Andres CarrizosaRice University
Legislative stability exists in the House of Representatives, but there is not a consensus on whether this stability is produced by a Preference Induced Equilibrium (PIE), which predict that if preferences are unidimensional an equilibrium can be found at the position of the median legislator, or a Structure Induced Equilibrium (SIE), which predicts that multidimensional legislative preferences must be constrained by institutions to produce stable outcomes. While both explanations predict that roll-call behavior will be unidimensional, they make competing predictions for less constrained behavior such as cosponsorship behavior. By relying on Bayesian item-response methods that allow for the assessment of dimensionality of both cosponsorship behavior and roll-call behavior on a comparable scale--which has been problematic in the past--I provide evidence in support of a SIE in the House. While roll-call behavior is unidimensional, only around 30% of cosponsorship behavior is explained by the first dimension of legislative preferences.
I have included more analysis about what including a second dimension is useful for. What can multidimensional models capture about legislative behavior that cannot be captured by unidimensional models? I explore this additional question.
pdf : 0.45mb
pdf : 0.19mb
Appendix: The Dimensionality of Cosponsorship Behavior in the House of Representatives