Frequently asked questions
Get direct answers about Akashic coverage, margins, estimates, custom Places, forecasts, downloads, API access, citation, and corrections.
Does Akashic cover every presidential election since 1876?
No. The complete continuous public record claim is 1892–2024. Returns from 1876–1888 exist only for part of the country and must remain labeled partial.
Why does a city or district have fewer historical elections than a county?
Sub-county geography depends on precinct and block-level source records. Those sources are much more recent and less uniform than county returns. The current sub-county product contract begins in 2004 and includes method flags where local detail is estimated or reconciled.
Is D+8 a vote share?
No. It is an eight-percentage-point Democratic margin over the Republican candidate. The winning candidate's vote share could be different depending on third-party votes.
Does a missing value mean zero?
No. Zero, null, absent, suppressed, partial, and estimated are different states. The data dictionary explains how to handle them.
Are all local election values officially reported at that exact geography?
No. Some are exact rollups of reported units; some are reconciled disaggregations from a larger certified total; some historical details are estimated; some remain missing. Akashic labels the method and preserves the authoritative parent total.
Can a custom Place have disconnected parts?
Yes. A user-created Place may contain geographically separated components. All components should persist through preview, save, edit, fork, share, and rollup.
Are forecasts official results?
No. Forecasts are probabilistic model outputs. Certified election results and forecast editions are separate products. Retrospective backtests are labeled research and are not counted as prospectively published forecasts.
Is the API free?
The public read interfaces and static JSON records do not require an API key. Cache responses, use bulk archives for large analysis, and follow the license and attribution requirements. Protected user, billing, admin, and write actions are outside the public read API.
Can I embed Akashic on another site?
Yes, through supported iframe widgets and oEmbed. Keep the built-in attribution and do not present the embed as Akashic endorsement of surrounding commentary.
How should I cite a figure?
Use the exact page or record URL, metric, geography, year/vintage, retrieval date, source/method qualifier, and Akashic as publisher. See the citation guide for examples.
Why can two pages show different demographic vintages?
Coverage and build pipelines differ by tier and source. Check each page's provenance and as-of metadata rather than assuming every module was produced from one survey release.
What should I do if I find an error?
Record the exact URL, field or chart, observed value, expected value, source supporting the correction, and retrieval date. Distinguish a source-data issue from a formatting, boundary, method, or stale-release issue. Use the repository issue tracker or the contact path published by Akashic.
Does Akashic endorse a party or candidate?
The documentation and editorial standard require the same descriptive and evidentiary standard for every party, candidate, and constituency. Data and forecast outputs should be assessed through their sources and methods, not inferred institutional preference.