Modeling Texas House voting patterns beyond raw agreement

A self-directed demonstration of how Polilens can normalize a full legislative vote record, model member-specific baselines, and turn the results into inspectable network views.

Allen Ye

Founder & Software Engineer

founder@polilens.io

modeled Republicans

1,586

qualifying divided GOP rolls

1,245

modeled GOP connections

cross-party pairs

The analytical problem

Raw agreement rewards easy votes and members who usually vote with everyone.

This demonstration asks which voting relationships exceed what a member's normal behavior and each roll call's difficulty would predict.

Normalize the record

Resolve members, roll calls, bill context, and valid vote positions across the session.

Model expected behavior

Estimate member- and roll-specific baselines instead of treating every agreement equally.

Test the relationships

Apply effect-size, uncertainty, multiple-testing, and bootstrap gates to each pair.

Make the result inspectable

Open portraits and connections to see the members, direction, and contributing roll calls.

Within-party model

Republican voting cohesion

Which Republican pairs vote on the same side more often than member-specific and roll-adjusted baselines predict?

3,741: pairs tested
1,245: modeled connections
201: default display

Open interactive demonstration

Republican ↔ Democrat

Directional cross-party voting patterns

Which Republican and Democratic pairs appear on the other party’s majority side more often than party-specific model baselines predict?

905: opposed-majority rolls
34: modeled pairs
48: modeled directions

Open interactive demonstration

Client relevance

The same workflow can answer organization-specific legislative questions.

The useful output is not limited to a network. The underlying process supports caucus analysis, issue coalitions, target-member research, vote-context reconstruction, and session-level behavioral briefs.

Discuss a scoped analysis

Legislative member and vote data derived in part from LegiScan datasets, licensed under CC BY 4.0. Polilens transformed the source data and produced the analysis shown here.