The PEOPLE, Plaintiff and Respondent, v. Sergio VENEGAS, Defendant and Appellant.
Appellant Sergio Venegas appeals from convictions for rape and other offenses committed on November 2, 1989. The principal issue on this appeal is whether DNA profile evidence was properly admitted at his trial. We reverse and remand the matter.
The victim, E., returned to her room at the Red Lion Inn in Bakersfield just before noon on November 2, 1989. Her arms were full. She placed some items on the hallway floor, unlocked the door to her room and entered the room with the remaining items still in her arms. As she came back to the open door, a man in light blue or white jeans, a white T-shirt and no hat pushed his way into the room. He told her he had a knife and not to look at him. The man was taller than E., heavily built, and appeared to be Hispanic. He had a “kind of Spanish, Hispanic” accent.
The man ordered E. to orally copulate him. He then raped her. He bound and gagged her with cloth and electrical cords, digitally penetrated her vagina on two occasions and cut her with a knife in several places. He searched the room and took a small amount of money from E.'s purse. About 2 p.m., he left. E. worked free from the gag and called for help. Motel employees summoned police.
E. was taken to the hospital. Vaginal swabs were preserved. Police also recovered the bed covering from the motel room, which appeared to have semen stains on it.
On November 2, 1989, appellant was at the Red Lion to apply for employment. He met briefly with a personnel clerk and obtained an application. The clerk was about to give new employees a tour of the motel, so she told appellant she would interview him when she returned.
As the personnel clerk toured the facility, she came upon appellant in a remote part of the motel, far from the personnel office but near E.'s room. When she returned to the personnel office, appellant was there. She interviewed him briefly, and he left. It was approximately 11:45 a.m. She said appellant was dressed in light gray or white cotton pants with pockets on the front of the thighs. He wore a jacket and a cap. The personnel manager identified appellant at trial.
E. did not remember the face of her attacker; she could neither identify nor exclude appellant. Upon viewing appellant in the courtroom, she said he was what she meant when she described the assailant as “stocky.”
A few days after the rape, police contacted appellant. They told him they had information he was involved in the rape and asked him to provide hair, saliva and blood samples. He initially declined, but the next day gave hair and saliva samples. He denied culpability and declined to give a blood sample. Police eventually obtained a search warrant for a blood sample, at which point appellant cooperated in providing the sample.
A Kern County Crime Lab criminalist compared a hair specimen recovered from E.'s body with a sample of appellant's pubic hair. Although there were some differences between the two hairs, he was unable to exclude appellant as the donor of the specimen.
Another crime lab criminalist tested appellant's blood, one of the vaginal swabs, and scrapings from the bed stain. Although she was unable to obtain a blood type from the vaginal swab, she was able to test the stain and appellant's blood. The results excluded one-half of the male Hispanic population as the possible source of the bed stain, but did not exclude appellant.
A Federal Bureau of Investigation (FBI) criminalist tested appellant's blood, the vaginal swabs and the bed stain to establish the DNA profiles of the samples.1 For three of the tests she performed, appellant's DNA matched DNA in the two crime-scene samples. A fourth test was inconclusive, but did not exclude a match. The criminalist performed a statistical analysis of the match information, resulting in an estimate of appellant's DNA characteristics as shared by one person out of 65,000 in the general population, and one out of 30,000 in the southwestern United States Hispanic population.
A defense genetics expert evaluated the FBI DNA information. He concluded that the data supported a conclusion that appellant's DNA characteristics were far more common than claimed by the prosecution. Depending on his choice of methodology, the defense expert estimated that one out of 35 or one out of 378 persons shared appellant's DNA profile.
After deliberating less than two hours, the jury on November 16, 1992, convicted appellant of all charges and found true all enhancements, as follows:
Count 1, forcible oral copulation (Pen.Code, § 288a, subd. (c)); 2
Count 2, forcible rape (§ 261, subd. 2);
Count 3, genital penetration with a foreign object (§ 289, subd. (a));
Count 4, genital penetration with a foreign object (§ 289, subd. (a));
Count 5, assault with a deadly weapon (§ 245, subd. (a)(1));
Count 6, residential robbery (§ 212.5, subd. (a)).
As to Counts 1–4, the jury found true weapon use allegations (§ 12022.3, subd. (a)); as to counts 1, 2 and 6 the jury found true great bodily injury allegations (§ 12022.7).
On November 17, 1992, appellant admitted a prior prison term enhancement as to all six counts. (§ 667.5, subd. (b).)
After denying appellant's new trial motion, the court sentenced appellant to a prison term of 65 years.
Appellant contends the prosecution failed to establish a general scientific consensus concerning the reliability of tests used to match his DNA with the DNA profile of the crime-scene samples. (See People v. Leahy (1994) 8 Cal.4th 587, 604, 34 Cal.Rptr.2d 663, 882 P.2d 321.) He also claims no such consensus was established concerning the appropriate method for the statistical analysis of DNA profiles. While not fully developed in the briefing, appellant's contentions have merit. Appellant's objections at the Kelly stage in the trial court were comprehensive, and the record adequately permits review of the contentions on appeal.
The Scientific Background
The published California appellate opinions concerning DNA admissibility contain exhaustive descriptions of DNA itself and of the process by which it is isolated and profiled. We will do our best to succinctly summarize the DNA story, finding it unnecessary to repeat many of the details previously set forth. (See People v. Wallace (1993) 14 Cal.App.4th 651, 17 Cal.Rptr.2d 721; People v. Pizarro (1992) 10 Cal.App.4th 57, 12 Cal.Rptr.2d 436; People v. Barney (1992) 8 Cal.App.4th 798, 10 Cal.Rptr.2d 731; People v. Axell (1991) 235 Cal.App.3d 836, 1 Cal.Rptr.2d 411.) We note also that a summary of the scientific aspects of DNA profiling is found in the report of the National Research Council, DNA Technology in Forensic Science (1992) (hereafter NRC report).3
Enormous numbers of genes appear on the 23 pairs of chromosomes in all nucleated human cells, including blood, saliva and sperm cells. (Pizarro, supra, 10 Cal.App.4th at p. 69, 12 Cal.Rptr.2d 436.) Genes for particular traits are located at the same chromosome sites in all humans. (Ibid.)
The molecular component of genes is DNA, a large molecule composed, in part, of patterns of pairs of four chemical compounds, or “bases.” (Pizarro, supra, 10 Cal.App.4th at p. 68, 12 Cal.Rptr.2d 436.) The overwhelming portion of these patterns of chemical bases is the same in all humans; they are the genetic code for features shared by most humans, such as arms, fingers, internal organs, etc. Other patterns of these chemical base pairs, located at known sites on particular pairs of chromosomes, vary from one person to another, and are responsible for genetic differences between people, such as eye or hair color. The base pairings at these sites vary within a limited range of possibilities, reflected, for example, by the limited range of (natural) human hair colors. (See id. at p. 69, 12 Cal.Rptr.2d 436.)
At still a third kind of site, base pairs occur in hundreds of different patterns, the purpose or effect of which variation is not known. It is these extremely polymorphic sites that are the subject of forensic DNA analysis. (Pizarro, supra, 10 Cal.App.4th at p. 69, 12 Cal.Rptr.2d 436.)
At the time of the proceedings herein, the FBI,4 by a complex chemical and physical process (see Barney, supra, 8 Cal.App.4th at pp. 806–807, 10 Cal.Rptr.2d 731), was able to isolate four pairs of polymorphic sites; each pair is on a different chromosome pair. (Pizarro, supra, 10 Cal.App.4th at p. 71, 12 Cal.Rptr.2d 436.) The FBI process breaks down DNA at these sites into segments reflecting the length of the repeating base-pair patterns, which can be anywhere from a few dozen base pairs to more than 10,000 pairs. The segments for each site are isolated and displayed according to their length, expressed in number of base pairs. The FBI, at the times relevant herein, was able to analyze segments ranging from 640 base pairs to 10,000 base pairs. Other segment lengths were disregarded.
The DNA base-pair segments are made radioactive and are placed under X-ray film for several days, creating a permanent visual record of the size of the segments. This film record is called an autoradiograph, or “autorad.” (Pizarro, supra, 10 Cal.App.4th at p. 70, 12 Cal.Rptr.2d 436.)
In the typical forensic setting, processed samples of DNA recovered from a crime scene are placed in the same testing medium with DNA samples from the victim and the suspect, as well as control samples of known base-pair length. Thus each forensic autorad, with certain variations that do not concern us here, consists of a visual display of the base pairs (bands) from a particular site on the DNA of a particular chromosome from each of the donor sources. (Barney, supra, 8 Cal.App.4th at p. 808, 10 Cal.Rptr.2d 731.) The bands are compared by the criminologist and analyzed by computer to determine whether the suspect's base-pair patterns are present in the crime-scene sample. (Id. at pp. 808–809, 10 Cal.Rptr.2d 731.).
As indicated, the FBI repeats this process for each of four sites, each on a different chromosome. The result is four separate autorads, each showing the base-pair bands for a given site on the same numbered pair of chromosomes from the crime-scene sample, the victim, the suspect, and the control DNA. There are two bands from each donor on each autorad, and, thus, eight sites (two from each of the four chromosome pairs) where the crime-scene sample may match the suspect's DNA. (Barney, supra, 8 Cal.App.4th at p. 808, 10 Cal.Rptr.2d 731.)
Thus, the first step in this system of DNA profile analysis, known as RFLP analysis,5 is the determination whether a suspect's DNA profile matches a sample recovered from a crime scene. If any of the suspect's bands are not present in the autorad of the crime-scene sample, the suspect may be excluded as the source of the questioned sample, barring technical deficiencies in the test that render the questioned sample inconclusive for a given band.6
If, on the other hand, all of the identifiable bands from the suspect's sample are also found in the questioned sample, the criminalist may be able to declare that there is a match between the two samples. In the present case, for the three autorads and their six measurable bands, the evidence showed there was a match.
The second step in DNA profile analysis is the statistical component. The benefit of using multivariable polymorphic sites for DNA profiling is that the likelihood of a coincidental match is rare. Quantifying the degree of rarity of a coincidental match is the goal of this second step in forensic DNA profiling.
Many of the laboratories that are able to do DNA analysis have compiled their own data bases reflecting the array of band sizes from a sample population. The FBI, for example, had a data base of less than 3,000 blood samples as of the time of trial in this case. These are grouped into Caucasian, Black, southeastern Hispanic and southwestern Hispanic data bases. The southwestern Hispanic data base was composed of approximately 450 samples.
Although there may be many persons with a particular band length at a particular gene site, the scientific literature says there is no known instance of two unrelated persons with an identical pattern of band lengths across four gene sites. (Identical twins have identical DNA profiles.) Accordingly, a researcher cannot simply turn to the data base and, by counting, determine the degree of rarity of the suspect's DNA profile. The data bases simply are not large enough to show any repeats, nor whether repeats would ever occur. (NRC report at 3–1.)
Scientifically accepted statistical theory, however, permits a calculation of the rarity of a suspect's DNA profile based upon the frequency in the data base of each of the six or eight matching base pairs comprising the suspect's profile. If the occurrence of a given base pair is independent of the remaining base pairs (unlike, for example, those determining black hair and brown eyes) (NRC report at p. 3–3), the frequency of a DNA profile can be calculated as 2(a x a′) x 2(b x b′) x2(c x c′), where a and a′ are frequency of the bands from one chromosome pair, etc., within a given data base. (The frequency is expressed as a decimal where 1.00 represents 100 percent.) This calculation of frequency employs what is known as the product rule. (Pizarro, supra, 10 Cal.App.4th at p. 70, 12 Cal.Rptr.2d 436.)
That simple calculation, in addition to assuming independence of the bands, makes several other assumptions. First, it assumes all bands are measured exactly. Second, it assumes the population from which the data base was drawn is homogenous, that is, that any Caucasian is equally likely to have mated with any other Caucasian, but not with a Black or an Hispanic. Third, it assumes the suspect and the actual perpetrator are members of the same data-base population. (See generally, NRC report at ch. 3.)
There is general agreement that the first two assumptions are not true, and that the truth of the third assumption may or may not be known in any given case. There is general agreement that something must be done in the statistical analysis to account for the uncertainty introduced by the three assumptions. The question becomes whether there is a generally prevailing scientific consensus of what should be done.
The evidence shows that the FBI resolved these uncertainties to its own satisfaction, and to the satisfaction of part of the scientific community, in certain ways believed to be conservative, i.e., making any given DNA profile appear more frequent in the relevant population, and hence less inculpatory of the suspect. Others in the scientific community disagreed with some elements of the FBI's resolution, and put forth other proposals to resolve the uncertainties still more conservatively. (See NRC report, ch. 3.) This dispute is recounted in some detail in People v. Barney, supra, 8 Cal.App.4th at 814–816, 10 Cal.Rptr.2d 731.
The 1992 NRC report is an effort to work toward a consensus on methodologies to account for the above-described uncertainties in DNA statistical analysis. The report is based on a study undertaken by the Committee on DNA Technology in Forensic Science, a committee of the National Research Council, which in turn is the principal operating agency of the National Academy of Sciences. (NRC report at pp. vii, ix; see Barney, supra, 8 Cal.App.4th at p. 816, 10 Cal.Rptr.2d 731.) This report made four recommendations that are germane to the present case. We summarize these briefly.
First, in order to determine whether the absence of completely homogenous mating patterns within a population results in statistically different subpopulations, 15 to 20 ethnically different populations should be sampled, with the results compiled in separate data bases.7 (NRC report at p. 5–13.) In order to account for the possibility of still further subpopulations un- or under-represented in the data base, for any band that results in a frequency of less than 5 percent the NRC report recommends use of a value of 5 percent. (Id. at p. 5–14.) (This modification of the calculation is known generically as the floor method.)
Second, for the interim when data bases are available for only three or four general populations, the increased likelihood of the existence of significant subpopulations makes it appropriate to use a floor of 10 percent for any given band-length. (NRC report at p. 5–14.)
Third, in order to account for the fact that an individual's genetic makeup may be composed of contributions from ancestors in differing general populations, the probability calculation should use the highest-frequency result for a given band in any of the various data bases. (NRC report at p. 5–12.) This is known as the “ceiling method.” The ceiling method not only accounts for ethnic diversity in individuals' backgrounds, but also, by reporting the probability of a coincidental match based on a hypothetical “mixed” genetic background, renders irrelevant the assumption that the perpetrator and the suspect are the same race. (NRC report at p. 3–13.)
Fourth, there is a measurement tolerance for samples on the same autorad and on different autorads, and this measurement tolerance varies from one lab to another. The NRC notes that the interautorad tolerance is generally greater than the intraautorad tolerance, so it recommends the match criterion for the data base be this interautorad tolerance. (NRC report at p. 2–12.)
To explain further, the FBI intraautorad tolerance (i.e., for samples examined in the same test) is plus or minus 2.5 percent. (This is known as the “match criterion.”) If a band from the crime-scene sample is measured at 1,468 base pairs, imprecision in the measuring technology makes it possible that the actual DNA segment is composed of anywhere from 1,431 to 1,505 base pairs. If the suspect's DNA measures 1,460 base pairs, it may actually be composed of between 1,424 and 1,496 base pairs. Because there is a range (from 1,431 to 1,496) in which these possible lengths overlap, the bands are considered a match. The FBI averages the measurements of the two matched segments, in our example, at 1,464 base pairs, and then goes to its data base to see how frequently that size segment appears in the population.
But again, the ability to accurately measure must be considered. Because the samples in the data base were each measured in a different test, the tolerance range is greater. The prosecution experts did not establish this tolerance for the FBI lab, but a defense expert testified it was plus or minus 4.5 percent (or a 9 percent tolerance). According to the NRC report (at p. 3–14), any segment in the data base between 1,398 and 1,530 base pairs (i.e., 1,464 plus or minus 4.5 percent) would fall within the match criterion and should be counted as a match in the data base.
The grouping of data base samples falling within the match criterion is called a “bin.” The probability of a coincidental match between the crime-scene DNA band and the matching DNA band of the suspect is essentially the bin size divided by the data base size. Thus, the larger the bin, the higher the expected frequency of a coincidental match.
The difference between these four recommended modifications of the statistical analysis and the FBI's own steps to modify the analysis in large measure constituted the basis for the court's conclusion in Barney, supra, 8 Cal.App.4th at pages 819–820, 10 Cal.Rptr.2d 731, that there was no general scientific consensus on statistical analysis of DNA profiles.8
In an effort to demonstrate general scientific consensus concerning DNA profiling, the prosecution took an interesting tack in the present case: it introduced expert testimony to the effect that the FBI's methodology for statistical analysis was more scientifically accurate than the NRC's proposed analysis, but that there was general scientific agreement that the NRC methodology produced a more conservative result and was suitable for forensic use pending further developments in the area. In other words, the prosecution sought to demonstrate consensus by simply adopting the principal alternative body of scientific thought.
The First Step: RFLP Methodology
Admissibility of expert testimony based upon the application of a new scientific technique involves a threefold process. (Compare People v. Leahy, supra, 8 Cal.4th at pp. 594, 604, 34 Cal.Rptr.2d 663, 882 P.2d 321.) First, the reliability of the technique must be established. Reliability means the technique must be sufficiently established to have gained general acceptance in the relevant field of scientific inquiry. (People v. Kelly, supra, 17 Cal.3d at p. 30, 130 Cal.Rptr. 144, 549 P.2d 1240.) Second, the expert furnishing the opinion on reliability must be suitably qualified. (Ibid.) Third, “the proponent of the evidence must demonstrate that correct scientific procedures were used in the particular case.” (Ibid.)
We conclude the evidence concerning the FBI RFLP methodology presented here meets none of the three parts of the Kelly test.9 In so concluding, we recognize that the timing of and interplay between a series of appellate court opinions have left a difficult path for this case to follow. Appellant was not arrested until after publication of Axell provided an arguable basis for admissibility of DNA profile evidence.10 Then, just before this case came to trial, Barney was published. In essence, Barney found scientific consensus on the RFLP methodology (Barney, supra, 8 Cal.App.4th at pp. 813–814, 10 Cal.Rptr.2d 731), but that various published scientific articles demonstrate a lack of consensus on the statistical analysis question. (Id. at pp. 818–819, 10 Cal.Rptr.2d 731.) Barney is dicta on this point, since the appellate court there ultimately found admission of the DNA evidence was not prejudicial. (Id. at pp. 825–826, 10 Cal.Rptr.2d 731.) But at the time, it was the primary guidance the trial court had. So the court ruled on August 11, 1992, that defendant was entitled to a Kelly hearing only on statistical analysis, since Barney “settled” the question of RFLP acceptance and admissibility.
The hearing proceeded accordingly, and when the prosecutor tried to introduce evidence of RFLP acceptance, the court directed her to focus solely on the statistical analysis issue.
After the trial court found the FBI statistical analysis was admissible if it conformed to the NRC methodology, appellant sought writ relief in this court. On October 13, 1992, we filed the Pizarro opinion, and on October 30, 1992, we issued an order denying writ relief to the present appellant: “To the extent the principles and reasoning of the [then-not-final] Pizarro case affect the ruling in petitioner's case, petitioner has failed to demonstrate he is without a remedy before the trial court in the ordinary course of the trial proceedings. He has failed to demonstrate why a request for a new [Kelly ] hearing to address the issue of whether the correct scientific procedures were utilized in this case would be an inadequate remedy.”
Appellant filed a new motion in the trial court on November 2, 1992, requesting, inter alia, another Kelly hearing in which the prosecution would be required to show “that the specific protocol employed by the [FBI] laboratory is accepted as reliable by disinterested scientists familiar with the procedure.”
At a hearing on November 2, 1992, the court acknowledged that Pizarro “perhaps might indicate” that the RFLP issue was still subject to the Kelly foundational showing. “Nevertheless,” the court added, “․ in this particular case, I remember sitting here as we went through the testimony of [the prosecution and defense experts]. [¶] Feeling a little bit of frustration because we were spending it seemed a lot of time talking about the [RFLP] procedures when the sole issue, I thought, was the statistical analysis methodology. [¶] ․ [¶] As I'm sitting here now, the thought that runs through my mind is ․ what more is there to say in terms of a Kelly–Frye hearing [on the RFLP methodology].” The court then placed the burden on the defense to establish, through further cross-examination of prosecution experts or through its own witnesses, that there was a dispute in the scientific community concerning the RFLP methodology.
In Pizarro, supra, 10 Cal.App.4th 57, 12 Cal.Rptr.2d 436 we acknowledged that the version of RFLP testing undertaken by Cellmark had been found reliable for Kelly purposes in Axell, supra, 235 Cal.App.3d 836, 1 Cal.Rptr.2d 411. However, what was missing in Pizarro was an adequate showing that the FBI's methodology was, when scientifically evaluated, essentially the same methodology as Cellmark's (Pizarro, supra, 10 Cal.App.4th at p. 78, 12 Cal.Rptr.2d 436), or that the FBI's methodology was considered reliable independently of Axell 's approval of the Cellmark methodology. (Pizarro, supra, at pp. 78–79, 12 Cal.Rptr.2d 436.) The testimony of the FBI's own scientist was not sufficient to demonstrate general scientific reliability as required by People v. Brown (1985) 40 Cal.3d 512, 533, 230 Cal.Rptr. 834, 726 P.2d 516; independent experts were required. (Pizarro, supra, 10 Cal.App.4th at pp. 79–80, 12 Cal.Rptr.2d 436.)
The prosecution introduced testimony of independent experts in the present case. However, notwithstanding the trial court's belief to the contrary, we have been unable to find anywhere in the record where the independent prosecution witnesses testified concerning general scientific acceptance of the FBI's RFLP protocols. Respondent's brief is similarly unable to direct us to any such testimony. Even if the FBI scientist in charge of the testing in this case, Audrey Lynch, gave some testimony about the reliability and general acceptance of the FBI RFLP protocols, the insufficiency of such “in-house” expert testimony was the explicit basis for reversal in Pizarro. (See Pizarro, supra, 10 Cal.App.4th at pp. 79–80, 12 Cal.Rptr.2d 436.) Accordingly, as in Pizarro, the record 11 here fails to support the necessary Kelly foundation for the FBI RFLP evidence.
The Second Step: Statistical Analysis
We now turn to the issue of statistical analysis of DNA profiles, once the RFLP methodology has resulted in declaration of a “match” between the crime-scene sample and the suspect's DNA sample. Here, by contrast to the evidence concerning the RFLP protocols, the prosecution was called upon by the trial court to make a full and thorough Kelly showing before it was permitted to introduce before the jury any statistical evidence concerning the degree of rarity of appellant's DNA profile.
The prosecution's witnesses did not testify that the NRC methodology for DNA statistical analysis was scientifically reliable. In fact, they were adamant that this methodology was wholly unreliable in that it produced unnecessarily conservative results that present a particular profile as far, far more frequent in the population than the profile is in reality.
Yet the evidence was also clear that the scientific community regards the NRC statistical methodology as forensically reliable. That is, there is general scientific agreement that DNA profiles provide a powerful tool for demonstrating that it is very unlikely that someone other than the matching suspect was the source of the crime-scene sample. There is general agreement in the forensic DNA community that the jury should not be denied some version of this evidence that a coincidental match is “very unlikely,” and that numerical statements of probability are a useful way of presenting this evidence. There is also general consensus that statistical analysis provides only an inexact estimate of the probability of a coincidental match, and that all inexactness should be construed and reported in a manner that implements the constitutional presumption of innocence.12 (NRC Report at 5–6, 5–7.)
Appellant has failed to convince us that the limited nature or potentially short duration of this scientific consensus of forensic reliability undermines the admissibility of such evidence. In other words, the evidence is persuasive in this case that the first two Kelly requirements have been met as to the statistical analysis methodology.
We therefore turn to the third element of the Kelly test, whether the evidence shows that the “correct scientific procedures were used in th[is] particular case.” (People v. Kelly, supra, 17 Cal.3d at p. 30, 130 Cal.Rptr. 144, 549 P.2d 1240.) We find that the evidence establishes just the opposite.
The fundamental starting point in this part of the discussion is that the FBI's calculation pursuant to its view of the NRC methodology resulted in a coincidental probability about one-half of the probability calculated by the FBI's own methods, i.e., 1 in 65,000 versus 1 in 30,000.13 This result is anomalous and particularly disturbing because of the reasoning largely underlying scientific endorsement for the NRC method; it is oriented toward producing a conservative result. If it is an acceptable methodology precisely because it is the most conservative, what are we to make of an even more conservative result from a less conservative methodology?
The answer seems to be that, while asserting at the trial below its adherence to the NRC methodology, the FBI in fact only partially performed its calculations in accordance with NRC recommendations.
To begin with the fourth NRC recommendation, concerning bins in the data base, we note the FBI ordinarily employs a wholly different method of binning, called the fixed bin method. In this method, the array of possible band lengths for a given site is grouped into 31 bins. (See NRC report at p. 3–14.) Actual bands in the data base will be distributed throughout the bins. The sample band is then matched with a bin; if the match “window” produced by the measurement tolerance overlaps two bins, the FBI uses the bin with the higher number of bands from the data base so as to give the suspect the benefit of the uncertainty. The probability of a coincidental match between the crime-scene sample and the suspect's DNA profile is the ratio of the number of bands in the matching bin to the number of bands in the data base.
Two points are important about the FBI's use of the fixed bin method as its normal analytical tool. First, the NRC report expressed approval of the fixed bin method as an alternative to the NRC's “floating bin” method. (NRC report at 3–14.) Thus, there is no need for the FBI to abandon that method in order to find consensus in the NRC methodology.14
Second, the prosecution placed in evidence a report by two FBI forensic scientists who studied the fixed bin and floating bin methods for comparative “conservativeness.” They discovered that a 5 percent floating bin (i.e., the plus or minus 2.5 percent window used in the present case) consistently produced a lesser probability of coincidental match using the FBI data base than did the fixed bin method. Only when they doubled the size of the floating bin to 10 percent (i.e., plus or minus 5 percent from the average of the questioned samples) did the floating bin results become more conservative than the fixed bin result.15
This result is fully consistent with the observation by appellant's genetics expert that the bin window employed in the present case should have been correlated to the tolerance for interautorad comparisons, not the intraautorad tolerance. Similarly, it is consistent with the NRC's admonition that, if a floating bin is used, its size must correspond with the lab's interautorad tolerance. (NRC report at 3–14.) Accordingly, we conclude that in this regard the (plus or minus 2.5 percent) floating bin procedure used by the FBI in this case was not the “correct” scientific procedure, i.e., the procedure about which there is general consensus concerning conservativeness and, thus, forensic reliability.16
We are satisfied there is substantial evidence in the record to support the trial court's finding of scientific consensus on the NRC statistical methodology and its rejection of defense expert Mueller's relatively minor disagreements with the NRC methodology. There is, however, no evidence to support the trial court's implied finding that the prosecution's DNA statistical analysis was performed in accordance with the accepted NRC methodology.
Without the erroneously admitted DNA evidence, the jury had before it only the ABO evidence, which excluded just 50 percent of the male Hispanic population as the possible source of the crime-scene sample. Given this rather insubstantial evidence of nonexclusion, and in light of the other evidence incriminating appellant, the weakness of which the prosecutor acknowledged throughout the trial, we must conclude it was reasonably probable appellant would have had a more favorable verdict had the trial court not erroneously admitted the DNA evidence. (People v. Watson (1956) 46 Cal.2d 818, 836, 299 P.2d 243.)
Retrial of this case is not precluded, since the erroneously admitted evidence was sufficient to permit a finding of guilt beyond a reasonable doubt. (Lockhart v. Nelson (1988) 488 U.S. 33, 40, 109 S.Ct. 285, 290, 102 L.Ed.2d 265; People v. Harvey (1984) 163 Cal.App.3d 90, 107–108, 208 Cal.Rptr. 910; People v. Mattson (1990) 50 Cal.3d 826, 853, fn. 16, 268 Cal.Rptr. 802, 789 P.2d 983.) In the likely event the prosecution is able to meet the Kelly threshold as to the RPLF portion of the evidence, it is also likely the FBI will be able to reanalyze its raw data by the fixed-bin or floating-bin method and present a statistical result achieved through the NRC methodology so as to meet the third prong of Kelly as to the statistical analysis of DNA profiling. Any new trial will be plenary, however, and nothing in the present opinion should be construed to prevent the prosecution from establishing the admissibility of some other methodology in light of any scientific consensus achieved in the time since the original trial. Similarly, if the NRC statistical methodology has been unable to retain scientific consensus on the issue of its “forensic reliability,” the defense is entitled to establish this at a new Kelly hearing as well.
Consciousness of Guilt
Appellant raises a final issue that merits only summary treatment. He contends the jury should not have been instructed that it could infer consciousness of guilt from appellant's refusal of a voluntary blood test.
After first refusing all cooperation, appellant agreed to submit saliva and hair samples, but declined to submit a blood sample voluntarily. The jury was instructed it could consider this later refusal as reflecting consciousness of guilt.17 Appellant contends this was error, since it penalized him for exercising his Fourth Amendment right to be free from warrantless searches.
This claim is conclusively resolved by People v. Roberts (1992) 2 Cal.4th 271, 311, 6 Cal.Rptr.2d 276, 826 P.2d 274: “The Fourth Amendment claim might require more scrutiny on a better record, but the record does not explain whether defendant refused to take the blood test for reasons within the scope of the Fourth Amendment. [Citations.] Defendant fails to establish that he exercised a protected right, and we must therefore reject his claim.”
In our case, there was no suppression hearing at which appellant testified, and he did not testify at trial. The motion in limine seeking exclusion of the evidence was not supported by a declaration from appellant explaining his refusal. The police officer who requested the blood sample simply testified that appellant declined, without any statement about appellant's reasons. Under these circumstances, appellant has failed to establish he exercised a protected right in refusing to give a blood test.
The judgment is reversed and the matter is remanded for further proceedings consistent with this opinion.
1. The trial court determined the criminologist's testimony was admissible following extensive Kelly (People v. Kelly (1976) 17 Cal.3d 24, 130 Cal.Rptr. 144, 549 P.2d 1240) hearings fully described in the discussion section, post.
2. Further statutory references are to the Penal Code unless otherwise noted.
3. The NRC report was referred to extensively by the witnesses in this case, although it was never introduced into evidence. We take judicial notice of the report.
4. Other forensic laboratories, such as Cellmark and Lifecodes, use similar, but distinct, processes.
5. Restriction fragment length polymorphism analysis. (Pizarro, supra, 10 Cal.App.4th at p. 69, 12 Cal.Rptr.2d 436.)
6. In the present case, for one chromosome both the questioned sample and appellant's sample contained more than 10,000 base pairs. Since none of the four bands in question could be measured reliably, that autorad neither excluded nor included appellant.
7. To take a hypothetical example, assume band-length X had a greater frequency among Slavic peoples, that persons of Slavic heritage were prevalent in great numbers in northern portions of the East and Midwest, and that they tended to marry within the Slavic–American community. If the Caucasian data base had been collected in Atlanta and Dallas, where there may be a dearth of Slavic–Americans, the overall frequency of band-length X would be less than if the database were truly representative of the Caucasian population of the United States. A data base compiled in Chicago might show that band-length X was dramatically more frequent than in the Atlanta/Dallas data base. The frequency-enhancement would be even more pronounced if the perpetrator and suspect were known to be from the Slavic population, and the data base had been developed entirely from that community.
8. In addition to these four recommendations designed to address the (untrue) assumptions of the simplistic application of the product rule, the NRC made two additional recommendations for forensic use of DNA statistical analysis. First, to help the jury understand that the statistical analysis is only one way of estimating DNA profile frequency, the NRC recommended that the jury be told the result of a frequency analysis obtained by the counting method. In other words, if the data base contains band information from only 200 samples, and no profile in the data base matched the suspect's, the counting method would report that the expected frequency of a coincidental match was 1 in 201 (the existing data base plus the new profile). Second, the NRC suggested that labs engage in externally administered proficiency testing, and that a lab's error rate be reported to the jury as part of the DNA profile evidence.
9. Respondent strongly urges, as it did in Barney and Pizarro, that DNA profiling is not subject to Kelly analysis. That issue having been fully treated and rejected in the recent Pizarro opinion, there is no occasion to revisit the issue in the present case.
10. The crime was committed on November 2, 1989. Axell was decided October 29, 1991. The complaint here was filed on November 11, 1991.
11. Although the parties in the present case introduced into evidence several scientific studies not part of the trial record in Pizarro, these studies only explain and discuss the relevant methodology. They do not discuss the issue on a Kelly basis, namely, general scientific acceptance of the methodology. Normally, as Pizarro points out, such an overview of the scientific status quo will be presented through expert witnesses who discuss and rely on the relevant scientific literature. The courts are ill suited to review the literature itself and reach a conclusion on general acceptance based solely on that literature. (Pizarro, supra, 10 Cal.App.4th at pp. 87–88, 12 Cal.Rptr.2d 436; see People v. Leahy, supra, 8 Cal.4th at pp. 602–603, 34 Cal.Rptr.2d 663, 882 P.2d 321; but see Wallace, supra, 14 Cal.App.4th at p. 659, 17 Cal.Rptr.2d 721.)
12. We note this concept of Kelly “forensic reliability” remains a somewhat unmapped region presenting potential dangers. Thus we find it appropriate to conservatively construe these statistics. In some cases, the strictures imposed on the forensic version of a scientific test may conflict with the trial court's power to control truth-seeking functions of the trial itself. In the case before us, for example, the prosecutor, without defense objection, was allowed to elicit from a witness (and repeat in closing argument) that “in the FBI laboratory, they have never seen two [un]related individuals who matched at three probes.” How does this empirical report of uniqueness of the DNA profile square with the court's limitation of the evidence to a 1 in 65,000 match?With these concerns in mind, we observe in the present case that the evidence demonstrates only a general “cease fire” in the relevant scientific community, and this cease-fire has resulted for the time being in general acceptance of the evidentiary value of DNA statistical analysis produced in strict accordance with the NRC methodology. In this instance, agreement on “forensic reliability” has produced evidence that will be helpful and not unduly confusing to the jury.
13. We wonder about the effect of these probability equations on the jury. For example, if the jury hears only one probability statistic from the prosecution, does it matter whether that figure is 1 in 30,000 or 1 in 60,000 or 1 in 300,000? These are all very low probabilities of coincidental match between the suspect and the perpetrator when compared with, to take an area with which a juror may have some familiarity, pari-mutuel odds at a horse track where 40 to 1 reflects a very low probability of the “suspect” winning the race. On the other hand, if the jury is told that DNA profiling routinely results in coincidence probabilities in the range of 1 in 2 million or 1 in 200 million, it may conclude that a probability of “only” 1 in 30,000 leaves room for a reasonable doubt about the perpetrator's identity. In the present case, the jury heard that ABO blood typing resulted in a 1 in 2 probability of a coincidental match, that the acceptable method of DNA analysis resulted in 1 in 65,000 odds, and that the FBI's unacceptable methodology produced better odds of coincidental match, at 1 in 30,000. The FBI scientist characterized the match as “rare,” as opposed to “common” or “extremely rare.”
14. It would be necessary to incorporate the floor/ceiling methods into the FBI's calculation, however. (See fn. 16, post.)
15. An independent prosecution expert testified he obtained a probability of 1 in 2,000 using the FBI fixed bin method and 1 in 100,000 using the FBI version of the floating bin method.
16. As noted, the FBI also analyzed the matched bands by the fixed bin method, and that method is approved by the NRC report. The jury was told that analysis resulted in a probability of coincidental match of 1 in 30,920. However, in reaching this result, the FBI did not employ either the floor procedure or the ceiling method recommended by the NRC. (NRC report, 3–24, 3–25.) Instead, it employed exactly the methodology about which the NRC report demonstrates a lack of scientific consensus.This lack of consensus involves the subpopulation issue and the perpetrator-ethnicity issue. Thus, even though the FBI does not have 15 or 20 data bases as recommended by the NRC report, and even though there is scientific debate concerning the effect of subpopulation on the data base, the FBI did not use the NRC's 10 percent floor in calculating coincidence frequency by the fixed bin method. In the case of its most conservative probability (1 in 30,920), it used a probability of 4 percent for one of the bands (probe D2544, band No. 2), instead of the 10 percent floor recommended by the NRC. If we correctly understand the mathematics involved, conformance with the NRC methodology in this instance would raise the frequency of coincidental match to 1 in 12,376. (Calculated as: 2(.255 x .183) x 2(.119 x .10) x 2(.104 x .175) = .0000808 or 1 in 12,376.)Further, both the FBI and the NRC report recognize there is a problem if the coincidental match probability is reported only in terms of a population base matching the suspect's ethnicity. The problem is that this assumes the actual perpetrator is the same race as the suspect (see People v. Pizarro, supra, 10 Cal.App.4th at pp. 93–94, 12 Cal.Rptr.2d 436) and it assumes that the suspect's genetic pool is entirely the same as that of the data base population group. (See fn. 7 ante.) As discussed above, the NRC report recommends the ceiling method—taking the highest frequency for each band from any available data base—in order to neutralize these assumptions. The FBI instead calculated a probability for each available data base, namely, Caucasians, Blacks, southeastern Hispanic, and southwestern Hispanic, and it reported all four probabilities for the jury's evaluation in light of the other prosecution or defense evidence. These results were presented to the jury, without defense objection, even though they were the result of the traditional FBI methodology rejected by the trial court for failure to satisfy Kelly.In the present case, this method of calculation resulted in probabilities from 1 in 30,920 (southwestern Hispanic) to 1 in 225,431 (Black). As we understand the application of the NRC ceiling method to the FBI's reported fixed bin probabilities, the composite (all race) probability of coincidental match would be approximately 1 in 26,000, not factoring in the floor concept, and 1 in 10,500 applying both the floor and ceiling concepts. (Calculated without floor: 2(.263 x .199) x 2(.124 x .04) x 2(.105 x .175) = .000038 or 1 in 26,316. Calculated with floor and ceiling methods: 2(.263 x .199) x 2(.124 x .10) x 2(.105 x .175) = .000095 or 1 in 10,526.)We have recalculated the FBI fixed bin analysis in light of the NRC floor and ceiling concepts not to provide a “correct” coincidence probability for use on remand. Rather, we have merely tried to show the nature of the FBI departures from the NRC methodology by using concrete examples. The data in the record do not permit similarly concrete examples of recalculation of the floating-bin approach using a corrected interautorad tolerance.
17. “If you find that the defendant was offered and refused a blood test after he had been made aware of the nature of the test and its purpose, such refusal is not sufficient, standing alone and by itself, to establish the guilt of the defendant, but it is a fact which, if proved, may be considered by you in the light of all other proved facts in deciding the question of his guilt or innocence. The weight to which such a circumstance is entitled and whether or not such conduct shows a consciousness of guilt are matters for your determination.”
VARTABEDIAN, Associate Justice.
STONE, (WM.A.), Acting P.J., and HARRIS, J., concur.