Textualism as pretext in the LGBTQ discrimination cases

 “We are all textualists now, [when it suits us]”

When a case is really close, really close, on the textual evidence, and I — assume for the moment I’m with you on the textual evidence. It’s close, okay? We’re not talking about extra-textual stuff. We’re — we’re talking about the text. It’s close. The judge finds it very close. At the end of the day, should he or she take into consideration the massive social upheaval that would be entailed in such a decision, and the possibility that — that Congress didn’t think about it — and that — that is more effective — more appropriate a legislative rather than a judicial function? That’s it. It’s a question of judicial modesty.

~ Justice Gorsuch in Harris Funeral Homes v. Equal Employment Opportunity Commission
(with interruptions removed for readability)

Last week, in Bostock v. Clayton County and R.G. & G.R. Harris Funeral Homes Inc. v. Equal Employment Opportunity Commission, the Court addressed whether discrimination on the basis of sexual orientation and gender identity, respectively, is discrimination on the basis of sex, and thus contrary to Title VII. There has been much commentary on these cases; we are now able to take a look at the numbers, both to predict the outcomes of the cases based on the disagreement gap, and to empirically examine the content of some of the arguments.

This is a horizontal bar chart of predictions of the oral argument based on the ScotusOA model. The Y axis is ordered by strength of prediction of a vote for Petitioner. The X axis is centered on a 50/50 prediction. Numbers to the left of the midline range from 50 to 100 percent in favor of Respondent. Numbers to the right of the midline range from 50 to 100 percent in favor of Petitioner. The chart shows predicted votes in favor of Petitioner in blue with bars extending from the midline to the right and votes in favor of Respondent in orange with bars extending from the midline to the left. Details of this prediction are described in the text.
ScotusOA prediction based on the oral argument in Bostock v. Clayton County (October 8, 2019)
This is a horizontal bar chart of predictions of the oral argument based on the ScotusOA model. The Y axis is ordered by strength of prediction of a vote for Petitioner. The X axis is centered on a 50/50 prediction. Numbers to the left of the midline range from 50 to 100 percent in favor of Respondent. Numbers to the right of the midline range from 50 to 100 percent in favor of Petitioner. The chart shows predicted votes in favor of Petitioner in blue with bars extending from the midline to the right and votes in favor of Respondent in orange with bars extending from the midline to the left. Details of this prediction are described in the text.
ScotusOA prediction based on the oral argument in RG & GR Harris Funeral Homes Inc. v. Equal Employment Opportunity Commission (October 8, 2019)

Our predictive model for Bostock shows Justice Alito clearly on the side of the respondent and Justice Ginsburg equally clearly on the side of the petitioner. The model is surprisingly equivocal about Roberts, Breyer, and Kavanaugh, and it suggests that Sotomayor, Kagan, and Gorsuch are leaning in favor of complainant petitioner. According to our model of Harris Funeral Homes, the justices lined up slightly differently in the transgender case. It shows Roberts and Gorsuch strongly in favor of petitioner (i.e. favoring a narrow reading of title VII) and Breyer, Kagan, and Ginsburg strongly favoring respondent. The model we use to predict votes based on oral argument is not particularly well suited to combined arguments, and this combined argument was especially confused. For example, the justices had multiple questions about bathrooms and gender identity for Pam Karlan in the first argument about the gay skydiver that would have made a lot more sense directed to David Cole in the second case about the transgender funeral home employee.

Others have suggested that Justice Gorsuch could be the potential swing vote in these cases. The reason for this is that, as a professed textualist, he could be expected to vote on the side of the Title VII plaintiffs, for whom the textual argument appears clear: if a woman who is married to or attracted to a man is treated differently to a man who is married to or attracted to a man, that would is clearly a difference based on sex. Similarly, if a woman who chooses to dress or identify as a man is treated differently to a man who chooses to dress or identify as a man, that is also a difference based on sex. Of course, as a conservative, Gorsuch’s ideological preferences pull him in the other direction. As such, these two cases are good tests of just how much of a textualist Gorsuch really is, or alternatively, how much of an ideologue. Our empirical analysis of the argument in Bostock v. Clayton County is encouraging for the plaintiff, but not in Harris Funeral Homes

Of course, if Gorsuch votes to deny gay and transgender individuals the protection that Title VII’s plain text evidently affords them, he will not admit that he is not being a textualist. Rather, as the initial quote suggests, Gorsuch is claiming that the cases are very close, and that it is permissible to stray from textualism under such conditions. The word cloud pictured below is based on what Gorsuch had to say in Harris Funeral Homes v. EEOC. It illustrates Gorsuch’s emphasis on the “close question” and signals that he is likely to vote for a narrow atextual reading of Title VII.

Word cloud as described in the text.
Word cloud based on Justice Gorsuch’s contributions to oral argument in RG & GR Harris Funeral Homes Inc. v. Equal Employment Opportunity Commission (October 8, 2019)

Text or pretext?

There are two massive problems with Gorsuch’s position: first his claim that the textualist case is a close one is implausible, and second, he is not asking the appropriate question.

First, the textualist argument on the other side is not simply weak; rather, it defies ordinary doctrinal logic. The argument goes that discrimination on the basis of sexual orientation or gender identity is not discrimination on the basis of sex because the question is: Is it lawful to treat a gay man differently to a gay woman, or a trans man differently to a trans woman? However, this phrasing of the question only works if you presuppose the answer to the question asked of the Court: whether using sexual orientation or gender identity as a basis for distinguishing is lawful under the statute. To phrase the question in terms of whether you are differentiating between different gay people in the right way presupposes that it is okay to treat gay women different to straight women, and gay men different to straight men. The textualist argument is not close; it is not even close to being close.

But the more damning problem is that a justice is not a textualist if he or she follows textualism only for the easy cases, and then if the case is “close” to the line, he or she can then turn to some other methodological approach to determine the difficult cases. The whole point of a judicial methodology is to apply it to differentiate between cases under and over the line. Looking to the text simply to ascertain whether a purported meaning is close enough to the penumbra of the text to justify then turning to purposivism to fill in the detail is contrary to the very notion of textualism. Put another way, it is not textualism if the key question is not what tool should be applied but rather when should the justice to apply it.

Consequently, gay and trans activists, and those who support them, placing their hopes on Gorsuch being true to his purported judicial ideology is likely to lead to disappointment.  In fact, if Gorsuch is going to support that side of the arguments to any extent, the models make clear that it will be to support the gay complainants but to disappoint the trans complainants. This is in spite of the fact that the textual argument is much stronger in Harris than Bostock: the complainant in Bostock literally was fired for wearing a dress, which would have been permitted to anyone born a woman. The members of the Court, like much of society, have had more time to get used to the idea of people being gay than people being trans, and so find it easier to recognize and reject discrimination against gay people than trans people. To the extent Gorsuch will stray from his ideology, it would be in favor of the text but it will hinge more on that greater tolerance of gay people, and consequently greater tolerance of gender discrimination against trans people.

Prediction:

Title VII plaintiffs (Petitioner in Bostock, Respondent in Harris): Breyer, Ginsburg, Kagan, Sotomayor

Title VII defendants: Alito, Gorsuch, Kavanaugh, Roberts, and Thomas.

Most likely to switch: Gorsuch, but only in Bostock.

How to predict doctrinally complex and morally murky death penalty decisions?

In its first case of the 2019 Term, the Supreme Court heard oral argument in Kahler v. Kansas, on the question of whether insanity is constitutionally required to be an available defense to capital murder. Petitioner argued that, in particular, the test of whether a defendant knows right from wrong is constitutionally required, and so Kansas’s consideration of insanity solely as to whether a mental defect prevents the defendant from forming the requisite mens rea, plus consideration at the penalty stage, is inadequate.

One easy way to know that a particular justice is not buying an argument like Petitioner’s is when the justice gives a detailed recitation of the almost always grisly facts of the death penalty case. Justice Alito did just that, reminding everyone that Petitioner Kahler not only killed his estranged wife and her mother, but executed his two daughters in a seemingly premeditated and cruel fashion, telling one to stop crying as he slaughtered her and her family—except for her brother, who had shown more sympathy for their father and so was spared the carnage, if not the trauma. But predicting Justice Alito’s likely vote in death penalty cases is usually pretty easy; predicting some of the other justices’ votes is more difficult, and requires more sophisticated tools. The figure below provides an assessment of the likely votes of each of the justices, based on everything they said at the oral argument, as well as what they have said in every other case they have heard in their time on the Bench.

ScotusOA prediction based on the oral argument in Kahler v. Kansas. Described in the text.
ScotusOA prediction based on the oral argument in Kahler v. Kansas

Like the Supreme Court, ScotusOA took the summer off from empirically analyzing Supreme Court oral arguments. As such, a quick reminder (or an introduction for the uninitiated) is warranted to explain how we generate our predictions. We measure various aspects of each justices’ behavior at a given oral argument, and how that behavior is tilted to one side or the other. We call this the disagreement gap: the difference in the treatment of the two sides is a strong predictor of how the justices will vote. For instance, interrupting one side more or talking more during the time of one side are both strong predictors of ultimately voting in favor of the other side. We then put those variables in the context (using Bayesian updating) of each justice’s behavior in general at oral argument, to create a prediction for each justice, based on their particular style at oral argument generally.

The Kahler oral argument illustrates the value of taking this contextual (Bayesian) approach: Justice Alito was seemingly the strongest advocate for the Respondent’s position—and we use the term “advocacy” advisedly, based on prior research showing that the justices engage in considerable amounts of advocacy during oral argument. Yet, Justice Ginsburg pips him by 1% in her likelihood to vote for Respondent. Now 1% is obviously not statistically significant, but the fact that the liberal Justice Ginsburg is measuring as effectively equally likely as the conservative Justice Alito to vote against a death penalty defendant is significant. This prediction is a product of that aforementioned context: Justice Alito is very often extremely uneven in which side he directs most of his questions and comments to, whereas Justice Ginsburg is very rarely so tilted to one side. In the Kahler oral argument, she only spoke during the time of Petitioner, and that makes us confident that she will vote for Respondent.

A look at the substance of Justice Ginsburg’s questions supports this conclusion. She opened the questioning (after the first application of the new Supreme Court rule that gives advocates two minutes of uninterrupted speaking time) with a tough challenge for Petitioner’s advocate, Sarah Schrup: whether a judgment of “guilty but insane,” which would result in institutionalization rather than imprisonment or the death penalty, would satisfy the Constitution. She followed up with hypothetical variations that prevented Schrup from answering the question by looking to collateral harms, and then Ginsburg asked about whether the history was more complex than Schrup was implying. All of these were indications of her position, but the strongest indication was the fact that she didn’t ask equally tough questions of the other side, as she typically does.

In terms of predicting the case rather than individual justice’s votes, three justices were harder to analyze on the basis of oral argument: the Chief Justice, Justice Kavanaugh, and the perennially silent Justice Thomas. (Thomas was in fact absent due to illness. He will apparently participate in last Monday’s cases on the basis of the briefs and the transcripts.) Thomas is easy to predict in a similar contextual way, but looking at his votes, which overwhelmingly favor the state in death penalty cases. But the two justices most likely to be the median of the Court, Roberts and Kavanaugh, were far more evenhanded than any of the other justices in their questioning and other conduct.

Roberts asked tough questions of both Schrup and Toby Crouse, the advocate for Kansas, but asked nothing of the more persuasive advocate for the government’s side, Elizabeth B. Prelogar, representing the United States. She directly answered the tough questions that Crouse had struggled with—most memorably, Justice Breyer’s question of how Kansas can justify excusing by reason of insanity someone who commits murder because he deliriously thinks his victim is a dog, but executing a person who deliriously believes that a dog commanded him to commit the murder. Prelogar may have satisfied the Chief’s concerns with that side of the argument. Thus, although the numbers show him close to 50:50 but potentially tilting for Petitioner, we predict him more likely to vote for Respondent.

Kavanaugh was more focused on the detail of doctrine and history, and hammered both sides with significant challenges. Whether assessed qualitatively or quantitatively, Kavanaugh’s questions do not reveal his likely vote in this case. Based on the oral argument alone, he really is too close to call. In terms of looking to past opinions, Kavanaugh was not involved in one of the two big death penalty cases last year, Madison v. Alabama. He voted with the conservative majority in the other case, Bucklew v. Precythe, but wrote a separate concurrence emphasizing the narrowness of the opinion—that is, he rode a fine line in the case, and gave little away about his general proclivities in this area. We expect something similar in this case: either a narrow majority opinion in favor of Respondent written by Kavanaugh, or a narrowing concurrence by Kavanaugh.

Prediction:

Petitioner: Breyer, Kagan, Sotomayor

Respondent: Alito, Ginsburg, Gorsuch, Kavanaugh, Roberts, Thomas

Full disclosure: Petitioner’s advocate, Sarah Schrup, is a colleague of Tonja Jacobi’s at Northwestern Pritzker School of Law, and Jacobi helped moot Schrup in preparation for the oral argument.