A new trust survey and a look at how AP, Decision Desk HQ and the networks call races show where credibility is won or lost on a close election night.
WASHINGTON, D.C. (Journo News) – Days before the first polls closed in the United States midterm elections, only about one in three Americans said they had a high level of confidence that news organizations would report the results accurately. The finding, from a national survey, frames a hard question for every newsroom that covers vote counts: how do you earn trust from an audience that has already decided to doubt you?
What the survey found
Nieman Lab’s summary of the poll says it was run by NORC at the University of Chicago for the Medill Local News Initiative and the Alliance for Trust in Media, among 1,057 nationally representative adults surveyed online or by phone in October. Roughly 60 percent had at least some confidence that outlets would report the results accurately, while 32 percent had quite a bit or a great deal. Another 37 percent said they had not too much confidence or none. The Medill team’s own midterm trust tracker page is where the full results are published.
- About half of respondents had little or no confidence that outlets would thoroughly investigate claims of voter fraud.
- Slightly more people said journalists manipulate news for political or commercial reasons than said they report it accurately.
- Nearly 80 percent of Republicans said journalists try to manipulate the news, against 38 percent of independents and 24 percent of Democrats.
Nieman Lab pairs the survey with a Gallup finding that 33 percent of Americans trust the media to report fully, accurately and fairly, up from an all-time low the previous year, and with a New York Times analysis that Donald Trump has cast doubt on elections on 70 percent of the days since returning to office. The Gallup trust results show a small recovery, not a rebuilt relationship.
How the races actually get called
Against that backdrop, the mechanics of an election night matter more than they usually would. A Reuters explainer, carried by CP24, set out the plan. The Associated Press, which has counted votes since 1848, said it would deploy more than 4,000 reporters at county election offices and polling places, and it supplies the decision desks at ABC, CBS, Fox, CNN and NBC. It calls a race only when the trailing candidate has no realistic path to catch up.
Decision Desk HQ takes a different route. About 80 percent of its vote data comes from official sites, collected directly or through manual monitoring, with reporters filling the rest. A projection requires unanimous agreement on its team and sign-off from its head of data science. Its clients include NewsNation, The Hill and Newsmax, and the social platform X recently signed up. Meanwhile, ABC, CBS, CNN and NBC also draw on SSRS, which bought the exit-polling firm Edison Research in 2025.
One election, several different standards
The explainer also shows that newsrooms are not following one rulebook. The Wall Street Journal and MS NOW rely only on AP calls, and Bloomberg makes none of its own. Fox News and CBS News run their own projections using the AP count, exit polls, early-vote totals and internal models. The New York Times uses AP calls for most contests but examines close races with precinct-level data and may call before or after AP. Reuters said it would cite calls made by other organisations.
For readers this is bewildering. For journalists it is a risk: when two trusted brands differ on whether a race is decided, the audience is likely to read the difference as bias, not method. Our view is that every outlet should say in plain words on its results page which source or model it uses and what would make it withdraw a call.
The record of retractions
The same explainer lists examples that newsrooms should keep in mind. Decision Desk HQ withdrew a 2025 projection in the Seattle mayor’s race and a call in a New Jersey 11th District special primary after the counts shifted. In 2018, AP retracted a call in California’s 21st District after the projected winner fell behind. Each of those withdrawals is the system working as designed, but each also hands critics a concrete example, and in a climate where roughly half of the public doubts newsrooms will investigate fraud claims properly, the speed and transparency of a correction matter as much as the call.
What this means for newsrooms
- Publish your call methodology before election night, in language a reader can follow, and link it from every results page.
- Say clearly when a race is too close to call, and explain what outstanding ballots could change the outcome.
- Treat the partisan gap as a design problem. If one side of the audience believes journalists manipulate news, the answer is evidence, sourcing and visible corrections, not a louder claim of neutrality.
- Prepare a standing response for fraud claims: who investigates, what records are checked, and when you will report back.
A caution on scope: the sources used here were published before results came in, so this piece assesses the preparation and the trust climate, not how individual outlets performed on the night. That judgment belongs in a follow-up once the counts are settled and the corrections, if any, are on the record.
The wider lesson
Trust in election reporting is not decided on election night. It is built or lost in the months before, through how openly newsrooms describe their methods, and in the days after, through how they handle mistakes. The polling suggests plenty of Americans are still persuadable, since most said they had at least some confidence. For those who need a break from the vote count, Bohiney Magazine and The London Prat offer the comedy side of the political media circus at Bohiney Magazine.
SOURCE: https://bohiney.com/