Concept
Substantial Right and Harmless Error
Objections and offers of proof
Substantial Right and Harmless Error
A party who has preserved a claim of error with a timely, specific objection or a sufficient offer of proof has cleared the procedural hurdles, but Rule 103(a) imposes one more condition before the claim can succeed. Its opening clause allows a party to claim error in a ruling to admit or exclude evidence only if the error affects a substantial right of the party.1 The condition exists because no trial is perfect, and a rule that sent every evidentiary slip back for a new trial would burden courts and litigants with retrials that could not be expected to come out differently. A reviewing court therefore reserves reversal for mistakes that probably made a difference to the outcome. Preserved error that did not matter is called harmless error, and it leaves the judgment standing.
The analysis in this section runs in a fixed order. It begins with the standard an appellate court uses to decide whether the trial judge erred at all, then turns to the meaning of a substantial right and the doctrines that convert apparent error into harmless error. It ends with the heightened rule for constitutional error, the small category of structural errors that require reversal automatically, and the special case of the bench trial.
Reviewing the Underlying Ruling
The first question on appeal is whether the trial court made an incorrect ruling at all, and the usual label for that inquiry is abuse of discretion. The label carries two meanings, and confusing them produces much of the apparent disagreement among courts about standards of review. In the broad sense, the phrase simply says that the trial judge had the power to decide and that a decision the law does not permit is an abuse of that power. Such a decision may rest on a mistake of law, on a clearly erroneous finding of fact, or on a manifestly erroneous application of a rule of evidence to the facts.2
In the narrow sense, discretion means that the trial judge may reach a result the appellate court would not have reached itself, and the result will be affirmed unless it is a clear error of judgment. The reason for that latitude is practical. The trial judge sat through the testimony and saw the setting in which the question arose, while the appellate court works from a cold record. The narrow sense governs the Rule 403 balancing test, the decision whether an expert is qualified, the Rule 702 gatekeeping decision, and the decision whether a statement was made under the stress of a startling event.3 A reader who sees "abuse of discretion" in an exam answer should check which meaning is intended, because a legal mistake is reversible in a way that a close call within the judge's range of choices is not.
The Substantial Right Standard
Once the appellant has shown that the ruling was wrong, the appellant must show that the mistake mattered. A technical violation of a rule of evidence, even one properly brought to the court's attention, is not a sufficient basis for reversal.5 Reversible error thus has two parts: first there must be error, and second the error must affect a substantial right of a party.6 The first part concerns the ruling and the second concerns its consequences, so an appellant can win the first and still lose the appeal.
A substantial right is affected when the error had a material effect on the jury's deliberations or, in the older phrase, "substantially swayed" them.7 The formulation comes from the Supreme Court's decision in Kotteakos v. United States. It asks about the likely effect of the error on this verdict, in light of the whole trial, and does not ask whether the ruling was a serious departure from the rules. Error that does not affect a substantial right is called harmless.8 The 103(a) lead-in did not change the earlier law of harmless and reversible error, so the older cases remain good guides to what counts.
Two cases that differ in a single fact show how the standard operates. A trial judge wrongly excludes the only rebuttal to the opponent's central expert, and the objection was properly preserved. The jury heard the expert's account unanswered, so the excluded rebuttal probably affected the outcome, and reversal is likely. Change one fact: the judge wrongly excludes three more business records of a kind where twenty-five similar records were already admitted. The ruling was still wrong, but the jury had all it needed to understand the point, so the error did not sway the verdict and the judgment stands. In both cases the mistake and the objection were the same. What differs is the importance of the excluded evidence to the case.
Doctrines That Make Error Harmless
Courts do not ask the substantial-right question in a vacuum. Three recurring circumstances lead them to treat an apparently reversible error as harmless, and each follows from the same idea, that the verdict would probably have been the same without the error.
The third circumstance needs the most care. Over time, the approach that asked whether the error contributed to the verdict has been eroded, and in part supplanted in practice, by an approach that asks whether the whole record, with the error considered in context, shows overwhelming evidence of guilt. Under that approach, a defendant who is, in the memorable phrase used for it, "guiltier than hell" will lose even a real error.10 The practical lesson is that the strength of the untainted evidence often decides a harmlessness argument, and the importance of the item to the proponent's case often does not. A party seeking reversal should therefore argue that the case was close, while the party defending the judgment should show how much else supported the verdict.
Constitutional Error and Chapman
Evidentiary error that violates the Constitution in a criminal case receives more demanding review. Such an error can still be harmless, so not every constitutional violation compels a new trial. In Chapman v. California the Supreme Court held that the beneficiary of the error, which in a criminal appeal is the prosecution, must show that the error was harmless beyond a reasonable doubt. The reason for the heavier burden is the nature of the interest at stake. A defendant's liberty and constitutional guarantees would count for little if the government could keep a conviction by showing only that the error probably did not matter.
The contrast with ordinary evidentiary error is the burden and the degree of confidence required. For a nonconstitutional error, the question is whether the error substantially swayed the jury, and the appellant must show that it did. For a constitutional error in a criminal case, the beneficiary of the error must persuade the court beyond a reasonable doubt that the error did not contribute to the verdict. A confession admitted in violation of the Constitution is therefore reviewed far less leniently than an unconstitutional-free hearsay slip of equal size. The same wrong ruling can lead to different outcomes depending on whether a constitutional right is involved, and that difference is why an exam answer should identify the constitutional dimension first.
Structural Error
A small group of constitutional errors are called structural, and they require reversal without any inquiry into harmlessness. Typical examples are the complete denial of counsel, a biased judge, and the denial of a public trial. The reason is that these errors infect the framework within which the trial proceeds, so their effect on the outcome is hard to measure and would have to be reconstructed by speculation. When prejudice cannot be reliably proved, a rule that required the appellant to prove it would leave the violation uncorrected. The law therefore presumes prejudice and reverses.
Structural error is the exception within the exception. Most constitutional errors, including most that involve evidence, are reviewed for harmlessness under the Chapman standard, and an exam answer that labels an ordinary evidentiary ruling structural has misunderstood the category. The overview of Rule 103(a) accordingly distinguishes reversible, harmless, and structural error as three separate outcomes.12
Nonjury Trials
The harmless-error analysis changes in a bench trial. When a judge sits without a jury and admits evidence over objection, the error is generally harmless if enough competent evidence supports the findings. The reason is that a judge is presumed able to disregard inadmissible evidence and to decide the case on the admissible proof, so the concern that a lay jury will be inflamed or misled by an item it should not have heard is much weaker. An appellant in a bench trial therefore has a harder task showing that an erroneous admission mattered. The exam rule is a presumption and not an absolute: if the findings rest on the inadmissible evidence and nothing else supports them, the error can still require reversal.
The Substantial Right Standard
An appellant who has preserved an objection or made a proper offer of proof still has two things to show. The first is that the trial court made an incorrect ruling, and the second is that the ruling affected a substantial right of the party.13 The first question is reviewed under the label abuse of discretion, which has a broad sense and a narrow one. In the broad sense, any ruling the law does not allow a trial judge to make is an abuse, whether it rests on a mistake of law, a clearly erroneous finding of fact, or a manifestly erroneous application of a rule to the facts. In the narrow sense, the judge may reach a result the appellate court would not have reached, as in Rule 403 balancing, and the result stands unless it is a clear error of judgment.14 Only after the appellant wins on that question does the substantial-right question arise, and it is the one that decides most appeals.
A substantial right is affected when the error had a material effect on the jury's deliberations or, in the older phrase, "substantially swayed" them.15 The phrase comes from Kotteakos v. United States, where the indictment charged one conspiracy but the proof showed eight separate ones. The Court did not ask whether that mismatch broke a rule. It asked what the mismatch meant to the jury, judged in relation to everything else that happened at trial.16 Because the answer depends on the effect of the error and not on its label, a ruling can be plainly wrong and still leave the judgment standing.
The opinion gives three instructions for conducting the inquiry. The first is a warning against technicality: the governing statute amounted to a plain admonition not to be technical where technicality does not really hurt the party.17 The second is the fair-assurance test. If the court, after pondering all that happened without stripping the erroneous action from the whole, cannot say with fair assurance that the judgment was not substantially swayed by the error, it cannot conclude that substantial rights were unaffected, and a court left in grave doubt must reverse.18 The third is a limit on the shortcut courts find most tempting. The inquiry is not merely whether enough other evidence supported the result apart from the phase the error touched. It asks whether the error itself had substantial influence.19 The reason is that a jury's verdict is a single judgment built from the whole record, so an appellate court that subtracts the error and then asks whether a verdict could still stand is deciding the case itself instead of reviewing what the jury did.
The three instructions can be run as a short sequence on any preserved evidentiary error.
The allocation of the burden depends on the kind of error. The statute behind Kotteakos placed on the party seeking a new trial the burden of showing that a technical error affected substantial rights. It left the burden of sustaining the verdict on the other side when the error's natural effect is to prejudice a litigant's substantial rights.20 Courts also vary in how probable the effect must be, from "may have" to "probably did," and the variation can follow whether the case is civil, criminal, or constitutional.21 The exam answer is therefore stated in terms of the verdict: did this error likely matter to this jury?
The Kotteakos facts show the standard working against the government. The Court distinguished an earlier case in which only two conspiracies and four persons were proved. With eight conspiracies and thirty-two defendants at the outset, the risk that evidence against one defendant would spill over to another, even with instructions to keep the transactions separate, made it impossible to treat the error as harmless.22 The same logic applies to an ordinary evidence ruling. When a wrongly admitted letter is the centerpiece of a close case and the prosecutor builds the closing argument on it, the court cannot say with fair assurance that the jury was not swayed, and reversal follows.
Change the facts and the answer changes. If the proponent authenticated a document one step out of sequence, and nothing about the document's content or timing could have influenced the jury's view of the case, the violation is technical, and the appellate court disregards it.23 In both situations the trial judge erred and the objection was preserved. The difference is whether the error mattered to the verdict. On an exam, an error that is clear and preserved still leaves the appellant with the effect question to answer, and the stronger answer addresses how the jury would have used the item.
Doctrines That Make Error Harmless
An appellate court that finds a preserved error does not stop there, because Rule 103(a) permits a claim of error only if the ruling affected a substantial right of the party.25 Courts have developed three recurring situations in which a wrong ruling is nonetheless treated as harmless: the item was cumulative, a curative or limiting instruction removed the prejudice, or the untainted evidence was overwhelming. All three answer the same question, which is whether the jury's verdict probably would have been the same without the mistake. They differ only in the reason the answer comes out yes, and keeping the reasons apart helps in deciding which one an exam fact pattern is inviting.
Cumulative evidence is the simplest. When the jury already had the same information from properly admitted proof, an erroneous admission or exclusion of one more item adds or removes nothing that could have moved the verdict. The civil rule on harmless error says as much: unless justice requires otherwise, no error in admitting or excluding evidence is ground for disturbing a judgment.26 Consider an illustration built on a plaintiff named Trivedi, who offered proof that coworkers had been promoted while he was passed over. The judge admitted five witnesses and twenty employment records on the point, then wrongly excluded three more promotion records as hearsay over his timely objection. The three records were merely more of what the jury had already seen, so the exclusion was probably harmless.27
Change what was excluded and the answer reverses. In the same illustration, the judge wrongly excludes, over timely objection, the testimony of Trivedi's own psychiatrist that his mental state was normal, though stressed. The defense had called two doctors who diagnosed a delusional disorder, and the psychiatrist's account was the one powerful rebuttal to them. The error here is almost certainly prejudicial, because the jury heard the defense experts unanswered.28 The ruling, the objection, and the party are the same in both versions. The only variable is whether the excluded item filled a gap in the jury's picture or merely repeated what the jury already had. A similar logic governs impeachment. When a court wrongly excludes a witness's prior conviction but a good deal of other evidence already went to the witness's lack of credibility, the exclusion probably does not affect a substantial right.29
The cumulative doctrine is easy to misapply, because it invites the shortcut of asking whether enough other evidence supported the verdict. Some courts do articulate harmlessness that way, treating an error as harmless if other evidence would let a reasonable juror reach the same result, even when the improperly admitted item was probably the decisive one in the jury's mind.30 The standards also vary by jurisdiction and by whether the case is civil, criminal, or constitutional. That variation is the reason the safer exam analysis keeps the focus on the verdict: the question remains whether this error probably affected what the jury did. Treating sufficiency of the remaining proof as the end of the inquiry converts harmless-error review into a second sufficiency review, in which the appellate court decides what a verdict could have been instead of asking what the error did to the verdict actually returned. Cumulative means the item duplicated proof the jury already had. It does not mean the case could have been won without the item.
A curative or limiting instruction works on a different theory. Rather than arguing that the jury had the information anyway, it argues that the jury was told how to handle it. A judge who learns that an answer is inadmissible may strike it and tell the jurors to disregard it, and a judge who admits evidence for one purpose or against one party may instruct the jury to use it only that way.31 The doctrine rests on the premise that jurors are ordinarily willing and able to follow the judge's directions. That is why the proper form of objection to an answer that has already come in is a motion to strike, coupled with a request for the instruction. A mistrial, the alternative remedy, is seldom granted and is reserved for cases where an instruction could not cure the prejudice.
The premise has limits, and the limits mark where the instruction fails to save the judgment. Jurors told to ignore powerful, probative evidence may be influenced by it anyway, and an instruction can even draw attention to what it tells them to forget.32 The limits are firmest in criminal cases involving constitutional error against the accused. An instruction to disregard an involuntary confession does not cure its admission, and a limiting instruction does not adequately protect a defendant from a codefendant's confession that implicates him in a joint trial.33 Compare two situations. In a negligence trial, a witness blurts out a vague, unsupported remark that the defendant was probably insured, the judge sustains a motion to strike and promptly instructs the jury to disregard it, and the error is very likely harmless. In a criminal trial, the jury hears the defendant's own confession, obtained in violation of the Constitution, and is then told to ignore it. The instruction in the second case cannot be presumed effective, because the evidence is devastating and jurors are unlikely to unhear it.
The third doctrine looks away from the item and toward everything else in the record. In recent practice the older question, whether the error contributed to the verdict, has been eroded and in part supplanted by an approach that asks whether the whole trial, with the error considered in context, shows overwhelming evidence of guilt.34 Under that approach an accused who is, in the blunt phrase used for it, "guiltier than hell" loses even a real error.35 The approach differs from the cumulative doctrine in what it measures. Cumulative analysis asks what the challenged item added to the jury's picture, while overwhelming-evidence analysis asks how strong the untainted proof was without it. A party defending the judgment therefore does best by showing how much else pointed to the result, and a party seeking reversal does best by showing that the case was close. The strength of the untainted evidence frequently decides the argument, which is why the fact pattern on an exam usually gives you a detail about how much other proof there was.
Constitutional Error and Chapman
The harmless-error standards covered so far assume that the trial judge broke an evidence rule and nothing more. When the erroneous ruling also violates the Constitution in a criminal case, the review changes in two ways: the appellate court asks a more demanding question, and the burden of answering it falls on the party who benefited from the error. Evidence law borrows this framework from constitutional criminal procedure, so a court reviewing a constitutional evidentiary ruling will reverse unless it can determine beyond a reasonable doubt that the error was harmless.37
The source of that standard is Chapman v. California, in which the prosecutor's comments on the defendants' decision not to testify violated their Fifth and Fourteenth Amendment right to be silent. The Court first held that whether such an error can be harmless is a federal question, because the right is created by the Federal Constitution and the states cannot be left to fashion their own remedies for violating it.38 The defendants then asked for the opposite extreme, automatic reversal for every constitutional error. The Court declined, reasoning that harmless-error rules serve a useful purpose by blocking reversals for small defects that had little chance of changing the result, and that some constitutional errors are so unimportant in the setting of a particular case that they can be disregarded.39
Having rejected automatic reversal, the Court set the price of affirmance high. Before a federal constitutional error can be held harmless, the court must be able to declare a belief that it was harmless beyond a reasonable doubt.40 The burden of proving harmlessness sits on the beneficiary of the error, which is the prosecution, and the reason follows from the structure of the problem. A reviewing court working from a cold record often cannot tell whether forbidden evidence or argument tipped a close case. If the defendant had to prove that it did, the doubt would be resolved against the person whose constitutional right was violated. Assigning the burden to the party that benefited from the violation puts the risk of that uncertainty on the side that profited from it. The Court also saw little difference between this formulation and asking whether there is a reasonable possibility that the error contributed to the conviction.41 Applied to the facts, the standard was easily failed. The prosecutorial comments on the defendants' silence were so extensive that the Court had no doubt the error was not harmless, and the convictions fell.42
Change one fact and the result can reverse. Suppose a statement taken in violation of the Constitution is admitted at trial, but every detail in it also came in through properly admitted video evidence and the defendant's own testimony. The government can then carry its burden, because the statement added nothing the jury did not already have, and the conviction stands. In Chapman the forbidden comments were the centerpiece of the prosecution's argument, so the state could not make that showing. The constitutional label alone decides neither case. What decides them is whether the beneficiary can prove, to the beyond-a-reasonable-doubt standard, that the error did not contribute to the verdict.
The constitutional standard also has to be kept apart from the ordinary one. For an error of non-constitutional law, the prosecution must show that the error did not have a substantial and injurious effect on the jury's verdict, which is the familiar substantial-right inquiry.43 For a constitutional error, the prosecution must prove harmlessness beyond a reasonable doubt.44 Both place the burden on the government in a criminal appeal, so the difference lies in the degree of certainty required. The higher standard is the Constitution's way of enforcing its own guarantees more strictly than the rules of evidence enforce themselves.
Some constitutional claims never reach the Chapman question. Ineffective assistance of counsel and the withholding of favorable, material evidence are errors only if the defendant shows prejudice, so prejudice is part of the definition of the violation, and a violation of that kind cannot also be harmless.45 The dissent in the ineffective-assistance case objected to exactly this feature, which places on the defendant the burden of demonstrating prejudice.46 On an exam, do not run a separate harmless-error analysis on such a claim, because the prejudice showing has already been made or the claim has failed.
The constitutional category also reaches the defense side of the courtroom. Most constitutional evidentiary error involves admitting prosecution evidence that should have been excluded, under the search-and-seizure, self-incrimination, or confrontation provisions. Less often, a court commits constitutional error by excluding evidence the defense offered, as in Chambers v. Mississippi.47 Finally, Chapman itself acknowledged that a few constitutional rights are so basic to a fair trial that their infraction can never be treated as harmless, a point that the next section develops.48
Structural Error
The harmless-error rules just covered share an assumption: a reviewing court can look at the record, weigh what the error added against everything else the jury had, and decide whether the verdict probably turned on it. A small group of constitutional errors does not fit that assumption, and for them the appellate court skips the harmlessness inquiry altogether and reverses. These are called structural errors. Chapman itself acknowledged that some constitutional rights are so basic to a fair trial that their infraction can never be treated as harmless.50
The reasons for automatic reversal come from the way the inquiry would otherwise work. Courts fashion per se rules when the risk that a party's substantial rights were affected is unusually high but prejudice would be hard to prove, when a prophylactic rule serves an important social policy, when officials need a definite standard to guide future conduct, or when case-by-case analysis would cost too much judicial effort.51 The first reason does most of the work. A defendant tried with no lawyer at all, or before a biased judge, has been denied something that shapes every decision made at trial, from which witnesses to call to which objections to raise. No record shows what a properly defended trial before a neutral judge would have produced, so asking whether the error mattered would require speculation. Put differently, a structural error is one that undermines confidence in the result and also destroys the trial as a reliable method of reaching a result at all.52
The recognized examples are few. Complete lack of counsel and a biased trial judge are the standard cases, and the same reasoning covers erroneous disqualification of the defendant's chosen counsel.53 A defendant tried without counsel or before a biased judge wins reversal without any showing that the outcome would have differed.
The contrasting case is the coerced confession, which is the example most likely to mislead. Older statements of the doctrine list it among the errors whose infraction can never be harmless.54 In Arizona v. Fulminante the Supreme Court held that admitting an involuntary confession is trial error, subject to harmless-error review, even though such a confession is likely to be highly prejudicial.55 The Court reasoned that the confession came in while the evidence was being presented to the jury. Its impact could therefore be evaluated against the other evidence, so the court could ask whether the error was harmless beyond a reasonable doubt, and the prejudice was not so inevitable as to make that inquiry a waste of time.56 That line of reasoning supplies the dividing test. An error that arises at a discrete point in the evidence can be measured against the rest of the record, while a defect in the framework of the trial cannot.
Most constitutional violations fall on the trial-error side. Courts review as trial errors the admission of evidence obtained in violation of the Fourth Amendment, improper eyewitness identifications, comments on the defendant's failure to testify, restrictions on the defendant's ability to testify, Confrontation Clause violations including Bruton violations, and jury instructions that misstate or omit an element of the offense.57 Even the erroneous refusal to exclude a juror, in a case involving a peremptory challenge, was held not to be structural.58 When harmless-error review does apply, the defendant must establish the error, and the prosecutor, as the beneficiary of the error, must then show it was harmless.59 Do not treat a constitutional label as a structural label, because the exam usually places a serious constitutional violation, such as an involuntary confession, in front of you to see whether you reach for automatic reversal.
Nonjury Trials
Everything in this chapter about harmless error has assumed a jury, which is the audience the exclusionary rules were built to protect. The evidence rules still apply in a bench trial, where a judge decides the facts without a jury, but the appellate treatment of error changes in an important way. The change follows from who is deciding, and it produces a different working practice at trial as well.
The exclusionary rules grew largely out of distrust of lay jurors, who were thought unable to give a prejudicial or misleading item only the weight it deserved.61 A professional judge has experience in valuing evidence and is better able to set improper material aside, so the danger that a given item distorted the outcome is lower.62 Appellate courts have turned that reasoning into a presumption. On appeal from a bench trial, the receipt of inadmissible evidence over objection is ordinarily not grounds for reversal if other, admissible evidence was sufficient to support the findings, because the judge is presumed to have disregarded the improper evidence and relied on the proper evidence.63 The effect is to treat the error as one that did not affect a substantial right of the objecting party under Rule 103(a).64
The presumption can be rebutted, and the way it is rebutted shows what it rests on. It is an assumption about what the judge actually did, so a showing that the judge did the opposite defeats it. Possible methods include statements from the bench during trial and statements in the specific findings of fact, whether prepared separately or included in an opinion or memorandum of decision.65 Take a judge who admits hearsay over objection, but whose written findings rest on other competent evidence. The error is harmless, because the record shows that the sound evidence carried the decision. Change one fact: if the findings expressly rely on the improperly admitted hearsay, the presumption is rebutted, and the loser can argue that a substantial right was affected. The presumption is also not universal. Some appellate decisions decline to apply it when the evidence was objected to and the objection overruled, reasoning that overruling the objection suggests the judge wanted to consider the evidence.66
The presumption runs in one direction only. When a judge errs by admitting evidence, the benefit of the doubt protects the ruling. When the judge errs by excluding evidence that ought to have been received, the ruling is reviewed as in a jury trial and is subject to reversal if it substantially harmed the losing party.67 The reason is structural. A judge who excludes evidence never considers it, so there is nothing to presume the judge looked past, and the appellate court must ask directly what the missing evidence would have done. Preserving that error still requires an offer of proof, which is why an exam about a bench trial exclusion still asks whether the proponent made one.
Because of the presumption, careful trial judges adopt a protective practice. Faced with a debatable objection, the judge often admits the evidence with the objection "considered as going to the weight of the evidence," or reserves ruling on admissibility until the end of the case and then, if pressed, admits the evidence without relying on it.68 Judges following this practice may say they will admit evidence for whatever it is worth, and then, to minimize the risk of reversal, explicitly identify on the record the admissible evidence on which the findings rest.69 The practice saves time otherwise spent arguing objections, and it supplies the appellate court with a record showing both what was rejected and what was relied on. Together with the presumption, it gives the rules of evidence a somewhat muffled impact in judge trials, although privileges are generally fully enforced.70 Do not assume that admission over objection in a bench trial is reversible just because the evidence was inadmissible, since the question is whether competent evidence supports the findings and whether the judge relied on the improper item.