An expert witness fails to bridge the “analytical gap.”
The Court of Special Appeals recently decided Matthews v. State, No. 3280, Sept. Term, 2018 (Feb. 25, 2021), a case where two homicide victims were killed by shotgun blasts delivered at close range. The sole issue at trial was the identity of the shooter. Writing for a panel that included Senior Judge Paul Alpert and Judge Kathryn Graeff, Judge Douglas Nazarian observed: “The story of this case is complicated and hard to follow.” Those complications were due largely to the cast of characters who served as trial witnesses.
The case is the first reported opinion dealing with the reliability of expert testimony issued after the Court of Appeals’ decision in Rochkind v. Stevenson, 471 Md. 1 (2020). But the panel’s application of Md. Rule 5-702(3)—with its requirement that an expert witness not fall into an “analytical gap”—did not depart from pre-Rochkind jurisprudence, confirming that even before Rochkind’s adoption of the Daubert standard, Maryland courts strictly applied Rule 5-702’s mandate that expert testimony be based on a sufficient factual basis. Unlike the case as a whole, the facts regarding the “analytical gap” issue luckily are not complicated.
Shortly after midnight, two dead bodies were found by the side of a dead-end street that doubled as an open-air drug market. The evidence at trial painted a picture of chaotic conduct in the hours preceding the shootings. Many of the witnesses were heavily inebriated, having consumed copious quantities of alcohol, drugs, or both. Some of the witnesses gave accounts that evolved over time. The witnesses’ levels of intoxication, coupled with their struggles with consistency and credibility, helped create the complicated story. We will stick to the facts necessary to review the expert testimony issue.
The prosecution presented two witnesses who observed Kirk Matthews in the area shortly before the shooting. The prosecution presented two other witnesses who identified Matthews as the man they saw in the area, carrying a shotgun. Another witness, Rico, said he was in the area that night, but denied seeing Matthews. Prosecutors then elicited testimony from Rico’s uncle, who said “that Rico told him he had seen Mr. Matthews shoot two people and drag them off the road, and that it was the worst thing he had ever seen.”
A defense witness testified that she had consumed her ordinary four or five White Russians that night. She said she saw a white man, walking down the road in front of her house, holding a shotgun, and he “was roughly 5’11” or taller, blondish hair, thin, and twenty-five or younger.” The Defendant Matthews, however, is an African-American, who stands about 5’8.”
In addition to the testimony of the persons present that night, there were videos that shed light on what had transpired. A nearby house had two security cameras, one of which was on the garage. “The cameras didn’t capture the shootings on video, but they did provide evidence and helped paint the picture of events on the night of the murders.” The State presented expert testimony from an FBI physical scientist, who along with her trainee studied footage from the garage camera and issued a report, “in which they used photogrammetry and reverse photogrammetry projection to identify the shooter from contemporaneous videos.”
After the jury found Matthews guilty of two counts of second-degree murder and related crimes, he appealed, raising three issues. He first argued that the circuit court erred in allowing Rico’s uncle to testify about the allegedly prior inconsistent statement Rico made to him. But by saying he did not remember telling his uncle that he had seen more that night than he had shared in his trial testimony, Rico “failed to admit having made the statement,” and extrinsic evidence of Rico’s prior inconsistent statement to his uncle was thus admissible under Md. Rule 5-613(b).
Matthews also argued that the circuit court erred in limiting his cross-examination of a witness who had been charged with several serious crimes. Before Matthews’s trial, that witness entered a plea agreement, although his sentencing was delayed until after Matthews’s trial. The circuit court allowed defense counsel to cross examine the witness about his plea agreement, as well as other aspects of the charges levied against him, but restricted the scope of that cross-examination. According to the Court of Special Appeals, that ruling gave Matthews the threshold level of inquiry needed to satisfy his constitutional right to confront witnesses against him, and the circuit court did not abuse its discretion in restricting further cross-examination.
That brings us to his third issue, which we examine more closely. At a pre-trial hearing, Kimberly Meline testified that she was a physical scientist with the FBI, who “specializes in multimedia evidence.” She “explained that photogrammetry involves taking measurements from photographs to determine how fast a vehicle is moving through a video, the length of a firearm, or how tall a subject is.” She further “explained that reverse photogrammetry projection is a specific type of photogrammetry that involves going back to a scene, recreating the image conditions, and then placing a calibrated measuring device where the subject was standing as a way of determining how tall the person was.”
Meline said she analyzed the video from the garage camera and concluded that one image could be used for her analysis. She and her trainee returned to the house, and using the same camera that had captured the “questioned image,” they “overlaid the live video with the questioned image using software, aligned the stationary items such as trees in both images, and placed a height chart in the location where the suspect appeared to be standing in the questioned image.” In short, “the overlay between the questioned image and the image with the height chart was used to estimate how tall the suspect was.”
She described “her process for estimating the uncertainty or error associated with the height measurements,” explaining that the margin of error was calculated by using “positional accuracy,” which “is determined by pixels in the image,” and “the resolution of the imagery.” She then “estimated the suspect’s height as 5’8”, with a margin of error of plus or minus 0.67 of an inch.”
There were, however, additional complications that affected the accuracy of her opinion. While she labeled the 0.67 inch margin of error as “calculated uncertainty,” she also described what she called the “incalculable uncertainty.” The latter was “based on factors such as the quality of the image (it was taken at night), the unevenness of the terrain, the body position of the individual, the inability to see his feet, and the head covering the individual was wearing.” She was concerned “about this particular examination,” because the subject was standing a “considerable distance” from the camera, and “because the uneven terrain would make a difference in whether the camera lens was actually at the position of the suspect.” She also noted her concerns about “the stature of the individual,” as there was no image where the subject was “visible from head to toe and at the full height of their stride.” She acknowledged she had no “scientific way of quantifying how these dimensions” affected her measurements and that the degree of uncertainty could be greater than 0.67 inches and could not be quantified.
The circuit court admitted her testimony, but the Court of Special Appeals held that decision was an abuse of discretion and reversed the convictions. When the case was tried, the Frye-Reed standard still governed the admissibility of expert scientific testimony. That standard worked together with Rule 5-702, which governs all expert testimony, to regulate the admissibility of expert testimony. While the Matthews’s appeal was pending, the Court of Appeals decided Rochkind, adopting Daubert, and that new standard now applied. But “the elimination of Frye-Reed”s focus on the acceptance of the methodology in the scientific community versus Daubert’s look at the overall reliability of the testimony” did not affect the outcome. Matthews did not challenge the general acceptance or overall reliability of reverse photogrammetry. Relying on Rule 5-702(3), he instead argued that Meline’s expert opinion did not rest on a sufficient factual basis. “This is still a reliability question, but it turns on the application of this universe of facts to the established method rather than on questions about the reliability of the method itself.”
The expert testimony was intended to measure the height of the person in the video so to help identify the shooter. As noted, the defense presented testimony that the shooter was at least 5’11” and white, while Defendant Matthews is African-American and 5’8”, “and the contemporaneous videos weren’t clear enough to allow a distinction even between these two possible suspects.” The expert testimony was crucial to the prosecution. The appellate court, however, found the testimony unreliable, as it rested on an inadequate factual foundation. The various missing variables acknowledged by the expert “prevented a reliably accurate height calculation. Put another way, the analytical gap between the data available for reverse photogrammetry projections and the conclusion Ms. Meline offered to the jury remained unbridged.” Defense counsel’s ability to cross-examine Meline about that gap did not cure that problem, as “it should not have fallen to the jury to work through the science on its own.” Thus, under either Frye-Reed or Daubert, expert testimony that does not rest on a sufficient factual basis fails to bridge the analytical gap between the data and the expert opinion, and is inadmissible, even if it uses a generally acceptable and reliable methodology.