CandiData

Social media platforms shape how citizens encounter political speech. Yet there is no single reliable open source mapping every elected official and candidate in the United States to their social media accounts across every major platform. The datasets that exist are fragmented, outdated, or federal-only.

You cannot study how platforms treat politicians differently from ordinary users if you cannot first say, authoritatively, which accounts belong to which politicians.

CandiData is that reference layer: a comprehensive, human-verified database of U.S. elected officials, candidates, and their social media presence, built for algorithmic transparency and social listening research.

The coverage problem it solves

Most political datasets stop at Congress, because Congress is easy — 535 people, well documented. CandiData goes all the way down.

Federal. All 535 members of Congress, the President and Vice President, and delegates from DC and the territories.

State. Governors, lieutenant governors, attorneys general, secretaries of state, treasurers, auditors, and legislators — nearly 8,000 state-level officeholders.

Local. Over 30,000 local officeholders from 984 counties and thousands of municipalities: county commissioners, city council members, mayors, judges, school board members.

Across all 50 states, DC, and five territories — Puerto Rico, Guam, the U.S. Virgin Islands, American Samoa, and the Northern Mariana Islands — covering 435 congressional districts and more than 6,600 state legislative districts.

Accounts, not just people

Politicians don’t have an account. They have several, doing different jobs, and conflating them produces bad research. CandiData distinguishes three channel types:

Campaign — used for electoral campaigning and political messaging. The primary focus of initial collection. Official office — government-issued accounts used in an official capacity (@POTUS, @SenatorSmith). Personal — private accounts that may carry political content but aren’t official or campaign channels.

Tracked across Facebook, X, Instagram, YouTube, TikTok, and Bluesky, plus Truth Social, Gettr, Rumble, Telegram, and Threads. Platform coverage moves as platforms rise and fall.

Each account record carries its URL, handle, verification status, and which researcher entered and verified it — the provenance that makes it usable as evidence rather than just as a list.

Modelling political life as it actually is

The data model handles the messy parts most schemas flatten. People hold multiple offices over time. They run in elections and lose. They change parties. Party affiliation is many-to-many with one marked primary, because states with fusion voting like New York need that. Offices belong to bodies; bodies have a level, branch, and chamber; districts link to offices via OCD-IDs so the data joins cleanly to everything else in the civic data ecosystem.

Unglamorous, and exactly the kind of reference infrastructure everyone else assumes already exists.