Employment discrimination doctrine is a mess, and one of the messiest parts concerns causation. Problems with causation have been the focal point of many articles in recent years, often in response to the “tortification” of employment discrimination law. You might think that there is nothing more to say, and that we’re just stuck with the mess. But James Macleod’s article, Ordinary Causation: A Study in Experimental Statutory Interpretation has persuaded me that neither is true.
In this article, Professor Macleod breathes fresh life into interpretation of Title VII by using the tools of experimental philosophy to explore the meaning of “because of” and other statutory causal language. What better way to determine the ordinary public meaning of a phrase, particularly a phrase in context, than to survey a representative sample of the population, ask whether a particular result was because of the reason described in the statute, and then share that information publicly? Professor Macleod did just that, and his article makes a case for this approach and then reports on his results.
The paper’s first two sections summarize familiar ground. Part I analyzes recent case law on statutory causation, and Part II summarizes why courts are properly concerned about ordinary meaning in legal interpretation. Recent years have seen significant decisions interpreting causal language in statutes. Two of these cases are employment discrimination cases: Gross v. FBL Financial Services, under the Age Discrimination in Employment Act, and University of Texas Southwestern Medical Center v. Nassar, under Title VII’s retaliation provision. In both cases, the Court held that the plain meaning of “because of” required plaintiffs to prove that an improper motive was a but-for cause of the adverse employment actions they suffered.
The Court’s own view is that that its job is to find the plain meaning of the statutory language. Professor Macleod agrees, arguing that this search for plain meaning is particularly appropriate for concepts like causation, which are core common law concepts and embody a sense of moral wrong. But Parts III and IV, which form the core of the paper, draw out the weaknesses of judges’ methodology for finding plain meaning, as well as explaining the benefits of the survey experiment method.
In Part III, Professor Macleod makes a persuasive case for the weakness of judges’ current methodology. Judges generally rely on introspection, hypothetical test situations (called intuition pumps), and dictionaries. Introspection is a poor tool because judges, like any of us, are subject to cognitive biases like motivated reasoning and confirmation bias. Because different hypotheticals, or intuition pumps, can point to different conclusions, judges may discount those that don’t match their initial intuition. This is simply another way that motivated reasoning and confirmation bias limit the power of individual reasoning. And dictionaries, though external, cannot use terms in context, which is often indispensable for understanding how language actually operates. Moreover, dictionaries often provide so many alternative meanings that a judge must pick from among them, often in ways that confirm initial intuitions. Confounding the operation of these biases, Professor Macleod notes, individuals are overconfident in their intuitions. We tend to think that we are representative of the population at large, even when we are wrong about what most people would think.
Part IV then explains the survey experiment method of finding plain meaning in context, which is already being used in litigation. Professor Macleod makes the case for why it is an improvement over other methods and reports the results of his own survey experiment: that courts have often reached the wrong outcomes about causation.
The key intuition underlying this approach is so straightforward, it almost seems too easy: “to find public meaning, ask the public.” But framing an experiment to discern that meaning is no easy task. Professor Macleod conducted a nationally representative survey of nearly 1500 jury-eligible lay people. Participants were randomly assigned to read a short vignette modeled on one of the cases the Court had adopted a but-for approach in. They were then asked about the cause of the result. In one vignette modeled on Gross, for example, participants were asked whether the protagonist terminated the employee “because of” the employee’s age in a situation where age and one other reason could have caused the outcome. Each vignette had four variations (participants were randomly assigned to which option they answered), in which the improper reason was necessary and sufficient, necessary but insufficient, unnecessary but sufficient, or unnecessary and insufficient. In this way, Professor Macleod was able to discern the public’s understanding of when someone is fired “because of” an improper reason.
The study found, essentially, six things. First, a sizeable majority of respondents found causation present in situations where but-for cause was absent–in fact even where independent sufficiency was also absent. Second, sufficiency’s presence or absence played a much larger role in responses than the presence or absence of but-for causation. Third, the substantial factor test seemed to be what a large majority of participants interpreted the statutory language to require. Fourth, the moral preferences of participants followed sufficiency rather than but-for causation. Fifth, participants in the minority were just as confident that their interpretation was the only right one as were participants in the majority. And Sixth, the results were responsive to small differences in changes of causal language even in highly blameworthy contexts, showing that language mattered. As a result of these findings, Professor Macleod concluded that courts have incorrectly interpreted the plain mean of “because of” as requiring but-for causation. Instead, “because of” pretty plainly means “substantial factor” to the overwhelming majority of people.
Professor Macleod acknowledges the limitations to this approach to statutory interpretation in some contexts. For example, survey experiments can test current usage and concepts, but not historical ones. So where the meaning of terms has changed significantly over time, say, for example, what sex or race are, this approach to discerning meaning may be less appealing to courts.
Professor Macleod has written an engaging article with far-reaching implications. In fact, I am already trying to think of ways to use his approach to better teach my courses. I look forward to more work by Professor Macleod in this area.