The authority of U.S. Supreme Court majority opinions rests largely on their use as precedents. Existing studies of citation patterns face two key limitations: dyadic citations are usually aggregated to the case level, and citations are treated as if they arise independently. This paper presents a method that addresses both issues and models citations at the dyadic, network level.
🗂️ A New Dyadic Network Approach
The citation exponential random graph model (citation ERGM) is introduced as a way to treat citations between opinions as a network tie rather than independent events. User-friendly software accompanies the model, enabling researchers to simultaneously estimate:
- the effects of observable case characteristics on citation formation, and
- complex forms of network dependence that shape citation behavior.
📚 What Was Analyzed: All Supreme Court Opinions, 1950–2015
- Network data include every Supreme Court case decided between 1950 and 2015.
- Citations are modeled at the dyadic (opinion-to-opinion) level rather than aggregated to cases.
- The accompanying software implements the citation ERGM for applied use.
🔍 Key Findings
- Strong evidence of network dependence processes in citation formation, including reciprocity, transitivity, and popularity.
- These dependence effects are both substantively and statistically as important as traditional exogenous covariates (case characteristics).
- Treating citations as independent or aggregating dyads to cases can obscure these network processes.
💡 Why This Matters
- Models of Supreme Court citations should incorporate both case attributes and the structure of past citations to accurately capture precedent formation.
- The citation ERGM and its software provide a practical tool for researchers interested in the generative dynamics of legal citations and the social structure of judicial precedent.
