Replication Data for: Analyzing the Rhetoric of Supreme Court Confirmation Hearings

The below data and code replicates the main and appendix materials for \"Analyzing the Rhetorical Behaviors of Supreme Court Confirmation Hearings.\" An important note for these materials: The author's preference for software packages and programming languages to complete different layers of the research design constitutes the incorporation of Microsoft Excel (data collection), STATA (data analysis), and Python (sentiment classification). All replication materials (including STATA do-files and .py files are included below). The author is happy to discuss any data limitations, as well as the procedure for classifying rhetoric using supervised machine learning processes like the one incorporated in this work (Naive Bayes). Motivation(s): Analyses of Supreme Court confirmation hearings routinely contend with hostility and political overtones becoming ingrained in the process. Accounts from political scientists, normative observers, and participants in these hearings themselves, often point to distinctive moments in the recent history of these hearings as the source of these developments. Chief among them is the contention that the rejection of Robert Bork in 1987 instigated a decades-long collection of grievances that plagued every subsequent hearing. Some observers have continued to support this framing by asserting that even more recent events like the infamous Thomas and Kavanaugh hearings, as well as the Republican Senate's refusal to grant Merrick Garland a hearing, built-from and expand-upon the grievances emanating from Bork. However, more recent literature exploring the contextual substance of interactions between committee members and nominees have questioned these contentions in favor of viewing rhetorical behaviors as a reflection of party dynamics and the balance of interbranch political power. Approach: I apply supervised machine learning using Naive Bayes to analyze the rhetorical behaviors of committee members during hearings spanning 1970 to 2020. Findings: My results ultimately suggest two core empirical findings. First, I find little evidence to support the contention that certain key events, Bork or otherwise, significantly impacted rhetorical behaviors. Any layer of analysis, regardless of how many of these aforementioned events are included in the research design, fails to discern any definitive moment where these events might explain longitudinal variation. Second, while the notion that grievances emanating from key events is the primary culprit fails to provide intuitive answers, I do find support for a conjoined influence of strategic party goals and the balance of interbranch political power. That is, rhetorical behaviors are often the result a committee member's position as an in (i.e., sharing the president's political party) or out-party member, as well as whether the hearing exists in unified (i.e., the Senate majority shares the party of the appointing president) or divided government. I find that the behaviors of in-party members alters significantly when they are in the minority. Under these circumstances, they become increasingly more probable to incorporate negative rhetoric, likely as a reflection of the out-party's position as the Senate majority now bearing the weight of real obstructionist consequences. I conclude this work by noting how these conditions and behaviors can produce negative consequences for the Court's institutional support.