Are there comprehensive datasets that track corrections and retractions across NYT, WaPo, WSJ, CNN, Fox and BBC over the last decade?
Executive summary
There is no evidence in the provided reporting of a single, comprehensive dataset that catalogs corrections and retractions across The New York Times, The Washington Post, The Wall Street Journal, CNN, Fox News and the BBC for the last decade; instead, the record shows a patchwork of outlet-specific correction pages, academic studies of limited scope, and watchdog or trade analyses that collect high‑profile examples rather than a unified cross‑platform archive [1] [2] [3] [4].
1. What the evidence actually documents: pockets, not a pan-out database
Researchers and watchdogs have produced focused compilations—an academic cross-sectional analysis examined COVID‑19 corrections recorded on the NYT’s corrections page, but it covers a specific topic and outlet rather than all major news organizations over a decade [1], Poynter has periodically compiled “top corrections” lists and annual roundups that survey notable media corrections rather than maintain a continuous, multi‑outlet database [2], and advocacy organizations keep curated lists of corrections for particular outlets but do so with explicit issue priorities (for example CAMERA’s tracking of NYT corrections) [4].
2. Comparative accounts exist, but they’re descriptive case studies
Trade and policy observers have written comparative narratives about how networks handled prominent retractions—Benton Institute’s “tale of two networks” contrasts Fox and CNN in a small set of high‑visibility cases and the institutional responses, but it is a journalistic or policy analysis, not a structured dataset designed for longitudinal cross‑outlet measurement [3]; similarly, Reuters and other outlets have historically reported on aggregate correction counts for individual papers as context, demonstrating that papers differ in correction volume and transparency practices [5].
3. Academic and replication data touch the subject but don’t fill the gap
Scholarly work explores the effects and dynamics of corrections and retractions—experimental work on how retractions affect public belief comes with replication data useful for methodological study [6]—yet these datasets are framed to test hypotheses (e.g., effects of retractions on belief) and typically draw on sampled cases or simulated feeds rather than serving as a comprehensive ledger of every correction issued by the six named outlets across a decade [6].
4. Public lists and Wikipedia chronologies capture controversies but not standardized metadata
Open compilations and encyclopedia entries chronicle controversies and notable retractions over time (examples include Wikipedia pages that list outlet controversies for NYT and CNN), but those lists are episodic, vary in sourcing and editorial standards, and do not provide consistent, machine‑readable metadata (dates, full text of original claims, correction text, why corrected) that a true cross‑outlet dataset would require [7] [8].
5. Reasons a consolidated dataset is missing in the record—and where biases live
A unified dataset would require standardized definitions (what counts as a “correction” vs. “clarification” vs. “retraction”), consistent outlet disclosure practices, and long‑term archival access; existing reporting shows outlets and observers handle corrections and retractions unevenly—some publish formal correction pages, some bury updates, and advocacy trackers may selectively publicize items that align with their interests—so aggregation without clear, transparent methodology risks reproducing bias or omitting cases [5] [4] [3].
6. Bottom line and immediate next steps for a researcher
Based on the supplied sources, no single, public, decade‑spanning dataset covering corrections and retractions across NYT, WaPo, WSJ, CNN, Fox and BBC has been identified; the alternative path is to combine outlet correction pages, watchdog compilations, and targeted academic datasets into a custom database while documenting inclusion rules and known blind spots—the sources here document the pieces but not the finished puzzle [1] [2] [6] [3] [4].