Download the data
All data are released under CC BY 4.0. Please cite the dataset and the version you used. Every version is kept permanently at its own address.
Current version: …
Versions
A new version is published whenever coding is revised, candidates update their sites, or a new election is added. Earlier versions stay available so results can be reproduced.
| Version | Election | Status | Notes | Files |
|---|
Codebook
The codebook defines the 10 main issues, the five positions, the "Other" topic labels, and the coding rules. In short: only the candidate's own campaign website is used; an issue counts as mentioned when mentioned = Yes; evidence_quote is at most 15 verbatim words and page_used links to the page.
Reading the files
candidates.csv: one row per candidate, with columns cand_id, candidate, party, state, district, office, campaign_url, site_readable, health_any, n_main. Use site_readable = True as the denominator for percentages.
coding_long.csv: one row per candidate × issue, with columns cand_id, issue, issue_type (main/other), mentioned, stance, approaches, evidence_quote, page_used, mention_count. Each candidate has exactly 10 main rows plus one row per other topic named.
issue_frequencies.csv, issue_stance.csv, state_summary.csv, state_issue_party.csv: summary tables derived from the two files above (shares by party, office and state, with Fisher exact p-values for party differences).
R users: fread("https://healthcarepolicies.org/data/releases/<version>/coding_long.csv") reads a file directly. The code for this site and the paper's R analysis scripts are on GitHub.