Siding With Junk Science Over Justice

On October 17, 2024, Robert Leslie Roberson III was scheduled to die by lethal injection in Texas’s death chamber for murdering his two-year-old daughter Nikki in 2002. Had he been put to death, he would have been the first American executed because a jury had accepted the “shaken baby hypothesis”—the argument that a young child suffered irreversible brain damage as a direct result of being shaken.

With his future still hanging in the balance—his conviction is raising important questions that have implications far beyond his own case. Perhaps the biggest question of all is whether a medical diagnosis can prove that a crime has taken place. Roberson’s story—and the scientific innovations that have occurred since Nikki’s death more than 20 years ago—are giving an increasing number of experts pause. His case is forcing us to confront the stark problem of why some courts—the U.S. Supreme Court among them—seem to show a greater commitment to junk science than to ensuring that justice be done.

The Tragedy of Bad Timing

Even long before his conviction, Roberson had led a miserable life. He’d dropped out of high school, washed out of the military and had nothing going for him. Though a grown man and a father of three children, he lived a life characterized by prostitutes, drug abuse and poverty.

His daughter Nikki’s mother was a homeless drug addict, so Nikki’s care fell to Roberson, who was hardly any more up to the task. Moreover, the two-year-old had a range of health issues, including breathing problems, and inability to hold food down and terrible bouts of diarrhea. Her temperature would spike—sometimes as high as 104.5 degrees. Roberson took her to doctors and gave her the medicine prescribed to her, but she remained sick.  

One day, Roberson walked into Nikki’s room and found her on the floor unconscious, blue and unresponsive. Blood was on her mouth. He took her to the emergency room and explained to the hospital staff that Nikki had fallen off her bed. While medical staff’s reaction to Nikki’s tragic death was strong and visceral, Roberson’s was much more muted. Roberson had a “flat affect” and seemed oddly indifferent to Nikki’s death. They began to suspect him. Police did, too. 

But the path to Roberson’s conviction was paved with questionable assumptions and problems. First, prosecutors didn’t know that Roberson is autistic and that his behavior was perfectly consistent with his condition. Roberson himself didn’t know, either: He wasn’t formally diagnosed until 2018, long after his arrival on death row. Therefore, he wasn’t able to explain his ostensibly strange behavior to them.

Then, during Roberson’s trial, uncorroborated claims dealt significant blows to his case. His current lawyers write that, despite lacking the proper training and credentials, an emergency room nurse examined Nikki for sexual assault and prejudiced the jury against Roberson by suggesting he had anally raped her. After the trial, the nurse’s claim was repudiated by qualified medical authorities.

Additionally, according to an account from the Texas Court of Criminal Appeals in 2007, Roberson’s live-in girlfriend, Teddie Cox, told the jury that Roberson beat Nikki and regularly shook her, leaving bruises. However, Cox may have had ulterior motives. Roberson wasn’t the first lover of hers accused of molesting children: Her daughter Rachel’s father was a pedophile doing time for molesting the young girl. Cox believed her parental rights to her child were in jeopardy, and giving prosecutors what they wanted might help her situation. Indeed, her description of Roberson as a sadist was precisely the tune prosecutors wanted the jury to hear. 

It was Nikki’s supposed cause of death, though, confidently explained to the jury by the state’s forensic experts, that sealed Roberson’s fate. She could’ve only been shaken to death, they said, because her autopsy revealed three intracranial injuries that, when found together, comprise the Shaken Baby Syndrome (SBS) triad: retinal hemorrhaging (broken blood vessels in the eyes), subdural hemorrhaging (severed “bridging” veins between the brain and the skull) and a swollen brain.    

In 2003, finding the SBS triad during an autopsy usually made getting a murder conviction a slam dunk. As one scholar put it, the presence of the triad was “in essence, a medical diagnosis of murder” and SBS triad testimony led to hundreds of convictions every year. By the time Roberson’s trial was over, everyone in the courtroom but Roberson, including his own lawyers, worked from the premise that he’d shaken Nikki to death. What had been accepted as scientific truth simply didn’t allow for any other explanation.   

However, since the time of Roberson’s trial, science has advanced dramatically. Many experts now believe the hypothesis of the SBS triad to be flawed, if not altogether false. SBS experts filed a brief with the U.S. Supreme Court in Roberson’s case last year explaining that “the scientific understanding of SBS has shifted seismically” and, as a result, “the triad-based diagnosis must be considered scientifically invalid.” Far from clearly showing child abuse, the experts claimed, we now know that “illnesses, medical disorders, and accidents” can “mimic” acts of violence. In a recent article on Roberson’s case, The Guardian reports that “more than 80 alternative non-violent causes of the symptoms have been identified, including short falls and illness—both of which were evident in Nikki’s case.” 

Roberson was doomed by bad timing. He had the misfortune of being tried during a dark age of child head trauma research. So were thousands of others, according to an oft-cited article by Northwestern University’s Deborah Tuerkheimer. She writes that decades of convictions based on now-debunked SBS medical testimony have caused a full-blown “criminal justice crisis” that will require a paradigm shift similar to the use of DNA in criminal forensics to be fully overcome. In the early 1990s, the Innocence Project began working to exonerate the wrongfully convicted via the new technology of testing evidence for DNA. The Innocence Project itself seems to agree that a similar effort driven by better scientific understanding is needed today: Its lawyers are now representing Roberson.

Follow the Science?

Unfortunately, the recommendations of medical experts have not been enough to convince the U.S. Supreme Court: Last year, the court decided not to review Roberson’s case, which kept his death sentence in place. If the court’s treatment of Shirley Ree Smith 12 years earlier is any guide, he wouldn’t have fared well even if the the high court had taken his case.

In 1997, prosecutors in California convinced a jury that Smith had shaken her seven-week-old grandson, Etzel Glass, to death. Though Smith’s trial had taken place when the SBS triad hypothesis was still widely accepted, the forensic evidence against Smith was particularly flimsy and confusing. Rather than concede that she’d shaken Etzel, the defense attempted to challenge it.

The prosecution’s medical experts claimed that aggressive shaking had severed Etzel’s brain stem, causing his death. However, during their examination they found neither the bleeding nor the swelling normally associated with SBS. The examination didn’t reveal a severed brain stem—not even a torn one. To account for the absence of these injuries, doctors testified that Smith had shaken her grandson so violently, and severed his brain stem so instantly, that “death occurred too quickly for visible trauma to develop.” In other words, this was a killing of such extraordinary rage that the baby expired before his brain had time to bleed or swell. So the state’s experts insisted to the jury that Etzel’s death was caused by injuries inflicted via his grandmother’s shaking—despite the fact that these experts never actually observed such injuries.

Even though medical experts who testified on Smith’s behalf gave strong rebuttals, the jury sided with the state’s witnesses and sent Smith—who had neither a criminal record nor a plausible motive to kill her grandson—to prison.    

Smith then argued to the U.S. Court of Appeals for the Ninth Circuit that even when interpreting the evidence presented against her in a light favorable to the prosecution, no reasonable juror could have found her guilty. She won. Her victory initiated an extraordinary back-and-forth battle between the Ninth Circuit and the Supreme Court—the former insisting she be freed, the latter that her conviction stand—that brought this California case to the justices in Washington, D.C., three times.

The Ninth Circuit recognized that Smith had been convicted by junk science, was almost certainly innocent and should be freed. Its findings couldn’t have been more compelling and straightforward: “Nothing in the physical evidence supported the prosecution experts’ testimony as to the cause of death.” Because the Constitution requires proof beyond a reasonable doubt before a jury can convict, and this jury couldn’t have rationally concluded that the evidence against Smith met that standard, Smith’s conviction ought to be reversed.   

But in its final ruling in this case, the Supreme Court, at the urging of then-California Attorney General Kamala Harris, decided to order Smith back to prison. Despite recognizing that “[d]oubts about whether Smith is in fact guilty are understandable,” the court wrote, the lower court’s determination to reverse Smith’s conviction “cannot be allowed to stand.” Smith’s trial had been fair, it maintained, and the Ninth Circuit had abused its authority by second-guessing her jury.

In her dissent, Justice Ruth Bader Ginsburg wrote that the court was trying “to teach the Ninth Circuit a lesson.” Her dissent both admonishes her colleagues for needlessly intervening in Smith’s case and explains that since Smith’s conviction in 1997, “[d]oubt has increased in the medical community” about SBS and that “in light of current information, it is unlikely that the prosecution’s experts would today testify as adamantly as they did in 1997.”

In both Roberson’s and Smith’s cases, the juries were almost certainly swayed by junk SBS science and overzealous prosecutors repeating the discredited findings of their experts. After their convictions, both Smith and Roberson argued that their constitutional right to a fair trial had been violated because the latest medical discoveries made it obvious that no rational juror would believe that prosecutors had proved beyond a reasonable doubt that their alleged victims had died from being shaken.

Fortunately for Smith, California’s governor at the time, Jerry Brown, was exceptionally committed to using his clemency powers to right the wrongs of his state’s criminal justice system, and in 2012, he granted her clemency. However, Roberson is from Texas, where those who administer clemency aren’t so scrupulous. Prior to his case, the Texas Board of Pardons and Paroles had recommended clemency in only one of 85 death penalty cases over the past decade. In that extraordinary case, Governor Greg Abbott accepted the board’s recommendation and commuted the sentence of Thomas “Bart” Whitaker, whose co-defendant, the actual shooter, received a lighter sentence and whose father, a surviving victim, had pleaded desperately for his convicted son’s life. The board unanimously rejected Roberson’s request. 

Despite his core constitutional claim already having been rejected at the highest levels, Roberson kept fighting to save his life in the courts. He filed another petition to the Supreme Court shortly before his October 17 execution date, which was rejected on jurisdictional grounds. However, the court’s most consistently antideath penalty justice, Sonia Sotomayor, attached a statement to the rejection sympathizing with Roberson and lamenting that his execution would perpetrate an outrage. “Current postconviction remedies often fail to correct convictions ‘secured by what we now know was faulty science,’” she wrote. “This case is emblematic of that problem.”     

A Refusal to Reform

Looking back, Roberson’s conviction should have been reversed for at least two reasons. First, there wasn’t enough credible evidence presented at his trial to sustain a conviction. Second, the junk science of an SBS diagnosis compelled his attorneys to say that Roberson had shaken Nikki. Not even truth remained as a viable defense. 

Preserving convictions based on an outdated understanding of SBS isn’t the only example of the Supreme Court failing to appreciate the exonerating power of new scientific developments. Even when confronted with revolutionary forensic discoveries that negate old methods, like genetic fingerprinting, the Supreme Court has chosen to deny requests from defendants for increased constitutional protections, leaving it to Congress and the states to implement reforms. In 2009, for instance, the state of Alaska refused to test the DNA on a piece of evidence in its possession that might have exonerated an inmate serving time for rape. The Supreme Court ruled that the inmate had no constitutional right to test that evidence, even at his own expense. Such rights, if they exist at all, were for Alaska to confer. Fortunately, Congress and the states have begun to rise to the occasion, doing what the highest court will not.

There have been some hopeful episodes in lower courts, as well, that suggest that future defendants may enjoy better protections in this area. DNA testing has advanced our understanding of how and when forensic evidence should be used in courtrooms. These tests are often smoking guns that absolutely contradict prosecutors’ theories of how crimes were committed by proving someone other than the accused left incriminating evidence at the crime scene. They’ve proved themselves to be irrefutable enough to transform at least one once-trusted science—forensic dentistry—into a pseudo-science and have jolted the entire criminal justice system by generally exposing forensic science and forensic scientists as less reliable than previously believed.

But the DNA Revolution hasn’t been able to directly help wrongfully convicted SBS inmates as yet. Those who shake babies are far less likely to leave testable biological evidence at the crime scene than rapists and those who murder using other methods, so DNA testing hasn’t been able to explode the SBS triad hypothesis as it has forensic dentistry and other unscientific forms of evidence. Instead, the defense is often compelled to resort to complicated anti-SBS medical testimony that, however accurate and impressive, lacks the same instant certainty and flair. 

Nor is it as obviously conclusive. The mountains of scholarly and popular literature attacking the SBS hypothesis doesn’t seem to have persuaded most of America’s pediatricians yet. In fact, studies show there’s still a long way to go: Of the nearly 3,600 exonerations listed on the National Registry of Exonerations, so far only 34 are SBS cases. 

Earlier this year, Roberson tried again with the Texas Criminal Court of Appeals. In a 159-page longshot petition, he claimed that, using the latest science, he can “explain precisely how Nikki died.” “Precisely how” may be a bit strong: He hadn’t discovered DNA from the actual perpetrator at the crime scene or a video recording of Nikki accidentally falling to her death. But his theory is certainly plausible enough to warrant stopping his execution and overturning his conviction.

He claims that Nikki’s death resulted from a convergence of factors: a severe case of pneumonia that had inflamed into sepsis; an impact from landing on her head after falling from her bed; and ingesting too much promethazine and codeine, both of which had been prescribed to her by doctors. (The FDA now restricts the use of these drugs to children Nikki’s age because it can weaken their respiratory systems.) Nikki’s system shut down completely. Her pneumonia and medication caused respiratory failure, which, like her head injury after falling from her bed, caused the intracranial bleeding that led to her SBS diagnosis. Her brain was de-oxygenated when it hit the floor, so the contents of her skull were bloody and swollen.

The court rejected his argument.

Let Justice Be Done ...

Considering the evidence presented during his trial, it’s hard to think that any jury would convict Robert Roberson today. And frankly, the fact that he remains on death row shows how our courts have yet to fully realize and accept when widely accepted science becomes junk science. 

The Supreme Court could have helped with the SBS junk science problem, perhaps sparing Roberson years of agony, back in 2011 with a strong forward-looking opinion vindicating the rights of Shirley Ree Smith. Instead, by ordering her back to prison, even while acknowledging she likely hadn’t killed her grandson, it announced that the Constitution provided no protection to the thousands condemned by an outdated hypothesis, one proven wrong by modern science. It’s likely that many of these inmates are innocent parents, falsely labelled baby killers, struggling to be believed as they grieve their lost children. Roberson, whose life now hangs by a thread, remains one of them.

There’s an old Latin phrase about the law—“fiat justitia ruat caelum,” or “let justice be done though the heavens fall.” The phrase always acquires a special urgency in capital cases, like Roberson’s, and is doubly fitting in his case because it also perfectly describes the problem the Supreme Court seems to have with the particular form of junk science that keeps him locked up. Though our understanding of these cases has advanced and improved, the court refuses to fix errors based on a hypothesis that no longer make sense and has condemned innocent people to lifelong prison terms and death sentences. Recognizing the evolutionary nature of science, and that new discoveries require us to right old wrongs, would bring us much closer to the ideal of true justice.