Forecasting Votes in Hellerstedt

On March 2nd, the Court heard oral arguments in perhaps the most publicized case of the Term – Whole Woman’s Health v. Hellerstedt (No. 15-274).  The case brings into question Texas’ new controversial law limiting who can perform abortions.

Prior to oral arguments expectations were already high that the Justices would split along ideological lines with the more conservative Justices – Thomas, Alito, and Roberts voting to uphold the law and the more liberal Justices – Breyer, Ginsburg, Kagan, and Sotomayor voting to strike down the law.  If this were the case then the outcome would depend on the “swing Justice,” Justice Kennedy.  Kennedy could create a stalemate by aligning with conservatives or could help strike down the law by voting with the more liberal bloc.  Noah Feldman provides some insight into Justice Kennedy’s possible position noting that Kennedy is the last remaining Justice on the Court from the Casey majority and according to Feldman, “He won’t accept an interpretation of that precedent that would be seen as vitiating it.”

Reports from oral arguments solidify these hypotheses regarding the Justices’ stances. Headlines after oral arguments include: “Supreme Court Appears Sharply Divided as It Hears Texas Abortion Case” (New York Times), Liberal Supreme Court Justices Just Slammed Texas’ Abortion Restrictions (Mother Jones), and “Liberal justices critical of Texas abortion law” (CNN).  The CNN report also underscored Kennedy’s position in the middle: “Justice Anthony Kennedy, whose vote will be crucial in this closely divided case, asked questions of both sides. At one point he suggested that the Court might remand the case back to the lower court to develop a fuller record.”

Using the qdap package in R, I performed an empirical analysis based on the oral argument transcript that more or less supports these expectations.  There are three important caveats to this analysis. First, it is based on a sample of text from one oral argument session without imputing any prior information about the Justices.  The application of these findings is thus limited. Second, part of the analysis looks at the sentiment of the Justices talking-turns (questions and statements).  The sentiment analysis dictionary is generic and so at best provides an extremely crude and sometimes likely inaccurate picture of the Justices’ true sentiments as I will describe below.  Finally, as usual (aside from the anomaly the other day), Justice Thomas was silent during oral arguments so his prior voting pattern is the only evidence we have on which to base any hypotheses about his likely voting direction in the case.

I split the oral argument transcript between the Justices talking-turns during petitioner’s and respondent’s arguments.  First I looked at the types of questions the Justices’ asked the petitioner (looking to strike down the law) which is depicted graphically below.


Initially I looked for clustering among the voting blocs.  This is immediately apparent.  In the top left quadrant Justice Roberts and Alito not only asked the most questions of petitioner (the darker shaded rectangles), but also asked generally more incisive questions (“where,” “what,” and “why”-based).  This evidence of amount of questions may be the most telling piece from the oral arguments (several studies including a recent piece from Professors Lee Epstein and William Landes, and Judge Posner (2009) argue that “Supreme Court Justices are more prone to question at oral argument parties against whom they will vote than parties for whom they will vote.”)  The more liberal Justices tended to ask fewer total questions of the petitioner and from the darkly shaded rectangles it is apparent that the types of questions were not typically aimed at grilling the petitioner (for example Justice Sotomayor focused on “were,” “may,” and “has”-based questions).   Note that Justice Kennedy appears more similar to the liberal-leaning Justices here than to the conservatives.


The questioning of the respondent stands in contrast to the petitioner’s questioning.  The boxes for the total number of questions are darkest for the liberal Justices with Justice Roberts asking the fewest number of questions and Justice Sotomayor the most.  The more incisive questions tend to come from the more liberal Justices as is evident from the top shaded rectangles.  The only outlier whose questioning pattern does not appear different in this figure is (perhaps unsurprisingly based on my comment in the last paragraph) Justice Kennedy.

The next two figures highlight some of these points and complicate them a bit as well.  These figures are also heat map with three important rows: one for total sentences, one for total words used, and below that, one for a Justice’s average sentiment.  First the figure for the petitioner’s questioning:


The totals for words and sentences correlates with the findings from the total questions discussed above, with the conservative Justices talking more than the liberal Justices.  For sentiment, while Justice Roberts has the most negative sentiment value here (as might be expected), Justice Alito is as positive as the most positive liberals.  After examining the oral argument transcripts, this more seems indicate a need for a specific oral argument sentiment dictionary than to present surprising evidence of Justice Alito’s position in the case.  Take, for example, Justice Alito’s statement on page 12 of the transcript, “Because­­ if you go through this ­­now we’re not talking about a huge number of facilities. I really don’t understand why you could not have put in evidence about each particular clinic and to show why the clinic closed. And as to some of them, there is ­­there’s information that they closed for reasons that had nothing to do with this law.” The wording itself does not necessarily connote a negative sentiment even though the statement is designed to press the petitioner.

Justice Robert’s focus on the proper test for the law in the case may lead to his negative sentiment.  The words he reuses during the petitioner’s questioning are “undue,” “burden,” and “obstacle” among others.  These all connote a negative sentiment even though they are really just aspects of balancing tests.

The figure for the respondent’s questioning corroborates the same positions.


The liberals in the middle of the figure have the most sentences and use the most words (the darkly shaded boxes) during respondent’s questioning while the boxes are lightest for Justices Alito, Roberts, and Kennedy.  While interestingly Justice Roberts has contrasting sentiment from the previous figure, here with the most positive average sentiment, the remainder of the Justices findings for sentiment are not as telling.  The liberal Justices are predominately in the mid-level of sentiment region and Justice Alito appears to have a highly negative sentiment.

The most significant takeaway comes from the Justices’ amounts of talking.  If prior positions hold true then the results for talking corroborate that Justices Roberts and Alito should support the respondent while Justices Kagan, Breyer, Ginsburg, and Sotomayor should support the petitioner.  Justice Thomas provided us with no additional information during these oral arguments and Justice Kennedy still remains a wildcard.

8 Comments Add yours

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s