Living investigation — last refreshed 2026-07-20: Fox News’s first-party inventory is established through the current crawl frontier, and the article-level evidence crawl is in progress. Results are provisional and update as verified records enter the completed span.
Operational rule: A name-matched record explicitly names by surname the person who was the sitting U.S. president when the article was published. This is a sitting-president name-match corpus, not every article “about” a president. Semantic aboutness is broader and requires separate analysis.
Current database snapshot
Obama II
3,451 total name-matched URL/headline records currently stored
85 records in the continuously completed date-span graph sample
3,484 of 439,202 inventoried pages inspected — 0.79%
Frontier: 2015-05-07
Negative 40.9% · Neutral 36.2% · Positive 22.9% · Score -9.4
Trump I
28 total name-matched URL/headline records currently stored
15 records in the continuously completed date-span graph sample
58 of 358,390 inventoried pages inspected — 0.02%
Frontier: 2017-01-21
Negative 14.3% · Neutral 53.6% · Positive 32.1% · Score +10.3
Biden
39 total name-matched URL/headline records currently stored
35 records in the continuously completed date-span graph sample
81 of 264,228 inventoried pages inspected — 0.03%
Frontier: 2021-01-21
Negative 25.6% · Neutral 48.7% · Positive 25.6% · Score +1.9
Trump II
31 total name-matched URL/headline records currently stored
31 records in the continuously completed date-span graph sample
31 of 73,859 inventoried pages inspected — 0.04%
Frontier: 2025-01-21
Negative 16.1% · Neutral 48.4% · Positive 35.5% · Score +13.6
Combined name-matched URL/headline records: 3,549. Inventory status: {"blocked": 16, "complete": 147}. Article retrieval errors: 0.
Headline wording by administration
| Administration | n | Negative | Neutral | Positive | Score | 95% CI |
|---|---|---|---|---|---|---|
| Obama II | 3451 | 40.9% | 36.2% | 22.9% | -9.4 | -10.6 to -8.2 |
| Trump I | 28 | 14.3% | 53.6% | 32.1% | +10.3 | -1.3 to +22.0 |
| Biden | 39 | 25.6% | 48.7% | 25.6% | +1.9 | -8.1 to +11.9 |
| Trump II | 31 | 16.1% | 48.4% | 35.5% | +13.6 | -0.5 to +27.7 |
Four-year term-aligned tone graphs
Lines begin only when the continuously completed sample reaches the opening quarter of that presidential term. The Obama II series is withheld because the verified first-party inventory currently begins too late to draw a truthful Year 0 line.
Negative headline share
Neutral headline share
Positive headline share
Overall headline tone score (−100 to +100)
Monthly and quarterly trends
Coverage volume
News, opinion, video, transcript and production-origin comparisons
| Administration | Type | Editorial class | Origin | n | Score |
|---|---|---|---|---|---|
| Biden | article | news | fox-produced | 38 | +1.9 |
| Biden | article | opinion | fox-produced | 1 | +0.0 |
| Obama II | article | news | fox-produced | 2669 | -8.1 |
| Obama II | article | opinion | fox-produced | 782 | -13.7 |
| Trump I | article | news | fox-produced | 25 | +10.8 |
| Trump I | article | opinion | fox-produced | 3 | +6.4 |
| Trump II | article | news | fox-produced | 30 | +11.7 |
| Trump II | article | opinion | fox-produced | 1 | +71.8 |
Provisional author lexical screening
Only author-administration groups with at least 30 qualifying headlines will appear. This screening cannot establish author bias.
Methodology and limitations
The primary inventory uses Fox News’s first-party HTML article sitemap, preserved with retrieval metadata and hashes. The separately published video sitemap and rolling RSS feeds are distinct discovery sources. The exposed article index begins at 2014, while observed 2014 day pages have so far yielded no article links; the 2013 portion of Obama’s second term remains an explicit source gap.
Publication timestamps—not URL dates or worker partitions—determine the administration, using noon Eastern boundaries on January 20, 2013, 2017, 2021 and 2025. Canonical aliases, redirects, identical content, identical headline text, updated versions, video/transcript versions and syndicated or wire copies are tracked separately. URL/headline records are not called unique headlines unless headline text is deduplicated.
When available, VADER compound scores classify wording as negative at ≤ −0.05, neutral between −0.05 and +0.05, and positive at ≥ +0.05. The reader score is compound × 100. Confidence intervals will describe uncertainty in mean lexical score. VADER measures wording—not factual accuracy, fairness, evidentiary support, newsworthiness, ideological viewpoint or political bias.
Every graph card reports both the total stored name-matched record count and the smaller continuously completed-date-span graph sample. Partial periods and crawl frontiers remain labeled. Straight news, opinion, contributor commentary, program recaps, video, transcripts, live blogs, wire reports and syndicated material will not be silently combined.
Evidence manifest SHA-256: b8fcfd3aefeb81bf6a33666b760abe9ef87fd0c0f9fe3dfcda3f24eef3fe0745
Snapshot SHA-256: 682d52fb1e6cc7dd8928b8a66eea64e40897107a5f4d30179847c93394f07dab