Chile's Twin Constitutional Bodies: Roll-Call Votes, Member Records, and Ideology Maps (2021-2023)

Two constitutional assemblies in Chile, working back to back between 2021 and 2023, now have their voting records and political profiles available for study. Together, these bodies ran 5919 roll-call votes on the floor. The first body, called the Constitutional Convention, operated from July 2021 to July 2022 with 155 members and produced 4738 votes. The second body, known as the Constitutional Council, ran from June to November 2023 with 50 members and held 1181 votes. Researchers only counted votes on real proposals, articles, and amendments. They left out procedural votes, non-binding agreements, and votes on internal rules.

The rules for getting elected looked very different in each body. The Convention had to follow gender parity, meaning equal numbers of men and women. It also set aside seventeen seats for indigenous peoples. Its electoral system made it easier for independent candidates to win seats. The Council worked under a more traditional setup. Candidates ran under party or coalition lists, and there was just one extra supernumerary seat for indigenous representatives, available only if a candidate cleared a minimum vote threshold.

Both bodies drafted new constitutional texts, but voters rejected each one. The Convention's proposal went down in a national plebiscite in September 2022, and the Council's proposal failed in December 2023. Even though neither text became law, the vote records still hold value. Researchers can dig into how members behaved, how coalitions formed, and how ideology shaped each round of voting.

To map out where each member stood politically, the team used three separate methods for each body. The W-NOMINATE model places legislators along ideological dimensions based on voting patterns. A Bayesian item-response model goes further by producing posterior means plus 95% credible intervals, which show the range of uncertainty around each estimate. A third dynamic model tracks how positions shifted within each body across four time periods, capturing movement rather than a single static snapshot. Running all three methods independently makes the results easier to cross-check.

The member-level data goes beyond ideology. It includes political party affiliation, the electoral pact each member ran under, the district or constituency they represented, their region, date of birth, sex, and whether they held a reserved indigenous seat. Each member also gets a political-sector label from a five-category scale: Left, Center-Left, Center, Center-Right, or Right. This rich mix of variables opens doors for comparative work on constitution-making, legislative behavior, and institutional design. Scholars in political science, public policy, and related fields can use the dataset to ask fresh questions about how rules shape representation and how ideology plays out when a country tries to rewrite its founding document.