Archive: A dataset of tabulated results from American ranked-choice voting elections

In the past two decades, a growing number of ranked-choice voting (RCV) elections have been conducted in various jurisdictions across the United States. However, tabulated results of RCV have been reported and stored in widely different styles across places and years, making it infeasible for researchers to perform systematic analyses of vote tabulations. We introduce ARCHIVE: a dataset of standardized tabulated results for over 7600 round-level candidate vote counts from 514 American RCV elections, 2004-2024. To construct the dataset, we develop a methodological procedure based on large-language models that semi-automatically collect, standardize, and store candidate vote counts while instantly validating the resulting information. Our dataset releases multiple levels of data, including election metadata, round-level attributes, and candidate-level information with consistent election identifiers, allowing users to address key questions in electoral competition under RCV. Our project and user interface are also available at www.archive-rcv.com. Our associated paper is available here: https://osf.io/preprints/socarxiv/27hgx_v1