The Two-Part Test

The Two-Part Test

The presumption of effective assistance introduced above gives a reviewing court its starting point, and Strickland then supplies two questions the court must answer. The first question concerns the lawyer. The defendant must show that counsel's performance was deficient, meaning that counsel made "errors so serious that counsel was not functioning as the 'counsel' guaranteed the defendant by the Sixth Amendment."¹ The second question concerns the effect of those errors. The defendant must show that the deficient performance prejudiced the defense, which means the errors were serious enough to deprive the defendant of a fair trial, one whose result is reliable.² Each question is a separate requirement, and a defendant who cannot satisfy both has not established a Sixth Amendment violation.³

The test has two parts because each part guards against a different mistake. The performance requirement ties the claim to what the Sixth Amendment actually promises, which is a lawyer who functions as an advocate. Without it, a defendant could win by pointing to a bad result even when counsel did everything a competent lawyer would have done, and since every defendant who raises this claim has already lost at trial, a result-based rule would put every defense lawyer on trial for the verdict.⁴ The prejudice requirement ties the claim to the purpose of the guarantee, which is a reliable outcome. Attorney errors come in an infinite variety and are as likely to be harmless in a given case as harmful, so a rule granting relief for every professional lapse would turn the Sixth Amendment into a disciplinary code that overturns sound convictions over mistakes that changed nothing.⁵ Requiring both showings reserves relief for the lawyering failures that actually damage the result.

The defendant carries the burden on both parts. A conviction entered after a trial at which counsel appeared is already a judgment, and the law begins from the position that counsel's assistance was effective, so the person attacking the judgment must produce the proof that overcomes that position.⁶ A claim that fails on either part ends there, because the defendant has not carried the burden on an element the claim requires.

Strickland also declined to turn either prong into a mechanical rule. The Court rejected any per se rule of ineffectiveness, that is, a rule under which a particular kind of lawyer failing automatically establishes the claim without further inquiry, and it described its standards as guides to one ultimate question: the fundamental fairness of the proceeding whose result is being challenged.⁷ The same reasoning explains why the Court refused to measure performance against a fixed set of detailed rules for counsel's conduct. No list can account for the variety of circumstances defense lawyers face or the range of legitimate choices about how to defend a client, and rigid guidelines could pull counsel away from vigorous advocacy.⁸ Prejudice follows the same logic. Because errors are as likely to be harmless as harmful, the general rule is that prejudice must be shown rather than presumed, and the few settings where it is presumed are exceptions taken up later in this chapter.⁹

The same test governs wherever the claim is raised. A defendant may press it on direct appeal, in a motion for a new trial, in state post-conviction proceedings, and in federal habeas corpus, the proceeding in which a federal court reviews whether a state prisoner's custody violates federal law, and the principles do not change with the procedural setting.¹⁰ Habeas review has one wrinkle. A state court's conclusion that counsel rendered effective assistance is not a binding finding of fact, because it applies a legal standard to what counsel did; it is a mixed question of law and fact.¹¹ Under the habeas statute as amended by AEDPA, the federal court's question is whether the state court unreasonably applied Strickland. The Court asked exactly that in Williams v. Taylor, where it treated Strickland as the governing standard and held that the Virginia court had applied it unreasonably.¹²

Because the two prongs are independent, a lawyer's performance and the defendant's fate can diverge, and a pair of scenarios that differ in one fact shows where the line falls. The fact that moves the answer is whether the steps counsel skipped would have mattered.

Both lawyers skipped the same steps, so the performance prong does not separate the two defendants. Prejudice does: in the first scenario the missed investigation had a payoff, and in the second it had none. A lawyer can fall far below professional standards and the defendant can still lose, because the remedy is tied to the reliability of the result and a lapse that left the result untouched leaves nothing to remedy. That is why many convictions are affirmed despite serious defense errors.¹³ It also explains a rule developed later in this chapter: a court that finds the prejudice showing lacking may deny the claim without deciding whether performance was deficient.¹⁴ The next section turns to the first prong and asks how a court decides whether counsel's performance was deficient at all.