Improper conduct: How the harmless error doctrine lets prosecutors’ mistakes slide
The testimony came nine days into the murder trial that would send Deonte Baber to prison for much of his life — testimony that was improperly introduced by the state’s key witness.
The Cincinnati man was accused of shooting a motorist who had accidentally hit a child. Hamilton County prosecutors built much of their case against him on grainy security-camera footage of the crime scene.
During Baber’s trial in May 2019, as prosecutors questioned a city homicide detective about that video, the detective testified that he had zoomed in on the footage, and saved screenshots showing the flare of gunfire.
“You could actually see the muzzle flash come out of the gun,” the detective told the jurors, implying that Baber was the blurry silhouette approaching the victim’s car.
In reality, there was no flash on the video.
Appellate judges later found this testimony, which unfairly implicated Baber, was inadmissible and false. But, relying on a controversial legal doctrine known as “harmless error,” they ruled that this and other improper conduct at trial was not egregious enough to warrant a new one.
Baber is one of scores of defendants in Ohio from 2018 to 2021 whose claims of prosecutorial misconduct were found to have merit, according to an analysis of state appellate decisions. But in the majority of those cases, appellate judges deemed what had happened at trial “harmless,” meaning it did not require a reversal. As a result, most of these defendants, including Baber, were left with no relief. Meanwhile, the prosecutors responsible for the improper conduct faced few, if any, repercussions.
The harmless error doctrine embodies the principle that defendants have a constitutional right to a fair trial, not a perfect one. Critics and proponents alike agree it is essential to prevent gridlock in already-congested court dockets.
But legal scholars have long debated whether the doctrine’s broad use is appropriate. Some doubt whether judges can apply it objectively; others question its uneven application among courts. Today, many experts said appeals courts across the country have expanded the doctrine so far beyond its original intent that some defendants are deprived of fundamental due-process protections.
“The effect of harmless error is to say, ‘there’s been a violation of rights, but we’re going to not [sic] do anything about it,’” said John Greabe, of the New Hampshire School of Law, a problem that he called “pervasive” in courtrooms nationwide.
Greabe said the doctrine has become so “regularly argued and applied in the appellate system” that the kind of improper conduct at trial not subject to its legal analysis is getting smaller and smaller.
In Ohio, there were roughly 4,700 criminal trials statewide between 2018 and 2021. During that four-year period, nearly 450 appeals included an allegation of prosecutorial misconduct, an investigation by Columbia Journalism Investigations, NPR and NPR member station WVXU and The Ohio Newsroom found.
Appeals involving prosecutorial misconduct are rare, but in Ohio about one in four claims ended in a ruling of improper conduct in that time — a ratio that suggests a systemic problem, experts said. Most of those confirmed cases of improper conduct – nearly 80% – were ruled harmless.
In Hamilton County, where improper conduct was affirmed at a higher rate than in any other major Ohio county, appellate judges followed the statewide trend: They found the conduct to be harmless in 17 of 22 total rulings from 2018 to 2021. But, as dozens of interviews and a review of select case records show, the improper conduct may not be so harmless.
For defendants like Baber, who has served four and a half years of his 18-years-to-life sentence so far, the consequences seem anything but.
“I just feel like I’m dying,” he said in an interview at Belmont Correctional Institution, about 100 miles east of Columbus. “Like I’ll never get out again.”
‘A perversion of the original theory’
Hamilton County’s appeals court ruled that Baber did not deserve a new trial in light of what it called “powerful and probative” evidence of guilt. But had the court considered his case 50 years ago, it may have reached a different conclusion.
Prosecutors are bound by legal and ethical standards meant to guarantee the right to a fair trial as established by the U.S. Constitution. In Ohio, prosecutorial misconduct occurs when they break those rules so egregiously that a court must overturn the conviction. Prosecutors can also engage in improper conduct that breaks those rules but does not warrant a reversal. That behavior can include everything from withholding evidence that would help defendants prepare their case to violating their due-process rights through inappropriate comments at trial.
Higher courts across the U.S. began using a harmless error standard to evaluate defendants’ appeals in the late 1800s, after problems arising from America’s adoption of the English common-law system had created a backlog of retrials.
Early on, the American legal system treated most trial mistakes, including those made by prosecutors, as though they eroded a defendant’s rights, according to researchers. Judges reversed convictions as a matter of course. But that led to what one scholar described in a 2010 article as “absurd” outcomes — granting a new trial for a convicted murderer whose indictment had grammatical errors, or vacating a thief’s conviction whose indictment did not specify that the stolen money was American.
The ensuing public criticism, as well as courthouses’ growing case dockets, led officials to adopt rules meant to limit what could prompt a retrial.
“Harmless error emerged not as a matter of principle, but really as a matter of judicial economy and kind of public relations,” said Daniel Medwed, a professor at Northeastern University School of Law, who has written about the doctrine and its effect on prosecutorial misconduct.
In the mid-1940s, after some states implemented their own rules, the U.S. Supreme Court set out to clarify how the harmless error doctrine should be applied. Under the legal analysis, appellate judges were instructed to determine whether a trial mistake occurred and, if so, was it severe enough to have affected the trial’s outcome. In 1967, at the height of the civil-rights era, the formal doctrine was established by the Supreme Court case “Chapman v. California.” It required prosecutors to prove “beyond a reasonable doubt” that the improper conduct did not contribute to the guilty verdict.
Since then, the court has issued several rulings that upheld the Chapman decision while seeming to contradict it. What has resulted in practice is a loose standard that instead weighs improper conduct at trial against evidence of guilt, experts said.
“It’s sort of a perversion of the original theory,” Medwed said, explaining this shift places the burden of proof on defendants.
The Supreme Court also expanded what can be ruled harmless. Misdeeds that once would have prompted a new trial now can be disregarded when appellate judges agree the evidence of guilt is sufficient, experts said. Misdeeds like not allowing a defendant’s attorney to question a witness. Or using coerced confessions. Or implying that a defendant’s silence indicates guilt.
That was the case for Daryl Robinson, a Cincinnati-area man who was convicted of murder and other charges stemming from a 2016 drug-related shooting. During the February 2017 trial, then-Hamilton County Assistant Prosecutor Michael Peck asked questions that led a city homicide detective to mention Robinson’s pre-trial silence.
Then, during closing arguments, the prosecutor told the jury that Robinson had had eight months before trial to consider what “bull crap lies” to say on the witness stand.
After Robinson appealed, in 2019, the three-judge panel deciding the case ruled that Peck’s conduct had “impermissibly encouraged the jury” to infer guilt from Robinson’s silence. But it did not require a new trial, the panel said. One of the judges, Charles M. Miller, disagreed.
“When you have a prosecutor commenting on someone’s right to remain silent, that’s almost always going to be, in my mind, a circuit breaker,” said Miller, a former appellate judge in the first district who wrote the dissenting opinion in Robinson’s case. “That’s just wrong.”
Medwed, who read the Robinson ruling at CJI’s request, said the burden should have been on the government to prove beyond a reasonable doubt that the constitutional mistake did not contribute to the guilty verdict. But in his view, the appeals court did not do that.
“Gradually we’ve shifted away from that model,” Medwed said. As a result, he said, “we’ve given prosecutors a little too much of a buffer, a little bit too much protection in our system.”
The Ohio Prosecuting Attorneys Association’s Louis Tobin disagrees. The director of the statewide lobby described the harmless error doctrine as a necessary tool for the criminal-justice system to correct unintentional errors made by prosecutors, defense attorneys and others at trial.
It is true that “there’s no bright line about what is and isn’t harmless,” he said, and that much of it comes down to what individual appellate judges think about the significance of a mistake on the trial’s outcome.
“Ultimately, harmless error is a simple matter of judicial economy. We can’t retry cases over and over again,” said Tobin. In most instances, he said, “harmless error is used exactly how it’s supposed to be used.”
Even critics recognize the harmless error doctrine helps reduce the backlog in overcrowded dockets, as well as the cost of redoing cases for minor errors.
But today the doctrine amounts to “a very, very big shield,” said Bennett Gershman, a former prosecutor who teaches at Pace University’s Elisabeth Haub School of Law, in New York. “If the evidence of guilt is sufficient, the court will say that the misconduct by the prosecutor really didn’t matter that much.”
‘Disregard’ that testimony
Hamilton County prosecutors suspected someone else before charging Baber in the deadly shooting of a driver in March 2017. The driver had struck and injured a 4-year-old in Cincinnati’s Walnut Hills neighborhood. On the surveillance footage that would become central in the Baber case, the boy’s father is seen dragging the motorist out of the car and beating him.
During the commotion, someone shot and killed the driver.
The boy’s father called 911. And when he thought he had hung up, he made a startling admission. “I killed him,” the father told his son, according to court records. “He dead, so you’ll be good. Your daddy got you.”
Police soon arrested the father and traced gunpowder residue on his hands. But, according to the state’s ballistics expert, he could not have been the shooter because of the angle that the bullet pierced the driver’s door. The father recanted his confession, and then became a key witness for the prosecution.
The security footage also showed a blurry figure, dressed in an orange and blue sweatshirt, running up to the driver’s car. A detective familiar with the neighborhood, working with a description of the shooter provided by the boy’s father, found a photograph of Baber wearing a similar jacket posted on his Facebook profile. While no weapon tied Baber to the shooting, another witness — who did not see the shooting — said she had recognized him wearing an orange and blue jacket.
Baber’s grandmother remembers the jacket was a local brand popular among neighborhood residents.
“Everybody that could afford the sweater had the sweater,” said Debra Headen, sitting in her living room, surrounded by plants that her grandson liked to lay beside and watch grow.
That’s the kind of kid her grandson was, she said. Quiet. A good kid. A smart person who was emotionally close with his mother and sisters. Not the type to murder someone.
“They really railroad him,” she said. “They really did.”
During Baber’s trial, when the homicide detective testified about the video, he insinuated that Baber was the fuzzy figure running toward and away from the driver’s car. He then mentioned the muzzle flash.
Michael Peck, the same prosecutor in the Robinson case, was questioning the detective. The prosecutor then referred the jury to exhibits of the detective’s screenshots, according to the trial transcripts. Prosecutors had turned over the video before trial but defense attorneys accused them of not disclosing those photos, as required by law. Nor did the photos depict a muzzle flash.
Baber’s defense attorneys sought a mistrial. The trial judge gave the jury instructions to forget the claim: “Apparently we don’t have a picture of what the detective just talked about, which was a muzzle flash,” the judge told the jurors. “You should disregard that part of the testimony.”
The judge disagreed that the proceedings had been compromised and declined to grant a mistrial.
The Hamilton County Prosecutor’s Office said in subsequent court proceedings that the prosecutors did not solicit the testimony. Nor did they have advance notice of the false evidence.
Peck, now a municipal judge in Cincinnati, did not respond to multiple phone calls, emails and a letter seeking comment for this story.
His co-counsel on the case, Ryan Nelson, a former Hamilton County chief assistant prosecutor who works in private practice, said he did not remember any issues involving a muzzle flash. He said that he and Peck turned over a copy of the video to the defense before trial and thus did not have to turn over screenshots taken from it.
“There was nothing to my knowledge that happened . . . that resulted in some sort of miscarriage of justice,” said Nelson, who defended the Baber jury’s decision.
But a subsequent request by the jury to magnify the video, made before its final deliberation, suggests that some jurors might not have followed the judge’s instructions to ignore the false testimony about the muzzle flash. About three days later, Baber was found guilty of murder.
“It’s the whole ‘Don’t think of a white elephant,’” said Wendy Calaway, an associate criminal-justice professor at the University of Cincinnati and criminal defense attorney. “And, of course, everyone’s thinking of a white elephant.”
In April 2021, the district appeals court concluded that the muzzle-flash testimony was likely fabricated and damaging to Baber. But while acknowledging the jury had asked to zoom in on the video, the court found the inadmissible evidence had not influenced its verdict. The court ruled that this and other improper conduct — including prosecutors withholding the identities of eight witnesses from the defense — were harmless.
Experts, like Medwed and Greabe, said the jury’s request to magnify the video alludes to a fundamental problem with the harmless error doctrine: That appellate judges use written transcripts and other case records to infer how the misconduct at trial may have contributed to a juror’s decision. Under the analysis, an appeals court must assume the jurors followed a trial judge’s instructions to disregard something improper. But that does not mean they actually did.
Juries don’t have to explain what evidence they rely on or why they ask to see certain evidence again. In Baber’s case, the jurors never said why they wanted to zoom in on the video. But given the detective’s false testimony and the facts of the case, Greabe considers it “the most likely” explanation.
“It’s the government’s obligation to exclude any possibility that the jury was affected by these errors,” said Greabe, who read the decision at CJI’s request. The court’s ruling, in his words, “falls considerably short of being persuasive on that.”
In an email, a spokesperson for the First District Court of Appeals declined an interview on behalf of the judges who decided Baber’s case. “The Court speaks through its entries,” the spokesperson said.
In Baber’s eyes, the misdeeds stole more than his freedom.
“I feel like I just got robbed out of my life and my son’s life,” he said.
Judges’ inability to agree on harmless error
The harmless error doctrine’s uneven application has meant that appellate rulings on prosecutorial misconduct can differ from one courtroom to the next, based on the judges’ understanding of the legal analysis. Even when presented with the same facts, appellate judges hearing the same case can reach strikingly disparate conclusions.
That is what happened to Andrew Lavender, a 25-year-old Cincinnati native whose murder conviction was recently upheld, despite disagreement among appellate judges on his prosecutorial misconduct claims.
Lavender was a sophomore at Robert A. Taft Information Technology High School. He was a good student with no disciplinary issues, an impeccable attendance record and no criminal history in August 2014, when he was accused of fatally shooting a father of five. One eyewitness, placing a call to 911, described the shooter as a middle-aged man. But detectives arrested Lavender, then 16, after a police informant said he had overheard the teenager bragging about “taking a hit” on the victim.
Three years later, Lavender would be tried as an adult.
During the December 2017 trial, then-Hamilton County Assistant Prosecutor David Prem flouted a court rule barring the use of unrelated evidence to invoke a person’s bad character: The veteran prosecutor relied on an old photograph and text messages not connected to the shooting to paint Lavender as a gangster involved in an alleged murder-for-hire plot, according to trial transcripts. One photograph depicted Lavender brandishing a pellet gun aimed at the camera; in his other hand was a smaller gun that prosecutors claimed could have been the same kind of .22 revolver they speculated was used in the murder.
In opening remarks, Prem told jurors that killing people was Lavender’s “family business,” according to trial transcripts, suggesting that his older brother, who was already convicted of murder, had been a hitman.
The prosecutors also showed the jurors dozens of months-old text messages in which Lavender mentioned selling drugs and firing a gunshot into a car to argue that the teen had become increasingly desperate in an attempt to make money. But records show none of those alleged incidents were linked to the murder.
Then, in closing remarks, Prem told the jury that those texts revealed who Lavender truly was, again improperly alluding to the defendant’s character. Lavender was found guilty of murder, and sentenced to life in prison without parole.
Tina Lavender, Andrew’s mother, gets what she called “a bad taste” in her mouth when remembering Prem and how, in her words, “he didn’t want to listen to the truth.” She described her youngest son as her “golden child,” a caring teenager who excelled at basketball and was beloved by his school.
When Lavender appealed his conviction in January 2018, he alleged that Prem’s comments and the use of outside evidence had violated his constitutional right to a fair trial. Nearly two years later, a three-judge panel issued contradictory rulings that both confirmed and discounted the alleged misconduct.
In the majority opinion, Russell Mock — a former appellate judge who has since returned to his job at the Hamilton County Prosecutor’s Office — dismissed Lavender’s misconduct claims. He concluded that the text messages were appropriately used to show motive for the killing, not an unrelated pattern of bad behavior.
But his two colleagues deemed those same acts improper. In a concurring opinion, retired appellate judge Beth Myers wrote that Prem’s reliance on inflammatory comments, old photos and text messages constituted improper character evidence. Still, she found this improper conduct would not have altered the trial’s outcome — it was harmless, she said.
Hamilton County Appellate Judge Pierre Bergeron dissented. Unlike his judicial colleagues, Bergeron ruled that Prem’s conduct was too egregious to “be dismissed as harmless,” as his December 2019 decision states, and tainted Lavender’s original trial enough to warrant a new one.
“Before we send this defendant to prison for the rest of his life, I would afford him a new trial limited to evidence probative of whether he committed the murder at hand,” Bergeron wrote.
Prem, who recently retired after a 36-year career, declined to comment for this story.
Experts agree that the contradictory rulings in Lavender’s case — which lacked consensus on what constituted the error and the harm — show just how subjective the harmless error doctrine is. Case Western Reserve University School of Law professor Andrew Pollis, an expert in appellate law who read the Lavender ruling, said that he could “go on for days” about the appeals he has seen that involve judges’ inability to agree on harmless error.
Pollis said, in such circumstances, the question is “how we can tolerate a system in which this key question of prejudice always seems so pivotal, and yet is so subjective?”
Reached by phone, Mock said that he “vaguely” remembered Lavender’s case and declined to comment further. Bergeron, through the appellate court spokesperson, declined to comment. Myers was not available for an interview for this story.
Lavender, who is serving his life sentence at London Correctional Institution, about 31 miles southwest of Columbus, remembers his trial as a surreal experience. Dressed in the inmates’ royal blue uniform, he said he gets a “scary feeling” when he replays in his mind all the accusations that the trial court judge allowed in the courtroom.
“I felt that it just solidifies they can do whatever they want,” he said, referring to prosecutors. “Whatever they feel they need to do to get a conviction.”
‘Who’s gonna care about’ the rules?
Across Ohio, there were 88 prosecutors whose conduct at trial resulted in 80 harmless error rulings from 2018 to 2021, according to the analysis by CJI and its partners. More than 20% of those rulings — and 20% of those prosecutors — were in Hamilton County.
One such prosecutor was David Prem. In 1999, two decades before Lavender’s appeal, the Ohio Supreme Court admonished Prem and his co-chair, calling them out by name for making inflammatory comments during a death-penalty trial. The court found the inappropriate comments harmless but warned that “if this kind of activity continues, it is just a matter of time before it affects the outcome of a trial.”
At the time, two justices rebuked not just Prem but prosecutors across Ohio. In the dissenting opinion, written by then-Chief Justice Thomas Moyer, the justices wrote that many of those attorneys were intentionally engaging in improper conduct at trial “safe in the belief that this court will continue to protest with no consequences,” their September 1999 decision states.
It was an unprecedented warning for Prem: Experts say appellate judges rarely identify in their rulings the prosecutors responsible for improper conduct. In the few instances when an appellate judge has complained about a prosecutor’s misdeeds, the judge did not name the prosecutor, the analysis by CJI and its partners shows. None of the other Ohio prosecutors involved in the 80 total harmless error rulings have been publicly reprimanded by the courts, according to a review of state appellate decisions dating back to 1992.
By the time Prem handled the Lavender case — three decades into his tenure at the Hamilton County Prosecutor’s Office — the longtime prosecutor had amassed five findings of improper conduct, court records show.
Prem was never disciplined for any improper conduct rulings — and it seems neither were his colleagues. CJI and its partners obtained the personnel files of 13 of the 18 Hamilton County prosecutors. None contained any mention of misconduct claims or documented disciplinary action related to improper conduct at trial.
The only prosecutor who faced any discipline was Ryan Nelson, the former prosecutor who handled the Baber case, partly because of his repeated failure to share evidence with defense attorneys before trial, records show.
Nelson acknowledged that he was “bad at a lot of things,” and said the office offered little training on legal and ethical standards, leaving him to figure things out for himself. Nearly a dozen other former prosecutors said the office rarely discussed improper conduct rulings with those prosecutors responsible for the actions.
The Hamilton County Prosecutor’s Office did not respond to multiple interview requests and a list of written questions for this story.
The Ohio Prosecuting Attorneys Association’s Tobin said prosecuting attorneys across the state want their assistant prosecutors to be well trained. But no one should be disciplined for an unintentional error, he said.
“Every elected prosecutor across the state would say, if they’ve got somebody on their staff who’s made a mistake or is making mistakes, they want them to get some training and get that stuff corrected,” Tobin said. “But to rise to the level of discipline for an error is not a system that we want to encourage.”
It’s the kind of systemic response that should concern the public, said Daniel Epps, a law professor at Washington University School of Law in St. Louis, Missouri, who has written about the harmless error doctrine. Epps believes the widespread use of the doctrine has done little to encourage prosecutors to follow the rules and safeguard defendants’ rights. And that puts the integrity of the criminal-justice system at risk.
“If we tell prosecutors over and over, ‘Look, here are the rules. But, by the way, if you break the rules . . . nothing will happen to you and your convictions won’t even be overturned because they’re all going to be harmless anyways,’ who’s gonna care about [the rules]?” Epps asked. “What incentive are you giving prosecutors to actually care?”
The question can haunt the family members of defendants, like Baber, whose claims of prosecutorial misconduct have ended in harmless-error rulings. Baber, now 31, has reached the legal limit for appealing his conviction, and has little chance of getting out of prison before 2037.
“It hurts,” said his sister, Destiny. “It’s like a piece of something is missing without him.”
For Baber, one of the most painful things is being separated from his six-year-old son who was born after his incarceration. Last spring, Baber met his son in person for the first time when his grandmother and sister brought the boy for a visit.
Baber’s grandmother still remembers how the prison’s visitation window divided the pair.
“When the child saw him, he wanted to definitely touch his dad,” Headen said. “The only time they have touched or seen one another is through that glass.”
This story is a collaboration from NPR’s Station Investigations Team, which supports local investigative journalism; Columbia Journalism Investigations, an investigative reporting unit at the Columbia Journalism School in New York; and NPR member station WVXU in Cincinnati.
Nick Swartsell is a reporter at WVXU/Cincinnati Public Radio. Patricia Martinez Sastre and Cameron Oakes reported this story as fellows for CJI. CJI fellows Gabriela Alcalde and Jake Kincaid, CJI research assistants Frances Howe and Jake Millman, and NPR senior producer Robert Benincasa and Roy W. Howard fellow Tirzah Christopher contributed to the data analysis. The Ohio Newsroom, a collaboration among Ohio public radio stations, provided editing and other support.