Concept
Beyond the Trial
Ineffective assistance of counsel
Beyond the Trial
The two-part test began as a rule about sentencing, so it was never confined to guilt-phase trial lawyering. Strickland v. Washington arose after the defendant pleaded guilty to three capital murders and his lawyer presented little mitigating evidence at the sentencing hearing.1 Because the decision under review was a sentence rather than a verdict, the Court described prejudice there as a reasonable probability that, absent counsel's errors, the sentencer would have concluded that the balance of aggravating and mitigating circumstances did not warrant death.2 The test has since spread well past that setting and now governs virtually all claims of ineffective assistance of counsel, with only a few unusual cases treated differently.3
The test travels easily because the reason for its prejudice requirement applies wherever a lawyer's error might have been harmless. A mistake at any stage matters constitutionally only if it affected the decision it touched, so the same two showings are needed at every stage: performance below the prevailing professional standard, and harm from it. The performance prong stays the same. What must be adapted is the benchmark for prejudice, since "a different result" means a different verdict in one setting and something else in another. The next three subsections take up sentencing claims, challenges to guilty pleas (beginning with Hill v. Lockhart), and plea offers that were rejected or missed because of bad advice. Each one applies the same test and asks what the lawyer's error changed.
Sentencing Claims
When the alleged deficiency goes only to the sentence, the two-part test stays intact and the benchmark for prejudice moves. The guilt finding is not under attack, so a different result means a different sentence, and the defendant must show a reasonable probability of a more lenient one.5 In a capital case that means a reasonable probability that, without counsel's errors, the sentencer would have concluded that the balance of aggravating and mitigating circumstances did not warrant death.6 The burden is the same reasonable probability used at trial, which falls short of a preponderance. The defendant need not show that a lighter sentence was more likely than not, only a probability sufficient to undermine confidence in the outcome.7 The performance prong is unchanged as well, so the strong presumption that counsel acted reasonably carries over into the sentencing hearing.8
The facts of the originating case give the clearest picture of a sentencing claim that fails. In Strickland v. Washington the defendant pleaded guilty to three capital murders and was sentenced to death after a hearing at which counsel presented little mitigating evidence.9 The Court found counsel's decision not to present more mitigation to be a reasonable strategic choice. It then added that, even if the performance had been deficient, the prejudice was insufficient, because the aggravating circumstances were overwhelming and the omitted evidence would not have altered the outcome.10 The reasoning is a comparison of two weights. A reviewing court asks what the sentencer actually had before it, adds what competent counsel would have supplied, and asks whether the balance would probably have shifted. When the aggravation is heavy and the omitted mitigation is light, adding it to the scale leaves the balance where it was, so no reasonable probability of a different sentence exists. That is the totality-of-the-evidence idea applied to a sentencing record, and the Court's fallback holding on prejudice shows the permission to resolve a claim without settling performance.
The capital cases in which the Court has found Sixth Amendment violations show what a sentencing claim looks like when the lawyer's own work is the problem. In Wiggins v. Smith, trial counsel failed to investigate the defendant's life history fully even though counsel had reason to believe the investigation would turn up mitigating circumstances. In Rompilla v. Beard, defense counsel did not look at the file on the defendant's prior conviction, which the prosecution had said would be central to proving an aggravating circumstance, and they skipped it even though family members and the defendant himself had suggested that no mitigating evidence existed. That same file held evidence of severe mental illness, child abuse, and alcoholism that counsel never read, and without reading it counsel were in no position to make a reasonable strategic choice.11 In Buck v. Davis the error was affirmative: the lawyer knowingly introduced a psychologist's opinion that the defendant was statistically more likely to act violently because he is Black, in a case where the death sentence turned on the jury's assessment of future dangerousness. These cases differ from Strickland in the quality of the lawyer's choice and in the bearing of the evidence on the sentencer's question. Strickland's counsel made a strategic judgment about material that would not have changed the balance, while the lawyers in these cases failed to learn what the record held or placed damaging material before a jury deciding dangerousness.
Sentencing claims can rest on a modest difference, so the size of the difference matters, and the Court has declined to set a floor. It refused to require substantial prejudice, in part because it is hard to tell substantial from insubstantial prejudice, and so any increase in the time a defendant will serve that counsel's deficient performance causes can be the prejudice needed.12 In Glover v. United States counsel failed to object to a legal error that affected the calculation of the defendant's prison sentence, and an error that added imprisonment, regardless of its length, plainly harmed the defendant.13 The reasoning is practical. A threshold would force courts to decide how many extra months count as substantial, and any line would be arbitrary, while every added day in custody is a concrete deprivation. The rule does not lower the reasonable-probability burden. The defendant must still show that the error caused the extra time, and in a miscalculation case the link is direct because the sentence follows from the number. What the defendant avoids is any argument about whether the difference was large enough to matter. Strickland and Glover therefore answer different questions: the first asks whether the omitted material would probably have changed the sentence at all, and the second declines to ask how large a proven change must be.
The lawyer's performance was equally poor in both cases, and what moved the answer was the connection between the error and the sentence. A miscalculation converts directly into time served, while omitted mitigation matters only if it would probably have moved a sentencer who had already heard similar material.
A court may begin at the third step and stop there when the aggravation is overwhelming and the omitted material is thin, as the Court did in Strickland. Two mistakes recur on this subject. The first is importing a size threshold and demanding substantial prejudice before added time counts, which Glover forecloses.14 The second is reading the capital standard to require proof that a life sentence was more likely than death. The test asks only for a reasonable probability, a probability that undermines confidence in the outcome, and the defendant need not show that a different sentence was more likely than not.15
Guilty Plea Challenges
Sentencing claims showed that the Strickland test keeps both prongs and changes only what counts as a different result. The plea process calls for the same adjustment, and the stakes are higher because the great majority of criminal convictions come from guilty pleas. In Hill v. Lockhart the Court held that the two-part test applies to challenges to guilty pleas based on ineffective assistance of counsel.17 The Court reasoned that the justifications for requiring prejudice in Strickland carry over to pleas: attorney errors come in infinite variety and are as likely to be harmless in one case as harmful in another, so even an unreasonable error must be shown to have had an adverse effect on the defense.18 Finality adds a reason specific to pleas. Because so many convictions rest on them, every new ground for setting aside a plea undermines confidence in the system's procedures and adds to the volume of judicial work.19
The first prong needs no new rule. In the plea setting it restates the standard of attorney competence from McMann v. Richardson and Tollett v. Henderson, under which the validity of a plea entered on counsel's advice depends on whether that advice was within the range of competence demanded of attorneys in criminal cases.20 That is a standard of reasonable competence, so a defendant is not entitled to perfect advice. A defense lawyer often cannot predict whether a court will exclude certain evidence or which witnesses a jury will believe, and the strong presumption that counsel acted within the wide range of reasonable professional assistance still has to be overcome.21 A plea claim therefore fails at the first prong when the lawyer's judgment about an uncertain trial was a reasonable one, even if it later proved wrong, because a defendant who takes the benefit of a plea accepts the risk that a competent lawyer's good-faith evaluation will turn out to be mistaken.22
The second prong is where the plea setting changes the benchmark. The defendant must show a reasonable probability that, but for counsel's errors, the defendant would not have pleaded guilty and would have insisted on going to trial.23 The adjustment follows from what a plea does. A guilty plea ends the case without a trial, so there is no verdict whose result counsel's error could have altered, and the proceeding the error may have changed is the decision to plead. Asking whether the outcome would have differed therefore means asking whether the defendant would have chosen the other path. As elsewhere, a reasonable probability is one sufficient to undermine confidence in the outcome, and it can be less than a showing that a different outcome was more likely than not.24 The defendant also need not prove that a trial would have ended in an acquittal, because the lost opportunity to go to trial is itself the harm.25
A trial defense still matters to the analysis, because Hill recognized that in many plea cases the prejudice inquiry closely resembles the inquiry in a challenge to a trial conviction. When the alleged error is a failure to investigate or discover exculpatory evidence, the question is whether finding it would have led counsel to change the plea recommendation, and that depends largely on a prediction whether the evidence would have changed the trial outcome. When the error is a failure to advise of an affirmative defense, the answer depends largely on whether the defense likely would have succeeded at trial.26 Hill also directed that these predictions be made objectively, without reference to how a particular judge or jury might have behaved.27 The strength of the hypothetical trial thus serves as evidence about what the defendant would have done, since a person holding a strong defense has good reason to reject a plea. It is evidence of that choice only, and the defendant does not have to prove it as a separate element.
Hill's own facts show the prejudice prong working as a screen. The petitioner pleaded guilty after counsel gave erroneous advice about his eligibility for parole under the sentence in the plea bargain.28 His habeas petition did not allege that, with correct information, he would have pleaded not guilty and insisted on trial, and it alleged no special circumstances suggesting that he had placed particular emphasis on parole eligibility in deciding to plead.29 The omission mattered because his mistaken belief would also have distorted his estimate of the time he would serve after a trial conviction, so nothing suggested that the error tipped the choice between the two paths.30 The Court therefore held the allegations insufficient on prejudice and held that no hearing was required.31 Because the claim ended there, Hill left open whether erroneous advice about parole eligibility can ever be constitutionally deficient.32 That sequence mirrors the order-of-analysis rule from earlier in this chapter, under which a court may resolve a claim on prejudice without grading the lawyer.
The contrast comes from Lee v. United States. A noncitizen who had lived in this country since childhood pleaded guilty to a drug charge after counsel wrongly assured him that the plea would not lead to deportation. He had no real defense and faced a longer sentence if convicted at trial, yet the Court found a reasonable probability that he would have rejected the plea and taken his chances at trial, because remaining in the country was of paramount importance to him.33 A claim with no winning trial defense thus succeeded where Hill's claim failed, and only one variable moved the answer: whether the defendant could credibly say that he would have insisted on trial. Hill's petitioner made no such allegation and showed no special weight on the parole information, while Lee supplied concrete evidence that deportation was the consequence driving his decision. Because the lost opportunity for trial is the harm, a defendant in Lee's position, who valued staying above everything, had every reason to gamble on trial even against long odds.34
A claim like Lee's presupposes deficient advice, and the performance side comes from Padilla v. Kentucky. The Court there held that advice about deportation is not categorically outside the Sixth Amendment, because deportation is so closely tied to the criminal process that it is uniquely difficult to classify as a direct or a collateral consequence of a conviction.35 Where the deportation consequence is truly clear, as when the removal statute makes it presumptively mandatory, counsel's duty to give correct advice is equally clear, and where the consequence is unclear, counsel need only advise that the charges may carry adverse immigration consequences.36 Padilla decided only that performance question and left the prejudice question for the state courts on remand.37 Hill and Lee together supply the test for that second question, which is why a plea claim built on deportation advice runs through Padilla first and Hill and Lee second.
The step that carries the most weight on an exam is the second, since facts about what the defendant valued usually decide the claim. Students lose points on this subject when they demand proof that the trial would have been won, which the plea-stage prejudice test does not require.38 The opposite error is to assume that wrong advice automatically harms the defendant, when the defendant must still show a reasonable probability of choosing trial.39
Rejected Plea Offers
The accepted-plea cases ask whether a defendant given correct advice would have rejected the deal and gone to trial. Bad advice can also push a defendant the other way. A lawyer who never passes along a prosecutor's offer, or who talks the client out of a favorable one on a mistaken view of the law, can leave the defendant facing a trial and a sentence far heavier than the one that was available. Missouri v. Frye and Lafler v. Cooper, decided the same day, hold that the right to effective assistance reaches plea negotiations that lapse or fail, even though the Constitution gives no right to a plea bargain in the first place.41 The reason for reaching negotiations is practical. Most convictions are obtained through plea bargaining, so negotiation is a stage at which a lawyer's advice can decide the outcome, and a defendant without competent advice cannot capture the benefits a plea agreement offers.42 The Strickland framework still governs, with a performance prong and a prejudice prong, and each prong is adapted to a defendant who never entered a plea at all.43
The performance rule comes from Frye. Galin Frye was charged with driving with a revoked license, a felony because of three earlier convictions for the same offense, and his lawyer never communicated the prosecutor's formal offers to him before they lapsed.44 The Court held that, as a general rule, defense counsel must communicate formal prosecution offers on terms that may be favorable to the accused, and that letting an offer with a fixed expiration date lapse without telling the client is not the effective assistance the Constitution requires.45 The rule is deliberately narrow. Bargaining depends so heavily on individual style that the Court declined to write detailed standards for how counsel negotiates, and it left any exceptions to the communication duty unexplored. A duty tied to formal offers is also easy to administer, because the offer's terms and processing can be documented, which gives a court something objective to measure counsel against.46 The same performance prong reaches the reverse error, since palpably deficient advice to reject an offer can also amount to ineffective assistance.47
Lafler presents the version in which the lawyer did communicate the offer. The prosecution twice offered to dismiss two charges and recommend 51 to 85 months, and the defendant, who had fired a gun at the victim, rejected the offer after counsel convinced him that intent to murder could not be proved because she was hit below the waist. After a fair trial he was convicted on every count and received a mandatory minimum of 185 to 360 months, and the case reached the Court with the concession that counsel's advice was deficient.48 The government parties argued that no prejudice can arise from plea bargaining when the defendant is later convicted at a fair trial. The Court found that position unpersuasive, because the prejudice alleged in this setting is having to stand trial at all when the better deal was available.49 The narrower view also ignores how the system works. Pleas dominate dispositions, and losing the benefit of an offer has real sentencing consequences, so the claim cannot be limited to protecting the fairness of a trial, and it survives even though the defendant stood trial.50
Prejudice in this setting is therefore measured against the lost bargain rather than against the verdict. The defendant must show a reasonable probability that, but for counsel's deficient advice, the defendant would have accepted the offer, the prosecutor would not have withdrawn it, and the court would have accepted it, producing a less serious conviction or a less severe sentence than the one imposed.51 Each part screens out a different kind of speculation. The first asks about the defendant's own choice, and because that choice must be reconstructed from information the defendant never actually had, a large gap between the sentence the offer carried and the exposure at trial counts as objective evidence of what the defendant would have done. In Baker v. United States counsel told the defendant his trial exposure was 21 years when the mandatory minimum on all counts was 57 years, the offer would have produced 15 to 17 years, and the Third Circuit found a reasonable likelihood he would have taken the deal.52 The second and third parts exist because an offer is only a proposal until the prosecutor honors it and the court approves it. A defendant who cannot show the deal would have come into being has lost nothing that counsel's error took away. Whether a prosecutor may cancel an offer or a court may refuse a plea is a matter of state law, and in Frye the Court remanded for the Missouri courts to decide those questions because they bear on prejudice: if the prosecutor could have canceled the agreement and Frye could not show a reasonable probability that it would have been honored, there was no prejudice.53
Frye's own facts supply the favorable pole on the first part. He ended up pleading guilty to a more serious charge with no promised sentencing recommendation, which supports a reasonable probability that he would have accepted the earlier, better offer had he known of it.54 The unfavorable pole comes from the other two parts, where a deal that would never have taken effect gives the defendant nothing to recover.
The contrast with the accepted-plea cases in the preceding section shows why the benchmark moves. In Hill and Lee the defendant gave up a trial, so the lost opportunity for trial is itself the harm, and the defendant need not show that a trial would have gone better.55 A defendant who rejected an offer has already had a trial, so that harm is unavailable, and the injury is the lost chance at a better outcome. The accepted-plea question is therefore whether the defendant would have gone to trial, and the rejected-offer question is whether the favorable deal would have taken effect and what it would have produced. The remedy follows from that injury. In Lafler the State had to re-offer the original deal, and if the defendant accepts it, the trial court decides how to amend the result, because outright specific performance of the original plea is not the remedy.56 The last section of this chapter returns to remedies in full.
The step students most often skip is the third, because the first two feel like the whole case once a lawyer has plainly erred. A defendant who proves bad advice and a willingness to plead still loses if the deal would never have come into existence. The more common mistake is treating a fair trial as a cure. A trial free of error does not answer the claim, because the injury alleged is the loss of the bargain.57