Note: This paper appears in the Proceedings from the Seventh International Symposium on Human Indentification 12-26 (1997), published by Promega Corp., ISBN 1-882274-58-X.
1. LINGERING ISSUES
1.1. The ceiling principle
1.2. Other methods of calculating random match probabilities
1.3. Match criteria and band shifting
1.4. Error rates
2. EMERGING ISSUES
2.1. PCR-based tests
2.3. Database searches
2.4. Likelihood ratios
The road to judicial acceptance of DNA identification evidence has been rocky. An initial period of enthusiastic acceptance gave way to widespread apprehension that the vanishingly small probabilities said to be associated with a matching type in an unrelated individual were exaggerated (National Research Council, 1996). At the same time that these concerns were being addressed, the technology of DNA typing has been making major strides. Thus, the road to judicial acceptance has been paved with good intentions, contradictory contentions, and new inventions.
As DNA identification enters its second decade of courtroom use, it is appropriate to survey recent legal developments in the use of DNA evidence in criminal cases. The paper focuses primarily on cases decided in the United States in 1996. It reveals that at least some controversies are subsiding, while others are emerging. Section one discusses the lingering issues -- the ceiling principle, idiosyncratic calculations of random match probabilities, match criteria and band shifting, and estimates of error rates due to laboratory mishandling or misanalysis. Section two turns to emerging issues -- PCR-based tests, claims that genotypes are unique, adjustments to random match probabilities due to database searches, presentations of likelihood ratios, and statutes governing the admissibility of DNA evidence. The survey of legal developments with respect to these matter is not exhaustive, but it should convey some sense of the extent and direction of legal activity.
1. LINGERING ISSUES
1.1. The ceiling principle
The method -- known as the ceiling principle -- that estimates the frequency of a DNA profile by combining the largest known allele frequencies from a potpourri of populations dates back to at least Lander (1991). Rumor has it that the National Research Council's first committee on the forensic uses of DNA technology endorsed the ceiling method for computing the frequency of VNTR profiles as a compromise between warring factions. (Anderson, 1992). Because the appeal of the method lies largely in the prudential judgment that it protects the rights of defendants while permitting some use of incriminating DNA evidence, it was roundly condemned by scientists seeking a procedure that could be justified on purely scientific grounds. (Aldhous, 1993). After repeated criticism, one committee member (Eric Lander) reported that the committee did not intend to preclude the use of the usual form of the "product rule," but merely to supplement those estimates with an upper bound that was high enough to accommodate speculation that population structure was a serious problem. (Lander & Budowle, 1994).
Any such intention, however, was not evident in the text of the report or in contemporaneous explanations and defenses of it from committee members, including Lander (1993). Faced with what seemed to be sharply conflicting expert opinions, courts reached various conclusions about the admissibility of the simple product rule, the ceiling version of the product rule, and the combination of the two. (Kaye, 1995). Aided by Lander's later recollections, and impressed with his sudden, public conversion to the view that the population genetics issues are no longer serious obstacles to admissibility, most recent cases have tended to regard concerns over population substructure as passé. An example is Armstead v. State, 673 A.2d 221 (Md. 1996). Michael Armstead was convicted of robbing and sexually assaulting a woman in her home. Among the evidence introduced against him was the testimony of several expert witnesses for the state who "presented both the product rule [1/480,000,000] and ceiling principle [1/800,000] calculations to the jury, explaining the rationale for each method." Id. at 225. Armstead had sought to exclude the former figure as "based on a purportedly outmoded method . . . rather than the newer 'ceiling principle' method, rendering the data so unreliable as to deny him due process." Id. at 224. On appeal, the Maryland Supreme Court deemed the possibility of departures from Hardy-Weinberg and linkage equilibrium to be one of the "significant potential problems in calculating the probability of a random DNA match" (id. at 236), but noted that "although there was significant debate across the country in both the scientific and legal communities . . . , this controversy has largely been resolved by recent scientific studies." Id. at 238. To nail down the point, the court added that "the debate over the product rule essentially ended in 1993, with the announcement in the scientific journal Nature that the 'DNA fingerprinting wars are over.'" Id.
A similar tack was taken in People v. Miller, No. 78011 (Ill. Aug. 2, 1996). Joseph Miller was convicted of a series of murders, robberies, and sexual assaults of prostitutes. Hair, fiber, blood, seminal fluid, and other evidence linked him to the bodies. The DNA statistics were 1/(465 million), 1/(1.1 trillion), 1/(466 billion), and 1/(1 billion). The Illinois Supreme Court observed that "while there has been some controversy over the use of the product rule in calculating the frequency of a DNA match, that controversy appears to be dissipating." Citing Lander and Budowle (1994), the court stated that "[t]he concerns enunciated by Lewontin and Hartl appear not to have been borne out by empirical studies," and it added that "[t]he most recent courts to consider the use of the product rule have concluded that it is a generally accepted statistical method for estimating the frequency of a DNA match."
Likewise, in State v. Copeland, 922 P.2d 1304 (Wash. 1996), a jury convicted William Copeland of murdering and raping a woman who had been beaten, strangled, and stabbed with a knife and a barbeque fork. The case against Copeland included RFLP tests conducted by the FBI, which used the product rule to estimate the probability of a random match to be between 1/(2.8 million) and 1/(3 million) for various databases. At a pretrial hearing on the general acceptance of the product rule, Copeland produced two experts, Dr. Lawrence Mueller and Dr. Seymour Geisser, who questioned whether VNTR alleles and genotypes were in the proportions that would be expected in a homogeneous, randomly mating population. The trial court found these experts to be "uninformed" about the work of researchers who were much more active in studying the population genetics of VNTRs. Id. at n. 5. The Washington Supreme Court agreed with the trial court that the defense testimony "was neither persuasive nor credible." It observed that
Although at one time a scientific dispute existed among qualified scientists, [it] was short-lived. [T]he FBI has collected data from around the world, and one of the most vociferous opponents of use of the product rule has joined with an FBI scientist in declaring that the DNA wars are over.
Neither Anderson, Miller, nor Copeland referred to the May 1996 report of the National Research Council Committee on Forensic DNA Science: An Update, which emphasized that data collected since the 1992 report reveal that there is no need for the ceiling principle and proposed methods more solidly grounded in population genetic theory to arrive at (on average) conservative figures for the probability of a random match. However, at least four other appellate opinions have cited this development.
In State v. Morel, 676 A.2d 1347 (R.I. 1996), the Rhode Island Supreme Court also was persuaded that:
More recently, numerous scientific articles have addressed the issue of population substructuring and concluded that it does not have a significant impact on final calculations of the probability of a match. Two principals in the debate over forensic DNA typing have concluded that there is "no remaining problem that should prevent the full use of DNA evidence in any court." Eric S. Lander and Bruce Budowle, DNA Fingerprinting Dispute Laid To Rest, Nature, vol. 371, Oct. 27, 1994, at 735.
676 A.2d at 1353. The court fortified this rather sweeping claim with the more specific observation that:
In a follow-up study requested by the director of the FBI to resolve the controversy, the NRC reexamined these issues. The NRC's conclusion was that: "The abundance of data in different ethnic groups within the major races and the genetically and statistically sound methods recommended in this report imply that both the ceiling principle and the interim ceiling principle are unnecessary." The Evaluation of Forensic DNA Evidence, at 5-35 (National Research Council 1996).
Id. at 1354. That the Rhode Island court would uphold the convictions of David Morel for a series of sexual assaults despite the attack on the product rule estimate was a foregone conclusion. After all, the three-locus match was said to exclude 99.9% of the population according to the simple product rule, or 99.8% according to the interim ceiling version. Id.
A Florida Court of Appeals reached more provocative conclusions in Clark v. State, 679 So. 2d 321 (Fla. Ct. App. 1996). The defendant was convicted of "lewd assault." The "primary evidence . . . was the result of DNA testing which established, to a disputed degree of likelihood, that the defendant was the father of the victim's child." As the court described this evidence, it rested on the simple product rule. Unlike Anderson, Miller, or Morel, ceiling figures were not presented:
[T]hree nationally recognized experts testified for the state, that DNA match probability calculations rendered under the "product rule" were properly deemed admissible -- without reference to and, indeed, to the exclusion of the so-called "ceiling principle" reflected in a 1992 preliminary report of the National Research Council. . . . We conclude not only that the order below is fully -- indeed, essentially uncontradictedly -- sustained by the evidence, but that the record as well as developments since the hearing, especially the contents of a 1996 report of the National Research Council, The Evaluation of Forensic DNA Evidence -- effectively demonstrates that the "ceiling principle" has no validity whatever and that, contrariwise, product rule calculations are appropriate as a matter of scientific fact and law.
The denigration of the 1992 report as merely "preliminary" is questionable, and the suggestion that the 1996 report demonstrates that the ceiling method "has no validity whatever" cannot be reconciled with that report's recognition that the interim ceiling method is one way to produce an upper bound on the random match probability and its discussion of ways to avoid abuses in ceiling calculations.
The Washington Supreme Court relied on the 1996 report in holding that use of the interim ceiling method could not justify upsetting a conviction. In State v. Jones, 922 P.2d 806 (Wash. 1996), William Jones was convicted of raping and robbing a 77-year-old woman. The state introduced expert testimony of a DNA match with a ceiling probability of 1/(1.2 million) and a simple product rule probability of 1/(8 million). The defendant argued that both calculations rested on methods that were not generally accepted. As to the interim ceiling principle calculation, Dr. Lawrence Mueller testified that it was inconsistent for the 1992 NRC report to advocate an interim measure with arbitrary ceilings of 10% while also proposing research on the actual ceilings in a set of 15-20 populations. The Washington Supreme Court was not impressed with this testimony. It noted that "most of the criticism of the interim ceiling principle is that it is too conservative and that accounting for subpopulations is unnecessary," and it concluded:
Considering the scientific literature, the expert testimony in this case, and the conclusions of other jurisdictions, the overwhelming view is that while the interim ceiling principle is artificially conservative, it has been accepted for what it is, i.e., a method for calculating probabilities which was biased in favor of defendants in order to compensate for any possible effect due to substructuring in human populations.
The court then invoked the 1996 report:
The National Research Council formed a new committee to update and clarify principles concerning population genetics and statistics as applied to DNA evidence. . . . The prepublication report concludes that "the abundance of data in different ethnic groups within the major races and the genetically and statistically sound methods recommended in this report imply that both the ceiling principle and the interim ceiling principle are unnecessary." Prepublication Report at 5-35.
Use of the interim ceiling principle will likely decline nationally following issuance of the final report, and in Washington it will likely be abandoned given our holding in Copeland that use of the product rule is admissible in calculating frequency probabilities. Despite the new committee's rejection of the interim ceiling principle, however, we see no reason to require retrial of defendants solely on the ground that the interim ceiling principle was used to calculate probabilities. Not only has the method been widely accepted, given its admitted limitations, but a defendant who has benefitted from its use it in no position to complain given our approval of the product rule, which is generally less favorable to a defendant.
Like the Washington Supreme Court, the Arizona Supreme Court read the 1996 Report as indicating that it is not reversible error to admit an interim ceiling frequency, but it expressly left open the question of the admissibility of simple product rule computations. Robert Johnson was convicted of raping a storekeeper, and a major part of the evidence was an analysis from the Arizona Department of Public Safety laboratory that concluded that the random match probability, as estimated with the interim ceiling principle, was 1/(312 million). In State v. Johnson, 922 P.2d 294 (Ariz. 1996), the Arizona Supreme Court relied on commentary and news reports in science journals, law review articles, and other cases to decide that the interim ceiling methods was generally accepted as a procedure for arriving at an upper bound on the random match probability. And, although the Arizona court misconstrued the 1992 NRC Report [see Kaye (1997)], it correctly described the 1996 NRC Report:
This report concludes that sufficient data has been gathered to make the conservative approach of the ceiling principles no longer needed. Id. at 5-32. It further concludes that alternative methods, primarily the product rule, are now appropriate. Id. We fail to see any prejudice to a defendant in results produced by a method that, if biased, is biased in the defendant's favor.
These recent cases suggest that the courts will admit a variety of estimates. The simple product rule has gained new life, at least in situations where the class of plausible suspects is as broad as a racial group. It remains to be seen whether courts will accept the affinal method advocated in the 1996 NRC Report for cases in which the class of plausible suspects is largely limited to a subpopulation. Unless persuasive, published criticism of the affinal procedure emerges, computations using the method should be admissible in United States courts.
1.2. Other methods of calculating random match probabilities
Although the simple product rule is regaining legal acceptance and the product rule with ceiling frequencies being understood as a very conservative alternative, parties can be expected to resort to other methods as well. In State v. Johnson, for example, the defendant insisted (in vain) that the interim ceiling method had to be supplemented with a direct count of the number of matching genotypes in the databases assembled by the Arizona Department of Public Safety.
In People v. Dalcollo, 669 N.E.2d 378 (Ill. Ct. App. 1996), the Illinois Court of Appeals confronted an unusual calculation. Although an FBI analyst had testified that the probability of a random match was 1 in 60 million, "defendant called Dr. Pravatchai Boonlayangoor to testify as an expert in forensic DNA analysis." Dr. Boonlayangoor somehow determined that "the probability of a random match could be from 1 in 745 to 1 in 4,212." According to the court, "Dr. Boonlayangoor admitted that his method of calculation was not recommended in the  NRC Report and was even more conservative." In fact, "[h]e also admitted that his method of calculation was used by no other scientist."
The 1996 NRC Report makes an effort to limit efforts to exploit ambiguities in the 1992 report's description of the ceiling principles, but the type of creativity exhibited in Dalcollo may continue to surface when counsel can find nothing better.  A more common device for raising the random match probability a few orders of magnitude is to discard a few probes and then use the ceiling principle. In State v. Marcu, No. A-4962-92T1., 1996 WL 592934 (N.J. App. Div. Oct. 17, 1996), for instance, Lifecodes performed VNTR tests at five loci and reported that the frequency of the profile in the Caucasian population would be 1/(4.7 billion). Dr. William Shields testified for the accused that the interim ceiling version of the product rule gave the figure of 1/(3.4 million), and that if one probe that he considered unreliable were discarded, then the frequency would rise to 1/6899. Dr. Seymour Geisser was willing to go higher still. He testified "that in the absence of better evidence of the statistical independence of the DNA print patterns developed from separate probes, it was inappropriate to combine the statistical probability of a random match of DNA print patterns even by use of the ceiling principle." Disregarding the results of two probes that another of defendant's experts, Dr. Peter D'Eustachio, concluded were unreliable, Dr. Geisser maintained that "the only reliable population frequency statistic to use" was 1/678. Although the court noted that "there seems to be limited, if any, support for Dr. Geisser's view that the probability of random matches of separate DNA print patterns should not be combined at all," the Appellate Division observed that "[t]he defendant remains free to present conflicting expert opinion testimony regarding population frequency calculations." Despite its citations to the 1996 NRC Report, this court apparently concluded that estimates ranging from 1/678 to 1/(4.7 billion) all had a sufficient scientific foundation to be admitted in court.
1.3. Match criteria and band shifting
Some of the early cases with powerful attacks on DNA evidence raised questions about the statistical criteria for declaring a match between two VNTR alleles and corrections for bandshifting. See United States v. Yee, 134 F.R.D. 161 (N.D. Ohio 1991) (rejecting the argument that the FBI's match window was not generally accepted), aff'd sub nom. United States v. Bonds, 12 F.3d 540 (6th Cir. 1993); Caldwell v. State, 393 S.E. 2d 436 (Ga. 1990) (monomorphic probe for bandshifting acceptable); People v. Keene, 156 Misc.2d 108, 591 N.Y.S.2d 733 (Sup. Ct. 1992) (alleged match held inadmissible where Lifecodes failed to use a monomorphic probe closest to the diagnostic probe to correct for band-shifting and experts differed on the general acceptance of this technique). The simplistic argument that wide match windows (windows that are wide enough to keep the probability of a false-negative match for a single allele very low) are too broad (because they have a substantial probability of allowing a false positive for that one allele) continues to be made -- and rejected. See Armstead v. State, 673 A.2d 221 (Md. 1996). For criticism of the argument, see Kaye (1995).
A more subtle argument arose in O'Dell v. Netherland, 95 F.3d 1214 (4th Cir. 1996). Joseph O'Dell was convicted of a murder committed while he was on parole after having been in prison for many prior offenses. After exhausting his appeals and remedies in the Virginia courts, he petitioned for a writ of habeas corpus from a federal court. The United States Court of Appeals for the Fourth Circuit reasoned that under the law governing these petitions, O'Dell was not entitled to post-conviction relief on the basis of arguments not made on the direct appeal of his conviction unless he could establish his innocence. The key to O'Dell's claim of innocence was a DNA test on five-year-old evidence samples. At trial, extensive blood antigen and enzyme tests were introduced, linking the victim's blood to O'Dell. In the post-conviction DNA tests on these samples, however, the VNTRs did not fall within Lifecodes' 1.8% match window. But band shifting was suspected, and additional tests with a monomorphic probe established a match. According to the court of appeals,
[t]he Commonwealth's experts . . . testified that the LifeCodes data demonstrate that the blood on the blue jacket matched Helen Schartner's blood, for two reasons. First, as O'Dell's experts were forced to largely concede, the DNA patterns on the blue jacket fell well within the state laboratory's and the FBI's match criterion of 2.5%. And second, as O'Dell's experts again had to concede, "band shifting" and "partial degradation," had occurred in the samples and could account for the differentials -- which is why LifeCodes performed further tests and ultimately concluded the blue jacket was a "match."
This account suggests that Lifecodes' experts were not content to report that because of technical problems, they could declare neither a match nor an exclusion. The reliance on other laboratories' match windows, which presumably are broader because the measurement error is greater in those laboratories, seems misleading. The defense experts' characterization of the testing as "inconclusive" seems fairer -- and was consistent with the state's position that O'Dell failed to establish his innocence.
The Fourth Circuit also described an attempt to denigrate the 1992 NRC report's uneasiness with corrections for band shifting by dismissing every recommendation in that report as unacceptable to the scientific community: "the Commonwealth's expert [described the] committee [as] issuing 'recommendations,' not 'accepted by the scientific community generally,' and currently under revision because 'over 300 distinguished scientists' petitioned for their modification due to inaccuracies." Id. at 1248. This is hardly a fair description of the work of the committee, which includes many sensible recommendations. If the court's characterization of the testimony about match windows and the 1992 report is accurate, then the willingness to distort matters is not confined to defense experts like the one in Dalcollo. In any event, O'Dell demonstrates that questions about match windows and band shifting still can arise. In particular, the suitability of monomorphic probes to correct for band shifting has yet to be resolved decisively in the case law.
Another dubious argument about match windows was offered in Satcher v. Netherland, No. Civ. A.3:95CV261, 1996 WL 596270 (E.D.Va. Oct. 8, 1996). Michael Satcher had been convicted of rape and murder. He petitioned for habeas corpus relief partly on the ground that because he did not commit the offenses, carrying out his death sentence would be unconstitutional. To support the claim of innocence, in 1995 he had Lifecodes retest evidence samples first analyzed in 1990 by a laboratory in Tidewater, Virginia. He sent the results of the two tests to Dr. Aimee Bakken, "an independent molecular geneticist." The Tidewater laboratory had found a four probe match in 1990, but:
Dr. Bakken determined that, based upon one probe, the results of the Lifecodes test excluded Satcher as the contributor of the DNA taken from [the victim's] clothing. According to Dr. Bakken, the two bands for the D2S44 probe in the Lifecodes test differed respectively by 3.66% and 3.06% from the measurements of the two dark bands that the Tidewater Laboratory demonstrated in 1990, and this fell outside the ±2.5% window of acceptability used by the Tidewater Laboratory.
This too looks like an abuse of the matching procedure, because it uses the 2.5% interval based on intragel measurement error to make an intergel comparison of two samples. The former standard error is bound to be smaller than the latter, raising the risk of a false exclusion for the intergel comparison. Because the choice of a match window entails considerable arbitrariness, however, the view of "five Commonwealth experts discount[ing], as scientifically meaningless, Dr. Bakken's observation" also seems overstated. Given the fact that "the two tests were run 'on different gels, using different molecular weight sizing ladders, run by different individuals in different laboratories at different times,'" the issue is whether "the match data for all four loci between the two laboratories [are] 'remarkably consistent,'" as Dr. Neil Risch put it. If the disparities are modest, as the court's description of the testimony indicates, then "the 1995 Lifecodes test, in fact, confirms the four-probe match found by the 1990 Tidewater test." Unfortunately, the pretense that there is a single "scientifically meaningful" cut-off for declaring a match promotes these tangential disputes about the size of "the" match window.
1.4. Error rates
Disputes over the chance that a reported match is the result of some error in collecting, handling, or analyzing DNA samples have been heard since the early days of forensic DNA work. Some analysts state or imply that false positive errors do not occur. See, e.g., Commonwealth v. Teixeira, 662 N.E.2d 726 (Mass. Ct. App. 1996) ("It may have been scientifically and statistically unsound for [FBI technician John] Quill to claim a laboratory error rate of zero, but that proposition is not so ordinary and obvious that a judge should be held to take judicial notice of it.") However, most observers are less sanguine. In the absence of definite proof that false positives are as rare as the frequency of matching DNA profiles in the general population, some commentators have argued that the frequency estimates should be excluded as unduly misleading and that the jury should hear only a statistic that merges the probability that a suspect's DNA would be reported to match the crime-scene DNA (due to laboratory error) with the probability that a randomly selected person would have the same DNA profile as the crime-scene DNA. (Koehler et al., 1995; Thompson , 1995; cf. Hagerman, 1990; Lempert, 1991).
The argument that the false positive error probability makes the random match probability irrelevant or prejudicial surfaced very clearly in Armstead v. State (discussed in § 1.1). A man convicted of robbery and rape contended on appeal that it was unconstitutional to introduce the random match probability against him because "the magnitude of the laboratory error as compared to the odds of random DNA matching renders the odds of random matching meaningless." As the Maryland Supreme Court described the argument:
Essentially, Petitioner argues that the DNA testing procedure is inaccurate due to lab error in 0.7% of cases. This means that in seven cases out of one thousand, an erroneous match may be found. Given this, he argues, it is improper to say that there is only one chance in 800,000 that the DNA match was "random" because there is at least a 0.7% chance of erroneous matching due to laboratory error.
There are some difficulties with this argument. To say that "there is only one chance in 800,000 that the DNA match was 'random'" is to misconstrue the meaning of the random match probability. Roughly speaking, that probability is the chance of the same genotype existing in an unrelated individual who did not commit the offense. It is not the chance of innocence given that an individual has the incriminating genotype. But this correction of the commonly heard transposition fallacy does not affect the claim that the false-positive probability swamps the random match probability.
One problem with merging or substituting the false positive probability is that accurate estimates of the risk of a false-positive laboratory result are not easily obtained. The 1996 NRC Report, for example, questions whether proficiency testing can provide such an estimate , and Armstead noted that: "The laboratory error rate of 0.7% for Cellmark was based on two errors identified in proficiency tests conducted in 1988. Since that time, the laboratory protocol has been revised to address these problems." Thus, the Maryland court was satisfied a more traditional legal response to the risk of error. Due process is preserved, the court stated, by that fact that "the jury was fully informed of the laboratory error rate and the Petitioner had a full opportunity to address this on cross-examination."
As Armstead may indicate, most courts will continue to allow admission of the random match probability on the theory that it can help the jury evaluate the possibility that the DNA match reflects the true genotypes but is merely coincidental. Whether or not the random match probability is admitted, however, defendants legitimately may seek to establish that laboratory or handling error might account for the evidence against them. As soon as the Maryland Supreme Court relied in Armstead on the proposition that evidence of the probability of a false positive is admissible, the Maryland Court of Appeals encountered a claim that a defendant had been denied the opportunity to engage in vigorous cross-examination as to the risk of such error. The state elicited testimony in Williams v. State, 679 A.2d 1106 (Md. Ct. App. 1996), about "the procedures for PCR testing, and the steps taken to ensure that technicians at Cellmark properly perform the tests." The state's expert "explained that Cellmark technicians undergo blind 'proficiency tests' given by independent forensic associations" and testified "that Cellmark had no errors in any of the PCR proficiency tests." During cross-examination,
defense counsel asked the expert about testing errors and incidents of contamination that occurred apart from the proficiency tests. Specifically, defense counsel queried whether any tests at Cellmark ever had been contaminated from testing solution accidentally spilling over onto a sample and polluting it. Weber acknowledged that she could think of at least one occasion where such spill-over contamination of a sample had occurred while she was performing a test. Defense counsel then sought to ask Weber how often such errors occurred at Cellmark generally, and the State objected on the ground that the question was irrelevant. At a bench conference, the following colloquy ensued:
[DEFENSE COUNSEL]: Your Honor, . . . I think that in order to make an
assessment about the effect of contamination in the laboratory, I am entitled to find out
how often it happened, if it happened very often with this technician, if it was discussed
in the lab, if the lab took steps to deal appropriately with it.
* * *
[STATE'S ATTORNEY]: It does not matter if fifty percent of their cases were contaminated. What matters is this one.
* * *
COURT: Okay. I thought you were going to ask her, because you had some contamination in one instance, does it carry over to all the rest of them. That's what I thought you were going to ask her. Now, you're going back and rehashing the whole thing over again. I mean, I don't mind you asking her ... if the fact that she had a contamination before, does it carry over to everything they do.
[DEFENSE COUNSEL]: [We] also want[ ] to ask her about incidents of contamination in the lab by other technicians. She was allowed to testify about proficiency tests by other technicians. * * * So she ought to be able to be allowed ... to testify to other incidents of contamination.
* * *
[STATE'S ATTORNEY]: Your honor, what is the relevance of other cases?
COURT: Yeah, I --
[STATE'S ATTORNEY]: It's just not relevant.
* * *
[DEFENSE COUNSEL]: Your Honor, there's a more general issue here. When a witness says, I've never made an error, certainly I can ... I don't think I'm confined to simply saying well, what proficiency tests have you done. Obviously, I have to -- I have to be able to cross-examine on the basis --
COURT: Well, what do you want to ask her about the proficiency test?
[DEFENSE COUNSEL]: About how many times these errors have occurred.
COURT: To her.
[DEFENSE COUNSEL]: About how many times contaminations occurred, yes.
COURT: To her.
[STATE'S ATTORNEY]: Only to her.
COURT: She's -- that's what she answered. She's the one that answered.
[STATE'S ATTORNEY]: She can't speak for other technicians anyway.
* * *
[DEFENSE COUNSEL]: Your Honor, with all due respect, she already has testified as to the proficiency testing of all the other technicians at Cellmark. So why shouldn't she be able to be asked about contaminations by other --
COURT: Well, maybe she doesn't know.
[DEFENSE COUNSEL]: Maybe she does though. Let us ask the question.
* * *
[STATE'S ATTORNEY]: Nobody else touched this case, so what difference does it make?
At that point, the trial judge asked Weber whether she knew when other technicians in the lab made mistakes, and Weber responded that she did not. Because the question occurred at the bench conference, the jury did not hear the question or Weber's response. After Weber answered the question, the judge ended the line of inquiry, and defense counsel objected to not being allowed to further pursue the line of questioning.
The state's argument that the extent of contamination with other technicians is necessarily irrelevant to the chance of contamination in the case at bar seems too doctrinaire. If the same conditions in the laboratory that produced or allowed other instances of contamination were present here, then the inquiry is logically relevant. And because the technician already had testified to the proficiency testing record of other technicians, the Court of Appeals determined that "Williams should have been allowed to fully pursue questions regarding testing errors and possible spill-over contamination in the lab." More generally, it seems likely that most courts will give defendants great latitude in adducing statistics or pointing to past instances that might be indicative of laboratory error.
2. EMERGING ISSUES
Litigators have raised questions about match probabilities, laboratory error rates, match windows and band shifting for years, and section 1 has shown that these issues are lingering points of contention. Of course, not every aspect of every topic is fiercely contested in every case, but counsel probe any weak spot they can find (and some that are not so weak). As a result, occasional attacks on all these facets of DNA identification evidence will continue. Even when the courts conclude that the problems are not so serious as to warrant exclusion of the evidence, defendants have the right to identify any and all possible weakness in the evidence, and prosecutors have the duty to present the evidence fairly.
In contrast to arguments about the quality and accuracy of the testing in specific cases, generic questions of the admissibility of evidence in all cases can be expected to be prominent with respect to new DNA tests and statistical procedures that have received less courtroom exposure. This section discusses the admissibility of PCR-based tests, opinions of uniqueness, probabilities related to database searches, and likelihood ratios. It also describes recent litigation on the constitutionality of statutes designed to foster the introduction of DNA evidence.
2.1. PCR-based tests
No state supreme court has held PCR-based testing inadmissible for want of general acceptance or scientific soundness. See State v. Hill, 895 P.2d 1238 (Kan. 1995) (DQA); State v. Moore, 885 P.2d 457 (Mont. 1994) (DQA test admissible under Daubert), overruled on other grounds in State v. Gollehon, 906 P.2d 697 (1995); State v. Moeller, 548 N.W.2d 465 (S.D. 1996) (DQA test admissible under Daubert); Spencer v. Commonwealth, 393 S.E.2d 609, 620-21 (Va. 1990) (DQA); State v. Lyons, No. SC S41261, 1996 WL 585608 (Or. Oct. 11, 1996) (DQA test admissible under a helpfulness standard); State v. Gentry, 888 P.2d 1105 (Wash. 1995) (DQA test admissible under Frye); cf. State v. Harrison, 644 N.E. 2d 1243 (Ind. 1995) (error not to hold Frye hearing on PCR based method). These opinions discuss the need for quality control, and they emphasize (or overemphasize, some might say) the danger of amplifying contaminating DNA in forensic samples, but they conclude that the underlying science and technology is well established and that it is possible to obtain highly probative results.
A few cases display some confusion over the statistical analysis of some PCR-based systems. For example, the Oregon Supreme Court noted that
Applying statistics to PCR test results presents a fundamentally different issue from the controversy involving RFLP testing that was discussed in NRC (1992). Each method includes a substantially different basis for determining population frequencies. The RFLP method is considerably more discriminating than PCR, with the result being that statistics are considerably more significant under the RFLP method. Also, there is a significant difference as to the data base. The frequency data under PCR is but an estimate.
Id. at n.16. Yet, the only "fundamental difference" is that a PCR-based system that uses discrete markers does not involve the complications in "matching" quasi-continuous VNTR markers. The "basis for determining population frequencies" is the same -- applying a population genetics model to allele frequency estimates. The same kind of databases are used for both types of systems, and reported genotype frequencies or random match probabilities are always "but an estimate". As a result, the "statistics" are neither more nor less "significant" for a PCR-based system than for a VNTR-based system. But there is this difference, which the Lyon court seems to be alluding to: a single VNTR locus is usually more discriminating than a single locus in a PCR-based test.
Although most of the cases on the admissibility of PCR-based tests involve the DQA system pioneered in the early days of forensic testing, newer systems such as STRs will be generating court opinions in the coming years. Late this summer, DNA sequencing was introduced for the first time in a United States court, in Tennessee. In State v. Ware, hairs on a 4-year-old girl who was raped and murdered were linked to the defendant's mitochondrial DNA. A harbinger of things to come, Professor William Shields, told the Los Angeles Times that more validation studies are needed. (Los Angeles Time, 1996).
The 1996 NRC Report suggested that we may be entering a period in which courts will have to decide whether experts can give opinions that DNA genotypes are unique. In fact, such opinions have been appearing for some time. In State v. Hummert, 905 P.2d 493 (Ariz. Ct. App. 1994), rev. granted (1995), the evidence included a VNTR match -- but only at three loci. The Arizona Court of Appeals reported that Lawrence Pressley of the FBI testified that such a match meant that "[e]ither you're brothers, identical twins, or that would be a very unique experience." The state's other trial expert, Mary-Claire King, reached a similar conclusion. Although the first NRC committee, on which she served, wrote that opinions of uniqueness based on the limited number of loci then available were unjustified, she testified that "one can, by carefully choosing particular parts of the DNA that vary a lot between people, uniquely identify every person with just a sample of each person's DNA." Id. In a Washington case, "Dr. King went further and stated 'In my view, there's absolutely no doubt those two samples came from the same human being.'" State v. Buckner, 890 P.2d 460 (Wash. 1995). Her testimony prompted appellate courts to reverse the convictions in these cases.
A more easily defended claim of uniqueness came in State v. Bloom, No. C8-95-218, 1996 WL 33092 (Minn. Ct. App. Jan. 30, 1996) (unpublished opinion), on remand from 516 N.W.2d 159 (Minn. 1994). Professor Daniel Hartl testified that a nine-locus match indicated "that to a reasonable degree of scientific certainty, the DNA in the semen found in the carpet sample and in the vaginal swabs came from the same source as the DNA extracted from Bloom's known blood sample." As more and more loci come into play, opinions like Dr. Hartl's may come to replace the current emphasis on random match probabilities. Of course, these opinions must rest on sound scientific reasoning, and some perspectives on establishing uniqueness can be found in the 1996 NRC report.
2.3. Database searches
Databases of DNA from convicted defendants are growing rapidly. (E.g., Barash, 1995). In the United States, searches through state databases have produced "cold hits." (E.g., Fleeman, 1995; Vandendorpe, 1995). When a suspect is identified through a search of a DNA database rather than on the basis of other evidence, is the probability of a random match the appropriate figure to introduce in court?
The 1992 and 1996 NRC Reports offer different approaches to accounting for the search process. The 1992 Report proposes that database searches only be used to provide probable cause for acquiring a DNA sample from the individual picked from the database, that testing proceed at additional loci, and that "only the statistical frequency associated with the additional loci should be presented at trial (to prevent the selection bias that is inherent in searching a databank)." The 1996 Report seems to favor "the simple rule sometimes suggested by forensic scientists: multiply the match probability by the size of the database." However, a careful reading of the 1996 Report reveals that this multiplication is for testing the hypothesis that "the evidence sample was not left by someone whose DNA profile is in the database (or a close relative of such a person)" as opposed to the hypothesis that the named defendant's DNA is in the evidence sample. As to the latter hypothesis, an appendix suggests an adjustment might not be necessary. In any event, the 1992 Report's proposal has been criticized (Balding & Donnelly , 1996) because "in a situation in which exactly one matching individual is found from a database search, the strength of the DNA evidence against that individual is not reduced relative to the setting in which the suspect has been identified on other grounds and subsequently subjected to DNA profiling."
The defendant in State v. Bloom (discussed in § 2.2) was identified from a database search. The court allowed testimony about a random match probability (computed with the ceiling version of the product rule) of 1/(4.6 million). However, the opinion does not state whether this figure pertained to all nine loci, and it does not appear that the defendant raised any objection relating to the selection effect. With the expansion of databases, this issue is certain to be argued in the courts. Given some of the recent statistical literature on the issue, the 1992 NRC approach may come to be seen as unduly restrictive and the 1996 approach as unnecessarily conservative.
2.4. Likelihood ratios
There are alternatives to analyzing the continuous data from gel electrophoresis of VNTR fragments with the matching-and-binning approach that crudely converts the measurements into discrete packages. (See, e.g., Roeder, 1994). These are noted in the 1996 NRC report, but none have surfaced in U.S. courtrooms.
Judges in a very few exceptional cases have considered the admissibility of testimony as to likelihood ratios as opposed to random match probabilities. The only appellate opinion seems to be United States v. Thomas, 43 M.J. 626 (A.F. Ct. Crim. App. Sept. 12, 1995). Stationed in Germany, Air Force Sargent James Thomas was convicted by a court-martial of killing a Filipino girlfriend, dismembering her body, and setting it afire. Shortly before the body was found in a forest, Thomas had rented a car and entered a building that had an ax in its hallway. Dr. Werner Pflug, a serologist and microbiologist at the State Bureau of Criminal Investigation at Stuttgart, analyzed blood samples from the ax and from the rental car following its cleaning. Based on Amp-FLP typing, Dr. Pflug concluded that the samples taken from the victim, from the driver's seat and trunk of the rental car, and from the ax all contained similar DNA fragments.
Thomas provided powerful criticism from members of TWGDAM of the protocol followed by Dr. Flug. Arthur Eisenberg testified that "the Amp-FLP analysis of the Apo-B region relating specifically to the ax and any material derived from the car either by tape lifts or from the trunk must be viewed as uninterpretable, inconclusive, and no criteria either match or non-match can be applied to that analysis." Id. at 645 (dissenting opinion). Mitchell Holland agreed that American and Canadian laboratories would have reported the tests as inconclusive. Nevertheless, with two judges dissenting, the military courts held that the PCR-based testing was admissible.
Dr. Valerie Prenger, of the Baltimore School of Medicine, was qualified as an expert in the field of human population genetics. She examined "a database of just Filipinos and a database made up of a combined population of Germans and Filipinos" and testified, without objection, that, "conservatively, it was 76.5 times more likely that the samples from the ax and the rented car came from the victim than someone else in the Filipino population. Applying a conservative confidence level, it was 842.3 times more likely that the samples tested came from the victim than anyone else in combined population data base of Germans and Filipinos."
Because defendant had not objected to these computations at trial, only "plain error" could warrant a reversal on appeal. Thus, Thomas argued that
it was plain error to permit the introduction of statistics from a combined population study consisting of 77.5% Germans and only 32.5% Filipinos because those figures inflated the likelihood that the victim was the source of the questioned samples by nearly ten times. The general population statistics presented by Dr. Prenger were derived from the PCR Amp-FLP analysis of a combination of 344 German Caucasians and 100 Filipinos which is not representative of the population of Germany. Appellant asserts that combining the Caucasian German database with the Filipino data base inflated the exclusion statistic because the sample did not include Germany's many ethnic groups.
The military appellate court rejected this claim of error without coming to grips with the peculiar database used to estimate allele frequencies in the German population:
Dr. Prenger's qualifications and the methodology she used to reach her conclusions were unchallenged at trial. She used the acceptable mathematical model to establish her conservative probabilities. There was no evidence presented that rebutted her testimony. Her testimony was admissible under Daubert. There was no plain error.
Thus, Thomas illustrates how likelihood ratios might be presented in court, but it does not supply a careful analysis of their admissibility in general.
In People v. Simpson, No. BA097211 (Cal. Super. Ct. 1995), Professor Bruce Weir suggested in a report prepared for the state that the proper analysis of mixed stains required likelihood ratios. The defense objected vehemently, telling the court that "there was no precedent for using likelihood ratios in a criminal case" (Peter Neufeld, Transcript, June 13, 1995) and that "we know of no precedent for the admission of likelihood ratios in a criminal matter" (William Thompson, Transcript, May 15, 1995). Although the trial judge seemed to agree and the prosecution did not press for likelihood ratios, these representations ignore such cases as State v. Skipper, 228 Conn. 610, 637 A.2d 1101 (1994), Davis v. State, 476 N.E.2d 127 (Ind. Ct. App. 1985), State v. Klindt, 389 N.W.2d 670 (Iowa 1986), State v. Spann, 617 A.2d 247 (N.J. 1993), and State v. Jackson, 358 S.E.2d 679 (N.C. 1987). The defense argued also that "likelihood ratios are difficult to understand and are likely to confuse the jury." People v. Simpson, No. BA097211, 1995 WL 126286 (Cal. Super. Ct. Mar. 20, 1995) (Notice of Objections to Testimony Concerning DNA Evidence and Memorandum in Support Thereof). These arguments were never developed or addressed fully. Instead, Dr. Weir testified only to the denominators of the likelihood ratios. Ironically, Simpson's insistence on omitting the numerators resulted in an incomplete presentation that tended to overstate the probative value of the evidence. The 1996 NRC report suggests that the likelihood ratio is particularly appropriate in evaluating mixed stains, and presentations like those in Thomas and the one proposed in Simpson will necessitate deeper analysis by the courts.
A growing number of states have enacted statutes to dispense with the requirement that the proponent of scientific evidence prove that the underlying methodology is generally accepted in the scientific community or is scientifically sound. Statutes providing for the admission of police radar measurements, blood alcohol, breath alcohol, and parentage tests are widespread. In two recent cases, defendants challenged the constitutionality or the interpretation of comparable statutes designed to expedite the admission of DNA evidence in criminal cases.
Satcher v. Netherland, No. Civ. A.3:95CV261, 1996 WL 596270 (E.D.Va. Oct. 8, 1996), involved a Virginia provision that:
In any criminal proceeding, DNA . . . testing shall be deemed to be a reliable scientific technique and the evidence of a DNA profile comparison may be admitted to prove or disprove the identity of any person. This section shall not otherwise limit the introduction of any relevant evidence bearing upon any question at issue before the court. The court shall, regardless of the results of the DNA analysis, if any, consider such other relevant evidence of the identity of the accused as shall be admissible in evidence.
Satcher argued that this statute is "unconstitutionally vague because (1) it does not specify which of the various methods of DNA testing should be deemed reliable and provides no standards regarding who is qualified to run such tests, acceptable testing procedures, or chain of custody; and (2) it impermissibly shifted the burden to prove the reliability of its evidence from the Commonwealth onto Satcher." The federal district court reviewing Satcher's habeas petition, like the Virginia Supreme that previously rejected these attacks, held that the statute need not be as detailed as Satcher demanded. Interpreting the statute as simply dispensing with the need for judicial determination that VNTR testing is "a reliable scientific technique," these courts concluded that the statute "neither creates a presumption nor shifts the burden of proof. It merely creates a rule of evidence and does not determine the guilt of the accused."
A Maryland statute, which is limited explicitly to RFLP testing, provides "that DNA profile evidence is admissible if the proponent: (1) Notifies in writing the other party or parties by mail at least 45 days before any criminal proceeding; and (2) Provides, if requested in writing, the other party or parties at least 30 days before any criminal proceeding with: (i) Duplicates of the actual autoradiographs generated; (ii) The laboratory protocols and procedures; (iii) The identification of each probe utilized; (iv) A statement describing the methodology of measuring fragment size and match criteria; and (v) A statement setting forth the allele frequency and genotype data for the appropriate data base utilized." This statute apparently contemplates the routine admission of random match probabilities. Like the Virginia statute, it specifies that "[i]n any criminal proceeding, the evidence of a DNA profile is admissible to prove or disprove the identity of any person."
In Armstead v. State, 673 A.2d 221 (Md. 1996), a defendant convicted of rape and robbery appealed on the ground that this statute should have been construed to permit him to challenge the admission of DNA evidence as not generally scientifically accepted or as unfairly prejudicial. Armstead, unlike Satcher, did not argue that the statute placed an unconstitutional burden on him. He simply argued that he should have the chance to carry the burden. Conceding that the statute mandated admissibility without proof by the state of general acceptance, he insisted that he should nevertheless have the opportunity to disprove such acceptance or to show that the prejudicial impact of such evidence substantially outweighed its probative value. The court, however, interpreted "admissible" literally, as meaning that properly conducted tests are admissible without regard to general acceptance or prejudice. Furthermore, it held that notwithstanding the possibility of a mistake in any particular case, VNTR tests and simple product rule estimates are sufficiently valid to support the legislature's decision to make properly conducted DNA tests and associated product rule calculations admissible per se.
1. Dr. Boonlayangoor has provided remarkable testimony in other cases. In Franson v. Micelli, 645 N.E. 2d 404 (Ill. Ct. App. 1994), vacated, 172 Ill.2d 352, 355 (1996), the state introuced testimony that VNTR paternity tests using five probes yielded a likelihood ratio of over 29,000,000. The court of appeals held this paternity index inadmissible because of the dispute over substructure in the Caucasian population. According to the court, Dr. Boonlayangoor testified for the defendant that because only one to three probes are used in most cases, the use of an index based on five was misleading. Moreover, he testified that a database should include at least 100,000 subjects. Somehow, he was able to conclude that the five-probe match made it likely that the defendant was not the father of plaintiff's child. 645 N.E.2d at 406. In earlier cases, Dr. Boonlayangoor testified that far smaller paternity indices established that the tested men were fathers. Elder ex rel. Finney v. Finney, 628 N.E.2d 393, 394 (Ill. Ct. App. 1993) (PI = 9440); People ex rel. Forrest v. Winston-Bey, 625 N.E.2d 708 (Ill. Ct. App. 1993) (PI = 477,611). [BACK]
2. Given the difficulties in quantifying the risk of a false positive error, the committee recommends that "[w]henever feasible, forensic samples should be divided into two or more parts at the earliest practicable stage and the unused parts retained to permit additional tests" performed independently at a different laboratory. [BACK]
Aldhous, Peter (1993) Geneticists Attack NRC Report as Scientifically Flawed. Science 259:755-756.
Anderson, Christopher (1992) Academy Approves, Critics Still Cry Foul. Nature 356:552.
Balding, David J. & Donnelly, Peter (1996) Evaluating DNA Profile Evidence When the Suspect is Identified Through a Database Search. J. Forensic Sci. 41:603.
Barbash, Fred (1995) British Authorities Launch 1st National DNA Database, Wash. Post, Apr. 11, at A15.
Fleeman, Michael (1995) First DNA Database "Hit" Leads to Arrest. Orange County Register, Nov. 9, at A22.
Hagerman, Paul J. (1990) DNA Typing in the Forensic Arena. Am. J. Hum. Genetics. 47:876
Kaye, D.H. (1995) The Forensic Debut of the NRC's DNA Report: Population Structure, Ceiling Frequencies and the Need for Numbers, Genetica 96:99-105
Kaye, D.H. (forthcoming 1997) Bible Reading: DNA Evidence in Arizona, Ariz. St. L.J.
Kaye, D.H. (1995) The Relevance of "Matching" DNA: Is the Window Half Open or Half Shut? J. Crim. L. & Criminology 85:676-695.
Koehler, Jonathan J. et al. (1995) The Random Match Probability in DNA Evidence: Irrelevant and Prejudicial? Jurimetrics J. 35:201.
Lempert, Richard (1991) Some Caveats Concerning DNA as Criminal Identification Evidence: With Thanks to the Reverend Bayes. Cardozo L. Rev. 13:303
Lander, Eric S. (1993) DNA Fingerprinting: The NRC Report. Science 260:1221.
Lander, Eric S. (1991) Reply. Am. J. Hum. Genetics 49:899.
Lander, Eric S. & Budowle, Bruce (1994) Commentary: DNA Fingerprinting Dispute Laid to Rest. Nature 371:735-738.
Los Angeles Times (1996), Murder Case Puts New DNA Test on Trial, Aug. 30, at A23.
National Research Council Committee on DNA Forensic Science: An Update (1996), The Evaluation of Forensic DNA Evidence. Washington D.C.:National Academy Press.
National Research Council Committee on DNA Technology in Forensic Science (1992), DNA Technology in Forensic Science. Washington D.C.:National Academy Press.
Roeder, Kathryn (1994) DNA Fingerprinting: A Review of the Controversy. Stat. Sci. 9:222-247.
Thompson, William C. (1993) Lessons from the "DNA War," J. Crim. L. & Criminology 84:22.
Laura Vandendorpe (1995) DNA Detectives, Portland Oregonian. Mar. 8, at B01.
Armstead v. State, 673 A.2d 221 (Md. 1996)
Caldwell v. State, 393 S.E. 2d 436 (Ga. 1990)
Clark v. State, 679 So. 2d 321 (Fla. Ct. App. 1996)
Commonwealth v. Teixeira, 662 N.E.2d 726 (Mass. Ct. App. 1996)
Davis v. State, 476 N.E.2d 127 (Ind. Ct. App. 1985) `
Elder ex rel. Finney v. Finney, 628 N.E.2d 393, 394 (Ill. Ct. App. 1993)
Franson v. Micelli, 645 N.E. 2d 404 (Ill. Ct. App. 1994), vacated, 172 Ill.2d 352, 355 (1996)
O'Dell v. Netherland, 95 F.3d 1214 (4th Cir. 1996)
People v. Dalcollo, 669 N.E.2d 378 (Ill. Ct. App. 1996)
People ex rel. Forrest v. Winston-Bey, 625 N.E.2d 708 (Ill. Ct. App. 1993)
People v. Keene, 591 N.Y.S.2d 733 (Sup. Ct. 1992)
People v. Miller, No. 78011, 1996 WL 435673 (Ill. Aug. 2, 1996)
People v. Simpson, No. BA097211 (Cal. Super. Ct. 1995)
Satcher v. Netherland, No. Civ. A.3:95CV261, 1996 WL 596270 (E.D.Va. Oct. 8, 1996)
Spencer v. Commonwealth, 393 S.E.2d 609 (Va. 1990)
State v. Bloom, No. C8-95-218, 1996 WL 33092 (Minn. Ct. App. Jan. 30, 1996) (unpublished opinion), on remand from 516 N.W.2d 159 (Minn. 1994)
State v. Buckner, 890 P.2d 460 (Wash. 1995)
State v. Copeland, 922 P.2d 1304 (Wash. 1996)
State v. Gentry, 888 P.2d 1105 (Wash. 1995)
State v. Harrison, 644 N.E. 2d 1243 (Ind. 1995)
State v. Hill, 895 P.2d 1238 (Kan. 1995)
State v. Hummert, 905 P.2d 493 (Ariz. Ct. App. 1994), rev. granted (1995)
State v. Jackson, 358 S.E.2d 679 (N.C. 1987)
State v. Johnson, 922 P.2d 294 (Ariz. 1996)
State v. Jones, 922 P.2d 806 (Wash. 1996)
State v. Klindt, 389 N.W.2d 670 (Iowa 1986)
State v. Lyons, No. SC S41261, 1996 WL 585608 (Or. Oct. 11, 1996)
State v. Marcu, No. A-4962-92T1., 1996 WL 592934 (N.J. App. Div. Oct. 17, 1996)
State v. Moeller, 548 N.W.2d 465 (S.D. 1996)
State v. Moore, 885 P.2d 457 (Mont. 1994), overruled on other grounds in State v. Gollehon, 906 P.2d 697 (Mont. 1995)
State v. Morel, 676 A.2d 1347 (R.I. 1996)
State v. Spann, 617 A.2d 247 (N.J. 1993)
United States v. Thomas, 43 M.J. 626 (A.F. Ct. Crim. App. 1995)
United States v. Yee, 134 F.R.D. 161 (N.D. Ohio 1991), aff'd sub nom. United States v. Bonds, 12 F.3d 540 (6th Cir. 1993).
Williams v. State, 679 A.2d 1106 (Md. Ct. App. 1996).
updated 28 Dec. 1999