← Dashboard

Findings and limitations

# Trump mentions in Republican campaign websites, 2026

Generated: 2026-10-15T04:43:41.150395+00:00. See the interactive graph, the complete candidate list, and the printable chart.

## Collection status

The roster contains **487 Republican candidates** and **13 additional candidates with Republican affiliation listed by a source but another current party or independent label.** The latter group is excluded from the Republican summary and labelled in every file. The roster covers House, Senate, and nonvoting delegate elections this fall, including Louisiana’s November House open primary and declared Republican write-ins.

This is a sourced roster, not independent certification of every ballot. John Kitover’s fall write-in candidacy is explicitly unconfirmed and flagged in the candidate file; the other affiliation and website corrections are recorded in the methodology and amendments.

Archive indexes successfully retrieved for **508/508 campaign domains**. Selected **66,697 page-month versions**. Currently **8,134** versions have usable saved HTML and text, **465** failed validation/retrieval, and **58,098** selected versions remain pending.

Usable Republican homepage observations: **1,833/2,922** possible candidate-months. A missing observation is not evidence of zero mentions. Download failures, missing same-month archives, and absent campaign websites are distinguished in the data.

**Collection is incomplete.** Results below describe the observations collected so far. The resumable collection script refreshes the datasets and chart as it progresses. No national percentage should be inferred from this partial sample.

Published collection status: 58,098 selected page versions pending at the dataset timestamp. Local collection may have progressed since publication.

## Homepage comparison

Among **281 Republican candidates with usable homepages in both April and September**, **30** had fewer mentions of Trump, **21** had more, and **230** were unchanged. **58** had a lower mention rate per word. These are comparisons of the selected captures, not necessarily the first or last day of each month.

The trend chart uses the same **83** candidates observed in all six months. This avoids a changing mix of candidates driving the trend. See all endpoint changes and same-page comparisons.

| Month | Usable homepages | Candidates with any collected pages | Same-candidate mean mentions / 1,000 words |
|---|---:|---:|---:|
| 2026-04 | 293 | 318 | 3.296 |
| 2026-05 | 342 | 355 | 3.716 |
| 2026-06 | 319 | 344 | 3.670 |
| 2026-07 | 188 | 208 | 3.707 |
| 2026-08 | 272 | 288 | 3.504 |
| 2026-09 | 419 | 422 | 2.959 |

Largest observed decreases in the paired homepage sample:

| Candidate | Race | April mentions | September mentions | Change | April / 1,000 words | September / 1,000 words |
|---|---|---:|---:|---:|---:|---:|
| Ralph A. Alvarado | KY H06 | 6 | 0 | -6 | 9.934 | 0.000 |
| Gus Michael Bilirakis | FL H12 | 3 | 0 | -3 | 7.264 | 0.000 |
| Brian Mast | FL H21 | 3 | 0 | -3 | 7.481 | 0.000 |
| Joe Mitchell | IA H02 | 3 | 0 | -3 | 3.896 | 0.000 |
| Nicole Malliotakis | NY H11 | 6 | 3 | -3 | 5.055 | 2.523 |
| Alexandra "Alex" Mealer | TX H09 | 3 | 0 | -3 | 4.615 | 0.000 |
| Carlos De La Cruz Jr. | TX H35 | 3 | 0 | -3 | 8.596 | 0.000 |
| Thomas "Tom" McClintock | CA H05 | 2 | 0 | -2 | 9.174 | 0.000 |
| Young Kim | CA H40 | 2 | 0 | -2 | 6.849 | 0.000 |
| Clayton McLean "Clay" Fuller | GA H14 | 12 | 10 | -2 | 7.467 | 6.435 |
| Peter Allen "Pete" Stauber | MN H08 | 2 | 0 | -2 | 2.535 | 0.000 |
| Kurt Alme | MT S2 | 2 | 0 | -2 | 1.571 | 0.000 |
| Russell W. Fry | SC H07 | 4 | 2 | -2 | 8.282 | 5.865 |
| Charles W. "Charlie" Hatcher | TN H05 | 3 | 1 | -2 | 10.949 | 3.356 |
| Eric D. Flores | TX H34 | 2 | 0 | -2 | 4.425 | 0.000 |

Largest observed increases in the same paired homepage sample:

| Candidate | Race | April mentions | September mentions | Change | April / 1,000 words | September / 1,000 words |
|---|---|---:|---:|---:|---:|---:|
| Michael David Whatley | NC S2 | 2 | 13 | +11 | 4.494 | 20.602 |
| Ron Estes | KS H04 | 0 | 10 | +10 | 0.000 | 4.405 |
| Nicholas "Nick" Begich III | AK H00 | 1 | 4 | +3 | 1.647 | 1.372 |
| Thomas B. "Tom" Cotton | AR S2 | 0 | 2 | +2 | 0.000 | 3.373 |
| Christopher Henry "Chris" Smith | NJ H04 | 1 | 3 | +2 | 1.164 | 3.836 |
| Felix Barry "Barry" Moore | AL S2 | 8 | 9 | +1 | 21.918 | 24.324 |
| Dale Whitney Strong | AL H05 | 3 | 4 | +1 | 7.317 | 7.767 |
| Michael S. "Dr. Mike" Katz | DE S2 | 0 | 1 | +1 | 0.000 | 0.686 |
| Ryan Elijah | FL H07 | 0 | 1 | +1 | 0.000 | 0.972 |
| Laurel Lee | FL H15 | 3 | 4 | +1 | 2.078 | 2.597 |
| Sydney Gruters | FL H16 | 5 | 6 | +1 | 15.432 | 9.105 |
| Houston Gaines | GA H10 | 1 | 2 | +1 | 2.732 | 5.319 |
| Erin Houchin | IN H09 | 0 | 1 | +1 | 0.000 | 1.075 |
| Derek Schmidt | KS H02 | 0 | 1 | +1 | 0.000 | 1.174 |
| Julia Letlow | LA S2 | 0 | 1 | +1 | 0.000 | 3.731 |

## Interpretation and limits

A reduction in the count is evidence that the word appeared less often in those two captured homepages. It does not establish a campaign’s intent, whether its support for Trump changed, or whether mentions elsewhere on the website were removed. A news item rotating off the homepage, a site redesign, or altered navigation can also change the count. A lower rate can reflect more non-Trump text with no change in the mention count.

The broader site measure sums the distinct extracted page texts actually collected in each month. It is not a complete historical website reconstruction. The set of archived pages changes by month; missing pages are never counted as zero. Exact duplicate texts are counted once per candidate-month in the site totals. Counts on individual pages remain available in the page-level data.

The original prompt did not include an article link. The likely reference found in search was The Atlantic, “The Republicans Are Bolting From Trump,” September 25, 2026. The full article could not be retrieved through normal fetching or the supplied news proxy. This study measures the archived websites independently; it does not rely on the article’s numerical claims.

## Files

- Candidate roster: identities, offices, ballot labels, websites, source links, cross-check status, and notes.
- Monthly metrics: every candidate × month, including missing observations, raw counts, word counts, exact mentions per word, and mentions per 1,000 words.
- Page metrics: each selected page version and its capture timestamp, result status, file paths, and counts.
- Archive selection manifest: requested pages, chosen captures, and alternatives in the same month.
- Coverage summary: collection coverage and paired comparisons.
- `assets/snapshots/<candidate>/<month>/<snapshot>/`: original HTML, extracted text, Trump context excerpts, and provenance metadata. Compressed source payloads are also retained when decoding was necessary.
- `assets/cdx/`: original archive-index responses and request metadata.
- `assets/sources/`: saved roster pages, the source spreadsheet, cross-check profiles, and article-access results.
- Methodology: inclusion rules, archive selection, text extraction, formulas, and limitations.
- Reproduction instructions.

## Public website

This deployment includes the graph, metric downloads, printable charts, and the exact counted source texts. Raw archived HTML links open at the Internet Archive. Original HTML payloads, archive indexes, retrieval logs, and other source material remain in the local research project. The downloadable CSV file paths refer to that local project; text.txt and page.html evidence paths also work on this website.