SCCC — Scaled and Classified Congressional Communication

About the data

This website allows users to explore and download the data introduced in "Measuring Partisanship and Representation in Online Congressional Communications" (American Political Science Review 2026). The dataset covers the 111th through 117th Congresses (2009–2022) and contains messages sent by 958 Representatives and 120 Senators.

Each message was coded into six categories drawn from Mayhew (1974), Fenno (1978) and Russell (2018): advertising, credit claiming, position taking, constituent service, negative partisanship and bipartisanship. Categories are not mutually exclusive. Labels come from BERTweet classifiers trained on roughly 43,000 hand-coded messages.

The partisan score places each message on a −1 (most Democratic) to +1 (most Republican) scale using a class affinity model fit separately for each Congress. Partisan extremity is the absolute value of that score. Validation against DW-NOMINATE, CF scores, presidential support scores and human coders is reported in the article and its supplementary material.

Tweets were collected from the Twitter API, Facebook posts through CrowdTangle, and email newsletters from Lindsey Cormack's DCinbox archive. Because the platforms' terms of use do not permit redistribution of message text, the public files carry identifiers (tweet ID, post URL, newsletter ID) rather than text; researchers with API access can rehydrate messages from those identifiers. For more questions about data access, please email one of the authors.

Contact: mkistner@tamu.edu or michael.heseltine@sociology.ox.ac.uk. Code for this site and the release pipeline is on GitHub.

Explore

Filter by platform, Congress, party, chamber, state or member. Results are message-weighted: proportions are the share of messages in each category and the partisan score is the mean across scored messages. Newsletter figures are computed per sentence, the unit on which the classifiers and scaling model ran, so category shares are comparable across platforms.

Platform From Congress To Congress Party Chamber State Member

Share of messages by category

By Congress

By member

Download

Two member-level tables cover most uses. The message-level files below them contain every tweet, post and newsletter in the dataset, without the text.

Member-level tables

members.csv: One row per member and Congress: name, party, chamber, state and district, plus message counts, the share of messages in each category, and mean partisan score and extremity — pooled across platforms and separately for Twitter (Twt), Facebook (FB) and newsletters (NL). Start here for member-level measures.
summary.csv: The same measures in long format, one row per member, Congress and platform, with mean engagement. This is the table behind the explorer above.
Codebook: Variable definitions, classifier and scaling details, known limitations.

Message-level files

Each row is one tweet, one Facebook post, one newsletter or one newsletter sentence, with its platform identifier, date, member ICPSR ID, Congress, engagement counts (social media only), the eight classifier labels and the two partisanship scores. Newsletters were classified sentence by sentence: the newsletter files mark a category if any sentence carries it, while the sentence files, which the member-level tables and the explorer are built from, give the per-sentence labels. Join any file to members.csv on MemberICPSR and Congress.

Cite

If you use the data, please cite the article and the dataset.

Kistner, Michael, Michael Heseltine, Robert Alvarez, Maya Fitch, Lucas Lothamer, and Elizabeth Simas. 2026. "Measuring Partisanship and Representation in Online Congressional Communications." American Political Science Review, 1–18. doi:10.1017/S0003055426101841

Kistner, Michael, Michael Heseltine, Robert Alvarez, Maya Fitch, Lucas Lothamer, and Elizabeth Simas. 2026. "Replication Data for: Measuring Partisanship and Representation in Online Congressional Communications." Harvard Dataverse. doi:10.7910/DVN/19MIBB

@article{kistner2026sccc,
  title   = {Measuring Partisanship and Representation in Online Congressional Communications},
  author  = {Kistner, Michael and Heseltine, Michael and Alvarez, Robert and Fitch, Maya and Lothamer, Lucas and Simas, Elizabeth},
  journal = {American Political Science Review},
  year    = {2026},
  pages   = {1--18},
  doi     = {10.1017/S0003055426101841}
}

If you use the newsletter data, please also cite the DCinbox archive:

Cormack, Lindsey. 2017. "DCinbox—Capturing Every Congressional Constituent E-Newsletter from 2009 Onwards." The Legislative Scholar 2 (1): 27–34.

@article{cormack2017dcinbox,
  title={DCinbox—Capturing every congressional constituent e-newsletter from 2009 onwards},
  author={Cormack, Lindsey},
  journal={The Legislative Scholar},
  volume={2},
  number={1},
  pages={27--34},
  year={2017}
}