I wrote in September 2015 that the court of appeals’ view of the admissibility of retrograde extrapolation under Daubert did not look much different from its take on the admissibility of that evidence under old Rule 702. As of yesterday, it does. The court of appeals in State v. Babich, __ N.C. App. __ (2017), changed the green light for retrograde extrapolation testimony in DWI cases to yellow.
Where we were. The North Carolina Court of Appeals has, on several occasions, liberally approved the admission of retrograde extrapolation testimony. For example, the court in State v. Turbyfill, __ N.C. App. __, 776 S.E.2d 249 (2015), held that a field technician for the Forensic Tests for Alcohol branch of DHHS (FTA), who was trained as a law enforcement officer, was properly allowed to testify about a retrograde extrapolation he performed using an FTA form, notwithstanding his inability to explain the basis for his calculations. And in State v. Green, 209 N.C. App. 669 (2011), the court of appeals held that the trial court properly allowed the head of the FTA to testify regarding the defendant’s specific alcohol concentration at the time of an earlier accident based on various assumptions about how many drinks the defendant consumed before he submitted to a breath test some three hours later. In only one extreme case, when the head of the FTA testified about a retrograde extrapolation he performed based on the smell of alcohol, had the court of appeals found error in the admission of expert testimony on retrograde extrapolation. State v. Davis, 208 N.C. App. 26 (2010).
What happened yesterday. The court of appeals held in State v. Babich that the trial court abused its discretion in admitting expert testimony from a forensic chemist that the defendant’s alcohol concentration at the time she drove was between 0.08 and 0.10. The chemist reached this conclusion by applying a mathematical formula to Babich’s .07 breath test result that assumed Babich was eliminating alcohol at an average rate for the hour and forty-five minutes between the time Babich’s car was stopped and her breath was tested. Because no facts supported the expert’s assumption that Babich’s body was eliminating rather than absorbing alcohol during this time, the court held that her testimony was inadmissible under Daubert. (A chart depicting the rates of alcohol absorption and alcohol elimination for subjects in one scientific study is reprinted below.)
Facts. Babich was driving 83 miles per hour in a 45-mile-per-hour zone at 3:20 a.m. when a Wilmington police officer began to follow her. Babich sped up to 91 miles per hour, braked at an intersection with a red light, slowing her car down to 45 miles per hour, and then drove through the intersection at that speed, despite the red light. The officer pulled her over. After observing several signs of impairment, the officer arrested Babich for driving while impaired. Babich provided breath samples at 5:07 and 5:09 a.m., which revealed an alcohol concentration of 0.07.
Babich was charged with reckless driving to endanger, driving while license revoked, speeding, driving while impaired, and habitual impaired driving.
Trial. Bethany Pridgen, a forensic chemist with the Wilmington Crime Lab, testified at Babich’s trial as an expert on retrograde extrapolation. In addition to opining about Babich’s alcohol concentration at the time she drove, Pridgen testified that the formula she used could be applied only if alcohol was no longer entering the person’s bloodstream and the person’s blood alcohol level was declining. She acknowledged that many factors affect alcohol absorption, including when the person last consumed alcohol, how much was consumed, and whether the person consumed food. Pridgen had no information about those factors, but assumed for purposes of her calculations that Babich’s body was already in the alcohol-elimination phase when she was pulled over by police.
The jury convicted Babich of impaired driving, speeding and reckless driving. Babich stipulated to three prior DWI convictions, and thus also was convicted of habitual impaired driving. She was sentenced to 19 to 32 months imprisonment.
On appeal. Babich argued on appeal that the trial court erred in allowing Pridgen to testify about her calculations as they were not based on sufficient facts or data. The court of appeals agreed, holding that the testimony did not satisfy the Daubert “fit” test as it was not sufficiently tied to the facts of the case in a way that would aid the jury. The court cited several appellate court decisions from other states reaching the same conclusion on similar facts. The court held that an expert opinion on retrograde extrapolation that assumes the defendant is in the alcohol elimination phase must be based on facts that support that assumption. These facts can come from the defendant’s statements, the officer’s observations, from other witnesses, or from circumstantial evidence. But they cannot simply be assumed. When “at least some facts” support the expert’s assumption that the defendant’s body is eliminating alcohol, the testimony is admissible, and the issue becomes one of weight and credibility.
Babich did not win. Despite the court of appeals’ determination that the trial court erred, it did not grant Babich a new trial. The court determined that, even without the impermissible expert testimony, there was no reasonable possibility that the jury would have reached a different result. Thus, the court upheld Babich’s conviction and sentence.
This may be big. The Babich court’s conclusion that the trial court abused its discretion in admitting retrograde extrapolation testimony of the sort that state courts have routinely admitted for years may signal its willingness to more rigorously review other aspects of this testimony and the experts who provide it. Babich cites as support the Texas Court of Criminal Appeals’ decision in Mata v. Texas, 46 S.W.3d 902 (Tex. Crim. App. 2001) (en banc), which established a far higher standard for the admissibility of such testimony than has ever applied in North Carolina. Mata stated that while the science of retrograde extrapolation “can be reliable in a given case,” the expert “must demonstrate some understanding of [its] difficulties” and “an awareness of the subtleties of the science and the risks inherent in any extrapolation.” Id. at 916.
The fog of judicial credulity about what the State says is ‘scientific’ can be amazing. Dubowski’s article in the 1970’s warned about the dangers of extrapolating for court. Wilkinson’s graph (despite its outdated note that the common DWI standard is .10%BAC) pretty clearly shows the danger of extrapolating without knowing accounting for the assumptions involved.
Judicial vigilance and good defense work are essential if the State can’t be trusted to put forward the limitations of “science-y” sounding testimony as part of their evidentiary foundation. In NM’s State v. Marquez, police testimony linking FST’s with BAC was held to be harmful error because the very cloak of scientific language masking defective evidence was enough to unfairly influence a jury.
North Carolina’s checkered history of police testimony about extrapolation–e.g. Glover’s testimony (State v. Davis, 2010) about extrapolation starting with his assumption that smelling alcohol on a subject hours after an accident corresponded to a .02% BAC from which he could then extrapolate her BAC was properly stomped flat–twice–by the NC Ct. of Appeals, but is a great indication of how far some police ‘scientists’ will go to make up evidence, and the complicity of prosecutors in trying to present it.
The scientific reliability of FTA’s retrograde extrapolation is a myth. The blog’s chart is an example of the mythology in action. (To get to statutory units, divide the chart’s Blood Alcohol Concentration by 10.) In the hands of the National Institute of Alcohol Abuse and Alcoholism (“NIAAA”), the chart shows exquisite precision. The lines closely follow the data points. The four curves are remarkably similar. The legend hovering above the chart shows that four drinks certainly will put a person above 0.08. With data such as these, how could retrograde extrapolation not be reliable? Well, that “science thing” gets in the way.
NIAAA extracted the chart from a scientific study by Wilkinson et al. published in 1977. NIAAA added the legend. NIAAA didn’t mention that Wilkinson’s subjects drank alcohol on an empty stomach. Hardly representative of most drivers charged with DWI. NIAAA also didn’t say that the chart is based upon average alcohol concentrations. (Generally, averaging tends to hide variability and uncertainty, contrary to the scientific spirit of Daubert/McGrady. Consider that, on a planet average, we’re males living in Asia – and we’re sober.) Importantly, in its publication, NIAAA is silent about the other charts in the Wilkinson paper. The other charts tell a far different story about retrograde extrapolation. For example, about that four-drink peak: The subjects’ actual peak concentrations were highly variable, ranging from 0.065 to 0.112. Getting close to double. Lesson: When the subject is alcohol, humans – and defendants – are not averages.
Another example. One of the (only) eight study subjects showed a period when his alcohol elimination rate was about 0.004 grams per deciliter per hour (g/dL/h). Compare this rate to FTA’s rate of 0.0165 g/dL/h. How could this subject’s rate be so low relative to “the accepted” value? Clearly, his alcohol concentration isn’t going up. And it’s just barely going down. What’s going on? He’s on a plateau. The rates of alcohol uptake and elimination are roughly matched. Or, to use the FTA’s courtroom analogy, the flow of water from the spigot into the sink is about the same as the flow down the drain. The scientific literature shows plateaus are not rare. Retrograde extrapolation is not applicable to such plateaus. And science cannot yet predict whether a person is or will be on a plateau. To tell whether a person is on a plateau requires measurement of their alcohol concentration over time. This is what Wilkinson did. But if you measure a person’s alcohol concentration over time, there’s no need to do retrograde extrapolation in the first place.
One more. Even the blog’s chart shows that retrograde extrapolation has a limit. Around an alcohol concentration of 0.02, elimination does not give a straight line. The chart shows curves. To be valid, retrograde extrapolation must have a straight line. If applied at 0.02 or less, retrograde extrapolation will give invalid and high results. The bottom line: Those who would extrapolate from 0.02 are inflicting non-science. Possible implication: For those who might have been convicted based upon retrograde extrapolation from 0.02 or less, the chart is evidence of innocence.
In conclusion, it appears that NIAAA cherry-picked a chart and used it to effect a statistical illusion. In doing so, NIAAA advances the myth. The National Research Council in its 2009 report has stressed the need for forensic science to account for and to report uncertainty. NIAAA trumpets certainty and obscures uncertainty. The latter is critically important when the subject is “beyond a reasonable doubt.” Regrettably, NIAAA’s misleading chart has been published once again. Perhaps to NIAAA’s satisfaction.
This cat must be a prosecutor; the fact is, the only true way that retrograde exstrapulation can only be scientifically proven, is when; a person is tied down with wires set onto electronic mechanisms that will scinitifcally show that “while a person is driving” what “specific” BAC level s/he is at as a matter of fact. Remember, the scientists’ that developed this scientific technology graded it in the same true fashion. So, 2 hours for the BAC to peak then it starts to drop like a rock…regardeless of all the legal mombo-jumbo the state will nonetheless fail in provong it’s purdon of doubt. Rememeber, if just an inkling of reasonable doubt exist, the court or jury cannot convict. Peace.