New network model predicts bill passage in US House from cosponsorship influence

A new study from UCLA and Oxford leverages network science to explain why some bills pass the House while others fail. The researchers, led by Sarah Sotoudeh, Mason Porter, and Sanjukta Krishnagopal, built a cosponsorship network from congressional data, awarding 'influence' between pairs of representatives when they jointly cosponsor a bill that later achieves legislative success (e.g., passes the House or becomes law). This accrued influence metric captures the peer dynamics that traditional measures like degree centrality or PageRank miss. In their analysis, they found that bills sponsored by highly influential cosponsors—as defined by their network measure—are significantly more likely to pass. The model consistently outperformed off-the-shelf centrality measures in predictive accuracy, suggesting that the structure of collaborative relationships is a powerful signal for legislative outcomes.

The work, posted on arXiv (2406.19554), also includes a careful limitations discussion, acknowledging that influence is only one factor among many (e.g., party alignment, committee assignments, public sentiment). But the approach offers a quantitative lens on the 'art of the deal' in Congress: who you cosponsor with, and how successful those partnerships have been, matters for your bill's fate. For analysts, this could enable more data-driven lobbying strategies or better forecasting of legislative pipelines. The technique also generalizes to other collaborative decision-making bodies—not just legislatures—where peer influence shapes outcomes.