Some trial errors are devastating. Others are real errors but do not substantially affect the outcome. A valid method for measuring trial fairness must be able to tell the difference.
David Leroy Washington’s case produced one of the most important criminal procedure decisions in American law: Strickland v. Washington. The case established the modern framework for ineffective assistance of counsel claims. Under Strickland, a defendant must show not only that counsel performed deficiently, but also that the deficiency caused prejudice—meaning a reasonable probability of a different outcome.
Washington pleaded guilty to multiple murders in Florida and was sentenced to death. His lawyer spent very little time preparing for the sentencing hearing and failed to present witnesses who might have offered mitigating evidence. But the U.S. Supreme Court ultimately concluded that the omission did not create a reasonable probability of a different sentence. Compared to cases like George Porter’s, Washington’s mitigation evidence was weaker, the aggravating evidence was stronger, and presenting “good character” evidence could have opened the door to damaging rebuttal evidence from the prosecution.
My research asks whether the empirical method developed in Measuring Fairness reaches the same conclusion.
Testing Whether the Method Can Identify Harmless Omissions
The Washington case study was not designed to show that every failure to present mitigation evidence is harmful. It was designed to test whether the method can correctly classify an error or omission that did not cause substantial harm.
That matters. A valid method should not simply find prejudice everywhere. It must accurately classify both harmful and harmless errors. In statistical terms, it should minimize both Type I errors—mistakenly treating a harmless error as harmful—and Type II errors—mistakenly treating a harmful error as harmless.
The study used the same general research design as the Porter case study. Respondents were assigned to review either the sentencing case as it was actually presented or a hypothetical version that included the mitigation evidence omitted by Washington’s attorney. The analysis then used sentencing preferences and a jury deliberation model to estimate the probability of a death sentence in each condition.
The result was notably different from Porter. In Washington’s case, the omitted mitigation evidence did not meaningfully reduce the probability of a death sentence. The estimated effect was a .011 decrease in the probability of death—essentially no substantial harm. In fact, the analysis suggested that presenting the omitted evidence might have slightly increased the risk by allowing the prosecution to introduce damaging rebuttal evidence.
Why Correctly Identifying Harmless Error Matters
Fair trial analysis should not be understood as a tool only for defendants. It is a tool for distinguishing fair outcomes from unfair ones.
When a trial error or omission probably changed the outcome, justice requires correction. That was the lesson of the Porter case. But when a defendant received a fair trial or sentencing proceeding despite an error, justice is served by upholding the conviction or sentence.
Washington’s case illustrates that point. His lawyer’s preparation may have been deficient, but the omitted mitigation evidence was not strong enough to undermine confidence in the death sentence. Survey respondents continued to view the aggravating facts as overwhelming. Some even viewed the additional mitigation evidence as making Washington appear more deserving of death because it opened the door to a fuller account of his character and conduct.
This case study therefore supports the larger claim of Measuring Fairness: empirical analysis can help courts evaluate prejudice more accurately. The method should identify cases like Porter, where omitted mitigation evidence likely changed the outcome. It should also identify cases like Washington, where the omission did not cause substantial harm.
That balance is essential. Measuring fairness does not mean reversing every conviction or sentence affected by error. It means separating errors that likely changed the outcome from errors that did not.
The larger point is not that every trial error requires relief. The larger point is that fairness should be measured carefully enough to know when relief is warranted—and when it is not.