A Computational Perspective on Political Cooperation and Opposition in Democracy: A Longitudinal Multiplex Network Analysis of the Brazilian Chamber of Deputies, 2003-2025

Legislative polarization can coexist with increasing institutional interdependence when political competition and opposition intensify within expanding arenas of negotiation. This article examines the Brazilian Chamber of Deputies between 2003 and 2025 through longitudinal multiplex network analysis, combining substantive legislative co-sponsorship, lagged roll-call voting similarity measured by Cohen’s κ, shared committee participation, and dyadic generalized estimating equation (GEE) models. Substantive co-sponsorship ties increased substantially, and the largest connected component expanded, indicating greater structural integration. Nevertheless, weighted cross-party cooperation declined sharply after 2018 and recovered only partially after 2022. The association between previous voting similarity and subsequent cross-party tie formation strengthened across periods: the odds ratios for a one-standard-deviation increase in lagged similarity were 1.027 in 2004-2018, 1.339 in 2019-2022, and 1.440 in 2023-2025. The corresponding annual mean gaps in lagged Cohen’s κ between collaborating and non-collaborating cross-party dyads increased from approximately 0.013 to 0.088 and 0.161. These results distinguish network integration from politically cross-cutting cooperation: growing institutional interdependence may coexist with stronger political competition, opposition, and selective partner choice. The findings are associational rather than causal and remain subject to sampling, dyadic dependence, measurement, and short-window limitations.