Harmless Error: Explained
In our Explainer series, Justice Collaborative lawyers and other legal experts help unpack some of the most complicated issues in the criminal justice system. We break down the problems behind the headlines—like bail, civil asset forfeiture, or the Brady doctrine—so that everyone can understand them. Wherever possible, we try to utilize the stories of those affected by the criminal justice system to show how these laws and principles should work, and how they often fail.
In the spring of 2012, near the end of President Barack Obama’s first term, federal prosecutors in North Carolina were pursuing a case against Alejandro Garcia-Lagunas, whom they accused of trafficking large amounts of cocaine. At trial, defense counsel argued that Garcia-Lagunas, who had been renting a room in a small trailer for less than $350 per month, was a drug user, not a high-level dealer.
As the trial progressed, the government recognized that Garcia-Lagunas’s meager lifestyle was a potential weakness in its case. But rather than counter the defense with evidence, it resorted to a racist stereotype: A police detective testified that “Hispanic drug traffickers” live modestly because they send “the majority if not all of the proceeds back to their native countries.” Defense counsel objected, but the judge allowed the detective’s testimony, explaining that “based on his experience, most Latins [sic] send money home whether they’re drug dealers or not.” The government returned to the racist theme during closing argument, reminding the jury of the detective’s testimony, and arguing that Garcia-Lagunas did not “have an extravagant lifestyle” because Hispanic drug traffickers “package that money” and “send it back to their home country.” The jury returned a guilty verdict, and the judge sentenced Garcia-Lagunas to more than 15 years in prison.
On appeal, Garcia-Lagunas argued that the government’s use of a racist stereotype violated his constitutional right to a fair trial, and the government agreed. But Garcia-Lagunas never got a new trial. Around the same time that presidential candidate Donald Trump was verbally attacking a federal district judge for his “Mexican heritage,” the court voted to uphold Garcia-Lagunas’s conviction because the constitutional error was “harmless.”
For the judges in the majority of the 2-1 decision, the government’s heavy reliance on the detective’s racist testimony was not significant. They found the evidence of Garcia-Lagunas’s guilt to be strong, and his defense to be weak. So for them, it was “beyond clear” that the jury would have convicted Garcia-Lagunas, “even had they heard nothing of [the detective’s] improper testimony.”
The case of Garcia-Lagunas illustrates a disturbing feature of the American criminal system: Under a legal doctrine called “harmless error,” appellate judges routinely affirm convictions tainted by legal error whenever they feel confident that the person appealing the conviction is guilty. The doctrine receives far less attention than stories about the misbehavior of police and prosecutors, but as one of the most frequently invoked doctrines in all of criminal appeals, it has a profound influence on the behavior of everyone inside our criminal system, including police, prosecutors, defense attorneys, and judges.
One study of California death penalty appeals found that, during a 10-year period, roughly 90 percent of death sentences imposed by trial courts were upheld on appeal even though nearly three-quarters were infected by constitutional error. On the other side of the country, “harmless error” may soon decide the fate of 173 people on Florida’s death row, all of whom were sentenced in a way that the U.S. Supreme Court later decided was unconstitutional.
What kind of legal errors can be “harmless”? The long list includes the use of illegally seized evidence and illegally coerced confessions.
The precise rules of “harmless error” are confusing, and they depend on the type of error that has occurred. If the error does not affect a constitutional right, a court may reverse a conviction only if the error had a “substantial or injurious effect or influence in determining the jury’s verdict.” If the error does affect a constitutional right, the court must decide whether there was a “structural error” or a “trial error.” Structural errors, such as the denial of the rights to counsel, an impartial judge, or jury trial, can never be “harmless”; they result in automatic reversal. For trial errors, such as the admission of illegally seized evidence, a court must reverse unless the government proves the error was “harmless beyond a reasonable doubt.”
Over the last 50 years, the Supreme Court significantly expanded the list of trial errors, so a large number of cases require application of the beyond-a-reasonable-doubt test. Under this test, many courts mistakenly focus on whether the government’s untainted evidence would lead to a conviction in a hypothetical future trial. What they should focus on—and what many do focus on—is whether the government can prove the error did not affect the outcome of the trial that already happened. As the Supreme Court explained in Sullivan v. Louisiana, “the inquiry … is … whether the guilty verdict actually rendered in this trial was surely unattributable to the error.”
The Development of Harmless Error
Appellate review was not always so complicated. For much of American history, any error required reversal. The doctrine allowing a constitutional error to be “harmless” did not arrive until 1967. That year, in Chapman v. California, the Supreme Court created a couple of new rules: (1) constitutional errors can be “harmless,” and (2) constitutional errors require reversal unless the government proves the error was “harmless beyond a reasonable doubt.” The second rule placed a heavy burden on the government, suggesting that most errors would continue to result in reversal. The first rule came with an important caveat: the Court acknowledged that some errors were “so basic to a fair trial that their infraction can never be treated as harmless,” and it listed several examples of these structural errors, including the admission of a coerced confession.
The election of Richard Nixon shattered these limits on the Chapman decision. Earlier in the 1960s, with Earl Warren as chief justice, the Supreme Court decided several cases that expanded constitutional protections for people accused of crimes, including Mapp v. Ohio, Gideon v. Wainwright, and Miranda v. Arizona. During the 1968 presidential campaign, Nixon railed against these cases, and pledged to appoint judges who would combat them. After his victory, Nixon appointed four Supreme Court Justices, including Warren Burger, who replaced Earl Warren as chief justice, and William Rehnquist, who became chief justice in 1986 and remained on the Court until his death in 2005.
In general, the Burger and Rehnquist Courts did not radically alter the substance of constitutional protections established by the Warren Court, as many predicted they would. Instead, as Professor Carol Steiker explains, the Court “revolutionized the consequences of deeming conduct unconstitutional.” In the realm of “harmless error,” the Court diluted Chapman’s beyond-a-reasonable-doubt test and expanded the list of “trial errors.” So today, every error related to unconstitutional police conduct can be subject to “harmless error” review—even the admission of a coerced confession, which Chapman described as a structural error that could never be harmless.
The Problems With Harmless Error
There are many problems with “harmless error,” particularly when the doctrine is applied to constitutional violations.
1. Harmless error ignores important constitutional values.
Our country is haunted by 400 years of racism, yet Alejandro Garcia-Lagunas remains in prison because two judges thought the government’s racist stereotype was “harmless.” His case shows that a myopic focus on guilt can blind judges to broader social harm. By design, many constitutional rights protect values that are unrelated to truth and accountability—values like dignity, privacy, and equality. Guilt-based applications of “harmless error” ignore those values.
2. Harmless error encourages misconduct.
Years ago, during an appellate argument in the state of Washington, a judge asked a prosecutor why, in light of a rule that prohibits trial prosecutors from asking one witness whether another witness is lying, the prosecutors from his office continued to ask the prohibited question. “It’s always been found to be harmless,” he explained.
When the court published its decision, it rejected any suggestion that courts “wink at intentional and repeated unfair questioning by prosecutors under the rubric of harmless error.” But the court then affirmed the conviction, concluding that the prosecutor was “correct in relying on the doctrine of harmless error.”
None of this is surprising. By making the remedy for the violation of a legal right so difficult to obtain, “harmless error” erodes whatever protection the right is supposed to provide.
3. Harmless error freezes constitutional law.
In the 1990s, the Supreme Court ruled that “harmless-error analysis is triggered only after the reviewing court discovers that an error has been committed.” Yet courts routinely ignore this ruling, skip the question of error, and conclude that any error, assuming one occurred, was “harmless.” By refusing to decide whether a constitutional error occurred, courts fail to perform one of their most basic functions: clarifying the law that governs the actions of police, prosecutors, defense attorneys, and trial judges.
4. Harmless error usurps the fact-finding role of juries.
For many courts today, “harmless error” focuses almost exclusively on the perceived strength of the government’s evidence. If the government’s evidence appears strong, the court will not carefully consider the ways in which an error might have affected the jury’s verdict, as Chapman’s beyond-a-reasonable-doubt test requires. It will simply affirm the conviction.
By relying on the perceived strength of the government’s evidence, courts usurp the fact-finding role of the jury. This is a legal problem because the Sixth Amendment guarantees the right to a jury trial. And when a court determines that some of the government’s evidence was illegally admitted, the only existing jury verdict is tainted by that illegality: There is no longer any verdict that rests exclusively on legal evidence. So, if an appeals court sets aside the illegal evidence, and the remaining evidence allows any inference of innocence, there is a strong argument that the Sixth Amendment requires reversal of the conviction so that a jury, and not a group of appellate judges, can decide the question of guilt.
Focusing on the apparent strength of the government’s evidence also creates practical problems. The appellate judges did not attend the trial, and there is no way to ensure they read the entire trial transcript, which often spans thousands of pages. When they do read the transcript, they may fail to detect body language or sarcasm that changed the meaning of a witness’s testimony. And as Professors Keith Findley and Michael Scott explain, cognitive biases “contribute in powerful ways to a conclusion that the defendant was guilty.”
In many cases, that conclusion is wrong. In New York, for example, a court affirmed Jeffrey Deskovic’s conviction because he confessed to the crime, and his “multiple incriminatory statements” provided “overwhelming evidence” against him. But DNA later established Deskovic’s innocence. Police had coerced his confession.
5. Harmless error provides our punishment bureaucracy with a facade of legitimacy.
The United States is the world’s largest punishment bureaucracy. In numbers unprecedented in the recorded history of human life, we sweep people from their homes, push them through crowded criminal courts, and leave them to languish inside cages of metal and concrete.
The Constitution exists alongside the punishment bureaucracy, and to those inside the cages, it promised a lot: equal protection of the laws, due process, freedom from unreasonable searches. When an appeals court affirms a conviction, it signals to almost everyone in our society that we have kept our constitutional promises. And so the bureaucracy is able to retain mainstream respect, and few understand that the path to a government cage may have been littered with constitutional violations. As Judge Jerome Frank once wrote, “harmless error” transforms important legal protections into “beautiful verbal garlands concealing ugly practices we have not the courage … to contemplate.”
The Dubious Justifications For Harmless Error
The Constitution does not say that a violation of its protections can be “harmless,” but over the years, the Supreme Court has offered a couple of justifications for the doctrine it created. Neither is persuasive.
Claim #1: Harmless error is necessary to avoid the time and expense of retrials.
One of the primary justifications for “harmless error” is that it is necessary to conserve “scarce judicial resources.” In other words, trial judges are busy with a swollen docket of criminal cases, and if appellate courts reversed a lot of convictions, trial judges would have more work to do. As one scholar warns: “the resulting proliferation of retrials would significantly undermine judicial efficiency.”
One problem with this argument is that “harmless error” does not obviously reduce the amount of time that judges spend on criminal cases. It can save a trial judge from handling a retrial, but when applied correctly, with the rigor that its beyond-a-reasonable-doubt test requires, the doctrine shifts a huge amount of work to appellate judges. At each layer of review, at least three judges will hear the case, and each of those judges must examine the entire trial record to evaluate the intricacies of the government’s evidence, the defense evidence, and the possible effects of any legal error on the jury—a fact-intensive evaluation that is generally unnecessary when the only question for the judges is whether a legal error occurred.
A more serious problem is the assumption that trial judges need to process human beings into cages at their current rate, and that doing so efficiently is a worthwhile goal. Police make more than 10 million arrests every year, but a majority of those arrests result in dismissed charges, and less than 6 percent are for violent crimes. Even for people convicted of violent crimes, there is no evidence that putting them in cages is the best way to deter future violence. So, if police and prosecutors were inclined to shrink the size of criminal dockets, they could easily do so without putting anyone in harm’s way: They could arrest fewer people and pursue fewer convictions. Of course, many police and prosecutors are not so inclined. But if appellate courts reversed convictions whenever state actors violated the constitution, police and prosecutors might select and handle their cases more carefully. If they did, the time necessary for retrials would be time well spent.
Claim #2: Harmless error promotes public respect for the criminal process.
Another justification for “harmless error” is that it “promotes public respect for the criminal process by focusing on the underlying fairness of the trial.” But that respect is not earned. To most people, an affirmed conviction signals the absence of error. So, to the extent that affirming illegal convictions garners public respect for our punishment bureaucracy, that respect is not a persuasive justification for “harmless error,” but a symptom of its insidious consequences. It keeps the population ignorant of widespread constitutional violations, and it reduces the threat of resistance.
Claim #3: Eliminating harmless error could weaken constitutional protections.
A third justification comes not from the Supreme Court, but from legal scholars. They worry that eliminating “harmless error” could lead appellate judges, who are often reluctant to reverse criminal convictions, to define constitutional rights more narrowly. This is a serious concern, and in related contexts, broader appellate remedies have had precisely this effect. But the concern is not actually a defense of “harmless error”: It simply recognizes many judges’ strong preference for upholding criminal convictions. The solution is to appoint judges who will not distort constitutional rights to achieve their desired result.
Undoing the Harm
Every day, the doctrine of “harmless error” allows police and prosecutors to act with impunity, and it steals life and liberty from people inside our prisons. What can we do to push back? Here are some ideas:
1. Stop skipping the question of error, and apply Chapman with the rigor that its beyond-a-reasonable-doubt test requires. Judges should not affirm convictions without first deciding whether an error occurred. When deciding whether an error is “harmless,” they should not focus on the strength of the government’s untainted evidence. If the government cannot prove that a constitutional error did not affect the verdict, the court should reverse.
2. Grant executive clemency. State governors could use their clemency power to grant relief to anyone whose constitutional rights were violated on the way to a state prison cell. The president could grant similar relief to people in federal prisons.
3. Enhance state legal protections. The Supreme Court’s decision in Chapman applies to federal constitutional rights. States remain free to create rights that are more protective than federal rights, and they can create broader remedies for the rights they’ve created.
4. Pass federal legislation. In Chapman, the Supreme Court wrote that its “responsibility” to “fashion” the existing rule arose from “the absence of appropriate congressional action.” This language arguably creates an opening for Congress to act; it could neutralize the expansion of “harmless error” that occurred after Nixon’s election, or it could abolish the doctrine altogether.
5. Stop defending unconstitutional convictions. There is a wave of “progressive prosecutors,” but few have said anything about appeals. Prosecutors could decide not to invoke “harmless error” in any of their cases; when they agree that a constitutional error has occurred, they could simply concede error.
None of these ideas require complicated tinkering with an already complex legal doctrine. They ask the judicial branch to apply the law as it’s written, and they ask the executive and legislative branches for political courage.
Gabe Newland is a civil rights lawyer and public defender in New York, where he specializes in appellate and post-conviction litigation.