Take the campaign-finance data with you.

Point your AI at this page.

A browsing-only chat may be unable to fetch a ZIP or run calculations. In that case, download the archive and attach it to a chat with file analysis. Compression makes transfer smaller; it does not put the entire dataset into a model’s context.

Just the transaction table?

These two files contain transactions only. Use the ZIP for matching-fund classifications, committee/entity tables, candidate links and supporting evidence. Source text is data; spreadsheet software may interpret it as formulas if imported without text types.

Start an independent analysis.

1. Download and unzip the recommended bundle. Read README.md and schema.json.
2. Query tables/transactions.parquet. analysis.sql includes examples for cash raised and donation mean, median, quartiles and tied modes.
3. Join reviewed_matching_ids to separate City matching money. Use the reviewed candidate crosswalk and governor evidence for race links.
4. Cite the snapshot, the Lab and ORESTAR. Save your downloaded manifest and ZIP so the analysis can be reproduced after the next update.

-- After unzipping, run from the archive directory with DuckDB.
SELECT committee_id, max(committee_name) AS committee,
       sum(amount_cents) / 100.0 AS cash_raised_dollars
FROM read_parquet('tables/transactions.parquet')
WHERE basis = 'cash_contribution'
GROUP BY committee_id
ORDER BY cash_raised_dollars DESC;

Donations, payments, loans, refunds and unpaid bills are different accounting bases. Adding them together produces a misleading total. Transfers between committees can count the same money more than once statewide.

What “all the data” covers.

All current records held in our published Oregon ledger, not all ORESTAR history or a complete election cycle. Includes all observed filers, not only council/governor campaigns. Federal FEC records are excluded.

• Manual-export coverage is provisional; older-dated late filings, removals and amendments outside the supplied exports may be missing.
• Archived maps, balances, event interpretations and network tests retain their individual dates.
• No donor street addresses, donor-level ZIP codes, employer/occupation fields, unrestricted purpose descriptions or private research files are included.

Reported names and cities/states remain available. Donor groups are conservative filing matches, not verified unique people. Business records do not establish corporate size or ownership. A missing candidate is missing coverage, not $0 raised.

Curated data is available under CC BY 4.0; published editorial interpretations retain CC BY-ND 4.0. Underlying public records retain their source terms.