About ten years ago, a Texas jury sentenced George Powell to 28 years in prison for an aggravated robbery. Texas allows defendants to elect whether the judge or jury will determine the sentence in noncapital cases. Powell’s conviction was affirmed on appeal.
Powell’s conviction was based in part on a video of the convenience store robbery. The robber was wearing sunglasses and a baseball cap. A store clerk told the police that the robber was five-foot-six but testified at trial that the robber was five-foot-ten.
As ExpertPages explained in 2017, a retired police officer named Michael Knox testified that he used photogrammetry to determine that the robber in the video was more than six feet tall. Powell is six-foot-three.
After his conviction was affirmed, Powell sought habeas corpus relief. Part of his challenge was based on the dubious expert testimony upon which his conviction rested.
Post-Conviction Expert Testimony
At a post-conviction hearing, Knox testified at length about his undergraduate education and his law enforcement training in accident reconstruction. At the time of his trial testimony, however, Knox had no degree and no formal education in video analysis.
Knox also testified about the procedures he followed to estimate the robber’s height from the videotape. He regarded those procedures as being established by the science of photogrammetry.
In response, Powell called Grant Fredericks as an expert witness. Fredericks is a certified forensic video analyst who has taught courses for a number of law enforcement agencies, including the FBI National Academy.
Fredericks testified that a trained video analyst can use principles of photogrammetry to arrive at valid conclusions. He submitted a report to the Texas Forensic Science Commission in which he concluded that Knox’s methodology had reduced photogrammetry to “junk science.” The Commission agreed with that assessment.
After the trial, Knox responded to the Commission’s inquiry by “revising” his estimate of the robber’s height to at least five-foot-ten. Knox claimed that it was impossible to determine a maximum height. Fredericks testified that he did not understand the basis for Knox’s revised opinion.
Fredericks explained in detail why Knox’s methodology was unreliable. In particular, he criticized Knox’s failure to understand the difficulty of determining height from a compressed video image.
Based on his own methodology, Fredericks determined that the robber was no taller than five-foot-nine. In addition, Fredericks scanned a picture of Powell in a prison jump suit, inserted it into a frame of the video showing the robber, and concluded by comparing the two men that they could not possibly be the same person. Other witnesses criticized that methodology because Powell was not standing in an identical posture in both photos and his weight (although presumably not his height) might have changed during his imprisonment.
Knox used PhotoModeler software with which Fredericks is unfamiliar. Fredericks suggested to the Forensic Science Commission that it hire Dan Mills to do a peer review of Knox’s work because Mills has experience with the software that Knox used.
Mills concluded that Knox’s revised analysis (following the conviction) was better than his first analysis (upon which the conviction was based). He also concluded that both Knox’s opinion and Frederick’s opinion were highly subjective.
A habeas challenge under Texas law can be based on proof of actual innocence. The court was not convinced that the difference in expert opinions conclusively proved that Powell was not the robber.
The court found that Knox was a credible witness, notwithstanding that the Forensic Science Commission questioned his methodology. The court found that Fredericks was also a credible witness. Since the two experts used different methodologies, the court concluded that their differing findings were not necessarily contradictory.
In essence, the court thought that the subjectivity involved in both methodologies created a range of possible results and that Knox’s first and second opinions, as well as Frederick’s opinion, all fell within that subjective range. One might question how methodologies that are so subjective could possibly meet the Daubert standard, but that was not the question before the court.
One might also wonder how an opinion that the offender was at least six-foot-one, an opinion that the offender was at least five-foot-ten, and an opinion that the offender was no taller than five-foot-nine are not contradictory opinions, regardless of the methodology that was used. Ultimately, given the subjective nature of the opinions, the court likely believed that nobody can say with any certainty how tall the robber was.
No expert testimony established that the robber was as tall as Powell. Still, since the expert testimony could not conclusively establish that the robber in the video was not Powell, the court concluded that Powell failed to prove his actual innocence.
New Scientific Evidence
A Texas statute also permits a conviction to be vacated when new scientific evidence, unavailable at the time of the trial because of a change in scientific knowledge, excludes the defendant as the person who committed the crime. For example, if new techniques for analyzing DNA were not available at the time of trial, and the new analysis points to the defendant’s innocence, Texas law will give the defendant a new trial based on the progress of science.
The court concluded that Powell’s new evidence was available at the time of the trial. The court noted that Powell could have retained Fredericks as an expert witness to testify at his trial, so the evidence was not new.
If Powell had retained his own expert, he might have been acquitted. While the failure to do so might support a claim of ineffective assistance of counsel, it did not satisfy the Texas standard for vacating a conviction based on newly available scientific evidence.
Powell attempted to shoehorn Knox’s change of opinion into the statute by arguing that his new opinion was based on an advanced understanding of the appropriate scientific method. The court concluded that Knox based his revised opinion on different data (he chose different measurement points when he prepared his second opinion) and that a change of data is not the same as a change in science.
It should be apparent to even a casual observer that Powell’s conviction rested in large part on a questionable expert opinion. The case illustrates the need for defense counsel to counter expert evidence with the defendant’s own expert in every case.
Fortunately for Powell, the trial court recognized other ways in which Powell’s trial was unfair. The prosecution relied on a jailhouse snitch but withheld information from the defense that created substantial doubt about the credibility of that witness.
Nor did the prosecution disclose the promises it made to the snitch in order to induce his testimony. When the snitch testified that no such promises had been made, the prosecutor did nothing to correct his perjured testimony, in violation of Powell’s right to due process of law.
The trial court recommended that Powell receive a new trial. On the basis of that recommendation, the Texas Court of Criminal Appeals granted Powell’s habeas corpus petition.
In a just world, the Bell County’s District Attorney would concede that the evidence falls well short of proof beyond a reasonable doubt that Powell committed the crime and would drop the charges. Unfortunately, the DA has refused to acknowledge that his prosecutors engaged in misconduct to convict Powell and has promised to move forward with a new trial.