Political Threads in Legal Tapestry: A Computational Analysis of Executive Branch Legal Interpretation, 1934–2022

Abstract

How does the push and pull between law and politics shape the work of executive branch lawyers charged with providing “neutral” interpretations of the law?  To shed light on this longstanding question, this Article undertakes the first large-scale computational analysis of legal interpretation in the executive branch of the United States federal government.  Using a new dataset comprised of the texts of 12,879 pages of opinions issued by the Office of Legal Counsel (OLC) in the Department of Justice (DOJ), I use machine learning methods to explore markers of partisanship and ideology in these high-profile executive branch decisions. 

Several important findings emerge from this analysis.  Broadly, I detect a significant partisan gap in the language used by the OLC.  Yet this gap appears to have experienced limited growth in recent years, suggesting some degree of disconnect between the OLC and the broader, increasingly polarized political environment.  Focusing on more fine-grained distinctions in the use of specific words and phrases, I find that Democrats and Republicans have differed along a number of salient dimensions, including the types of legal arguments invoked, substantive policy-related language, and institutional issues involving the separation of powers.  One of the sharpest distinctions relates to executive power, with Republicans being more likely to adopt language associated with expansive presidential authority.  I also find evidence that Republicans have more closely associated executive power with various markers of authoritarianism, such as language related to violence, the armed forces, and nationalism.  But neither party appears to have consistently associated executive power with markers of democracy more than the other, complicating this picture. 

These findings have important implications for law and politics.  To start, the existence of significant partisan differences in the OLC’s language use casts doubt on the prospect that high-ranking executive branch lawyers will act as “neutral expositors” when rendering legal opinions.  At the same time, the apparent disconnect from broader trends in polarization arguably offers some sliver of hope for those who would prefer neutrality.  Another issue that has attracted both scholarly and popular attention relates to the role of executive branch lawyers in safeguarding against—or alternatively facilitating—democratic backsliding.  My results highlight the risks of relying on executive branch lawyers to uphold constitutional guardrails against presidential overreach, especially amidst attempts by certain political actors to increase politicization within the DOJ.  Such concerns underscore the potential importance of other institutions in preserving democratic norms.

Keywords: law and politics, empirical legal studies, machine learning, constitutional law, administrative law, separation of powers, legal ethics, American political institutions, bureaucratic politics

JEL Classification: K00, K30, K40