn't, but I think there are going to be cases where you would want to use more than 13 loci --
DR. FERRARA: Or less --
DR. CROW: -- it's going to be demanded by some attorney or something.
DR. FERRARA: Or have to use less than ten, too.
DR. CROW: Yes.
DR. FERRARA: Especially with mixtures, but that's --
DR. CROW: Yes.
DR. FERRARA: In NRC I also you mentioned about the identification of the brothers -- I' sorry. I'm -- but another point that NRC I came out sort of against searching a database and identifying -- or expressed cautions about identification of a sib from a data bank of convicted felons. And I personally am not so sure that's a good idea, because since our database was developed, then there's been two or three circumstances where we have searched a database, not made a match, but identified what clearly had to be a brother of someone who's a convicted felon. NRC I comes down opposed to it.
DR. CROW: Right. Well, you were [inaudible].
DR. FERRARA: Yes, I know.
DR. CROW: There just weren't very many loci, and 13 loci's a lot different. With 13 loci you can usually identify brothers.
DR. FERRARA: So this draft -- do you think it might be -- what you just said is worth --
DR. CROW: I said he was on the NRC I committee -- full responsibility for everything it said.
DR. FERRARA: I disagreed at the time, and, like with the ceiling principle, was seriously outgunned. But I'd like that aspect of this -- I just --
DR. CROW: Anyhow, what I wanted to say and did is that 13 loci is different than four, and it would be pretty hard with VNTRs to be sure you had brothers versus other
25 kinds of relatives, or even --
DR. FERRARA: But now with 13, is it worth stating more clearly that the 13 do offer this --
DR. CROW: I think we do state it.
DR. FERRARA: -- capability? Now, whether the individual state laws -- up until July 1 of this year, Virginia state law in that situation I just described would preclude me from telling the police -- God forbid, but I've been in this situation twice already -- I'm sorry. We didn't make a hit, period? And I'm sitting there not being able to say to a law enforcement agency, But look at this guy's brother.
DR. CROW: Uh-huh.
DR. FERRARA: Now, July 1 comes -- we got a law change from the last session that will allow me to then make that statement.
DR. CROW: In our report -- it's back in an appendix -- but there is an example of a pair of sibs, and I know they're sibs because they came from the people that we knew that were sibs. And it clearly shows this distinctive pattern that you expect from sibs and they could possibly have been half sibs as far the statistics were concerned, but that's pretty well ruled out. The likelihood ratio for sibs is maybe 1,000 times as large as for half sibs.
Now, not every example will work, but a great majority of the 13 loci ones would permit this.
DR. FERRARA: Is that in the appendix?
DR. CROW: Yes. It's way back in the appendix. Yes. All you have to do is read 50 more pages and you'll --
JUSTICE ABRAHAMSON: Any other comments from the commission?
JUDGE REINSTEIN: Not on that point.
JUSTICE ABRAHAMSON: Okay.
JUDGE REINSTEIN: But on page 9 --
JUSTICE ABRAHAMSON: Of which -- the big one or the little --
JUDGE REINSTEIN: Of the revision we got today.
JUSTICE ABRAHAMSON: Okay. The April 6 one, page 9.
JUDGE REINSTEIN: This is what you've identified as a footnote with the little squiggly brackets?
DR. CROW: What?
JUDGE REINSTEIN: It's what you've identified as a footnote, an example of a situation? Do you see it, line 14 through 22?
DR. CROW: (Perusing documents.) Oh, yes. The British example. Yes.
JUDGE REINSTEIN: Right. I talked to Chris about this before and I had not seen this until I just started reading it. But from what I understand -- and you pointed this out in your remarks -- the difference between the FBI database, which is 13 STR loci and the British one which was six-- it talks about the National Database's report of mistakenly identified an innocent man based on six STR loci.
Should there be an immediate comparison there to what -- that the FBI database has 13 and that it's the database that everything was supposed to do, right, in that case?
DR. CROW: Just exactly what it ought to do. Yes.
MS. BASHINSKI: And if you did the calculation -- you're saying do a parallel calculation --
DR. CROW: With 13 loci --
MS. BASHINSKI: -- for 13 loci. How would you --
DR. CROW: I don't have to do the calculation. I already know the answer. With 13 loci it would be very unprobable in a database of 700,000 to find a match.
JUDGE REINSTEIN: Because the news reports that came out afterward were very negative toward all this and talked about how it had made a mistake, and I forget who it was in the FBI who had said something to the effect that this was mind blowing and the ramifications would be severe, but it was without really going into what the differences between the FBI and the British --
DR. CROW: Right. I think, Ron -- but somebody could verify this -- but I think this has happened several times before in Britain. It just didn't reach the newspapers. They just found the wrong one and checked it out and that was the end of it.
But does anybody know that for sure? Am I right?
JUDGE REINSTEIN: I actually have a copy of one of the articles --
MS. BASHINSKI: -- and they won't tell you how many because they say the don't count them.
DR. CROW: Well, we could probably calculate how many they would find.
MR. SMITH: What is it you say when it happens here? That it shouldn't have or that we expected it all along to happen?
DR. CROW: Well, it depends on how many --
MR. SMITH: Just say one. Next Tuesday it happens here.
DR. CROW: I think you say -- I think it depends on what you'd expect in this particular case. It was either what you expected to find or it was a very unusual finding, or that somebody made a mistake.
MS. BASHINSKI: But the parallel calculation would tell you how much less likely it is.
MR. SMITH: But once it happens, the calculation -- it tells you how less likely it is in terms you're talking about, however relevant. Right?
DR. FERRARA: But I think though -- I don't want to put words in your mouth, Judge Reinstein, but I thought you were simply saying yes. It worked in that in this case they searched a database -- and in fact I have a copy of one article. The DNA profile was submitted to the National Database where a computer match to one of 660,000 previous arrestees whose DNA was one file, but after the suspect provided an alibi, the police asked for a retest.
This time a new technique which examined ten loci and had a one in a billion likelihood was used. The suspect's DNA did not match at all the loci, and he was released.
And I guess that's the point I'm saying with like our search of eight, or the U.K.'s search of six for
investigative purposes, but ultimately when a suspect is developed you use your full battery and the statistics. In this particular case -- it says here that everyone in the U.K. that's ever been convicted on six-point DNA profiling will want to apply to us to have their convictions reviewed.
I don't see how that follows.
But again, there's a difference between how the database is used and for investigative purposes orfor prosecution.
DR. CROW: Well, certainly these people in Britain are smart enough to know that with only six loci they're going to find quite a number of adventitious matches.
DR. FERRARA: They knew they'd been getting a lot of them.
DR. CROW: Right. And I -- although it wasn't in your article, I'm sure that it was the remaining four loci that clarified this, not the same six over again.
DR. FERRARA: Exactly.
JUSTICE ABRAHAMSON: Just to finish up on Ron's point, do you -- I think if I heard him correctly he might want you to add a sentence that says in the United States the database is 13 loci, and therefore the likelihood of this kind of error being made is decreased or something like
that.
JUDGE REINSTEIN: Or even putting out the number what it would have been --
DR. CROW: We could certainly just put out what the number -- what the expected probability is, and improbable things do happen, like perfect bridge hands.
JUSTICE ABRAHAMSON: Well, and as you all have said around the table, you've got the DNA but that you've got to have other -- other evidence in there will make it more likely, like access or the fact that he's across the world at that time.
But anyway, Norm and then Lisa. Go ahead, Norm.
MR. GAHN: Dr. Crow, because I have so much respect for you, and I like you so much, you know, I --
DR. CROW: I don't like that kind of -- here it comes.
(General laughter.)
MR. GAHN: And what's disconcerting -- and I'm really -- I say this from my heart. Come Tuesday I'm picking a jury in Milwaukee, Wisconsin, and it's a case that I filed against a guy eight hours before the statute of limitations expired, and they picked him out of a data bank six-probe RFLP. As I say, filed the criminal complaint and warrant six hours before the statute expired. It's all the evidence I have.
Got a search warrant. Got his blood, redid it, got my statistic. And I'm picking a jury Tuesday, and that's all I'm doing. I've had no other -- this case was six years old.
Are you stating I shouldn't be doing this?
DR. CROW: No.
MR. GAHN: It's the same -- I mean, if you were called as a defense expert in that case, what would you say? What would you say to me?
DR. DAVIS: They're VNTR.
MR. GAHN: They're VNTR. Yes, six. But they're the same ones. We analyzed the exact same ones.
DR. CROW: All right. But anyhow, the probability of a five-locus match with VNTRs is again awfully small, so I probably wouldn't have any worries about this. Unless your database is enormous, something like the reciprocal of this probability -- which it isn't --
MR. GAHN: But you can use -- even though the search in the convicted offender database -- I'm using the same loci, the same probes for it now to go to trial.
DR. CROW: All right. I probably wouldn't do it that way. I'd probably do it the way NRC II suggested doing: divide -- multiply your probability by the size of the database. But it's still going to be small.
But anyhow, let me say one other thing, too.You're talking again about a convicted felondatabase, and I think we just have to say, and should -- but I don't know how to quantify it except to be a base in and take the frequency of recidivism as the prior probability. But the courts I don't think will do that.
MR. THOMA: Norm, I think what Dr. Crow is saying though to Paul originally is you just can't use that to bolster the statistics because they're not -- there's nothing independent about what the second test is. That's all he's saying. He's not saying don't do it. He's not saying one is better than the other.
All he's saying is that is completely dependent on your first test, so there's no change. You don't have higher statistical -- less of a statistical probability running it twice than you would running it once, because it should come up the same every single time.
MR. GAHN: I thought you were talking about independent -- that it wasn't independent -- the analysis you did -- you got the five-six probe match out of the data bank, convicted offender --
DR. CROW: Uh-huh.
MR. GAHN: -- and I thought you were talking about when you re-analyze it that wasn't an independent --
DR. CROW: Well, if your re-analysis simply checks whether you made a mistake -- is that all you mean?
MR. GAHN: No. I took a new sample of blood and took it and reran the tests totally from the start.
DR. CROW: But you ran the same tests?
MR. GAHN: Exact same tests ran.
DR. CROW: I still agree with what I said a while ago on that point. Yes.
MR. SMITH: Do I understand what you're saying is that the probability somehow is reduced because you did the first test?
DR. CROW: No.
MR. SMITH: That's what it sounds like.
DR. CROW: I wouldn't want to say it that way, but let -- I'll state it the other way around, that if I repeatedly sample from a large database, sooner or later I'm going to find the wrong person.
MR. SMITH: Yes. There's some chance that that will happen.
DR. CROW: Yes.
MR. SMITH: But what he's -- never mind.
DR. CROW: With his case I think you go with what you have, and what I would probably do, as I said, is to take your probability, which would ten to the minus 15th, multiply it by the database size and that will bring it down to ten to the minus 12. It's still convincing.
JUSTICE ABRAHAMSON: Are we done with that? A piece of cake. And I did not listen and I will not read this transcript when it comes back in two years. I would suggest you put all these people down as witnesses on your side so that the other --
DR. CROW: I don't think he wants all of us.
JUSTICE ABRAHAMSON: Oh, okay. Just to keep you from the fence -- is this on the same topic? And then we've got Lisa. Go ahead.
MR. WOOLEY: Earlier today I heard from Jennifer Smith, and she basically asked me as a commissioner to leave out tons of your report, and Judge Reinstein focused on specific language she quarreled with and asked the lab managers here if they had problems with it. I was wondering what do the people think about the overall suggestion of leaving out all of that, because she indicated that it would be some sort of apocalyptic problem for people running labs to have that kind of stuff in.
And I'll be honest, I didn't quite see it as being such a big problem, but are we going to -- we were asked as a commission to make that consideration. Are we going to discuss that or are we all of the mind that we're going to leave it in?
JUSTICE ABRAHAMSON: I asked Dr. Smith and Bruce Budowle and Art Eisenberg, as my closing remarks to all the rest of you are, this report, both sections have page numbers and line numbers, and I asked all of them to feel free to mark it up and send it to Jim, or you can dictate something with line numbers and page numbers and language. I think that to the extent that Jim will agree with that, he'll do it and clear it with his committee and then bring it back, and to the extent he disagrees he'll bring that back. And I think that's all we can do.
And if they want to mark up whole pages they can do that and we'll come back and discuss it, but I would really urge all of you, as Jim has, to read it carefully, and if you want a word struck or you want proper made precise, put it in, and that will call it to everybody's attention. That's how we work on the court in looking at a draft, and I think that's a good way of proceeding.
DR. CROW: I don't to take this -- after all, I've got blood, sweat, and tears in this chapter. I hate to take it out. But the more important reason is that although Jennifer Smith said we said a lot of the same things, we said a lot of things that are new in here, too. And it's a little hard to talk about identical twins, for example, without having said something about the rest of the population first.
And even if it's repetitive -- and it is -- I think we should leave it in.
JUSTICE ABRAHAMSON: And you have to assume that people who read this are not that familiar with the background and everything else and they don't come to it with NRC I, NRC II, and everything else, and that you have to have a story that makes sense. So --
MR. WOOLEY: My sentiment was to leave it in. She didn't strike me as being persuasive enough that we should scrap the whole thing.
JUSTICE ABRAHAMSON: And so I do suggest that the draft be looked at carefully and then send in the comments.
DR. DAVIS: Let me make a simple suggestion. Leave it as it is, make up a nice one-sentence disclaimer at the beginning of the cloudy crystal ball of the future, throw that in, and if any defense attorney or anybody ever brings that up again and tries to cram it down somebody's
throat they can always say, You didn't read the first sentence. That's all.
Some simple little thing like that, because to go back and redo the whole draft and back through the committee again is mind-boggling. You'll never finish.
DR. CROW: That's right.
JUSTICE ABRAHAMSON: But I think that --
DR. CROW: My committee's just as contentious as this group. More so if possible.
JUSTICE ABRAHAMSON: -- fine tuning is always in order, and we'll listen to these comments.
Lisa?
DR. FORMAN: I just wanted to say something on the record, because a few things have been said about page 9 that indicate an intention that I don't think the commission means, and we're talking about the six-probe match being a mistake. We're talking about the language in this particular page, which probably needs to be softened somewhat. On page 9, we have the national British DNA database, "Is reported to have mistakenly identified an innocent man."
Well, that's not true. The database identified an appropriate profile and the database -- and we don't know if he was an innocent man or not. We know ultimately that he wasn't involved in thatparticular crime that they were searching for. We know that he was in the database to begin with, so he was there for something.
At any rate -- and Chief Justice mentioned that mistakes will happen. These aren't mistakes. This is exactly how the database is supposed to act. It is supposed to identify profiles and then the investigative process is supposed to identify whether or not that person was the person who was involved in this particular crime.
I think that Professor Smith, when he brings up what will happen when we find an adventitious -- I love that word -- match at 13 markers what will we do? I do think that this particular working group needs to address the idea that should that happen, what is the scientific approach to that? How many more loci would we have to look at before we felt that this was not an adventitious hit. It was an appropriate hit for this particular crime.
So that's actually the question that we want to be considering, not -- the database is not making mistakes. Those genotypes are right, if you haven't mislabeled the tube or anything. So I wanted to bring that up and clarify that these aren't mistakes. These are real.
DR. CROW: I take the point, Lisa. What I wrote is reported to have made a mistake. That's what the New York Times said. I think I'd rather just say take out the word mistake entirely.
DR. FORMAN: He was identified but ultimately shown not to be part of that particular --
DR. CROW: Yes.
DR. FORMAN: But it's not that particular sentence necessarily. It's the idea that we in this report don't want to appear to be giving credence to the notion that it was a mistake.
DR. CROW: Fair enough. Accepted with pleasure.
JUSTICE ABRAHAMSON: And I think that's the kind of thing that's really quite easily done, to take out the word mistakenly and a series of other things, and that's why you have a lot of people read it, because different eyes and minds read different inferences from the same language, and so the more of us who read it and make these kinds of comments, the better.
DR. CROW: Chris reminds me that I maligned the New York Times. I should have said USA Today.
JUSTICE ABRAHAMSON: All right. Anything else to come before the house before we are in recess until tomorrow morning at 9:00 a.m.? The same place. Anything else?
(No response.)
JUSTICE ABRAHAMSON: Okay.
(Whereupon, at 5:20 p.m., the meeting was recessed, to reconvene at 9:00 a.m., the next day, Monday, April 10, 2000.)

Previous	Contents	Next
Back to National Commission Main Page