The Times Launches 30,000-Voter Election Model After Exit Polls Misled in 2016
Updated
Updated · The New York Times · Nov 1
The Times Launches 30,000-Voter Election Model After Exit Polls Misled in 2016
1 articles · Updated · The New York Times · Nov 1
Summary
The Times said it will publish “Times Voter Trends,” a new demographic breakdown of every U.S. House and Senate race by age, race, education and gender in all 50 states.
About 30,000 people interviewed this year for Times/Siena polls will be recontacted to verify whether they voted and for whom, then matched with precinct- and county-level results and The Needle’s election-night estimates.
The project is meant to replace reliance on exit polls, which The Times said have repeatedly produced misleading election-night narratives, including faulty 2016 signals that showed Hillary Clinton ahead and overstated Trump’s support among white women.
Preliminary findings will be released late on election night and updated as more data arrives; The Times has already published retroactive estimates for the 2024 and 2020 elections.
As media outlets abandon exit polls for complex data models, will this quest for accuracy sacrifice the speed of election night reporting?
Could relying on pre-election survey respondents to predict election trends inadvertently amplify the very polling errors journalists are trying to escape?
With early voting changing the landscape, can a hybrid model of recontacted surveys truly eliminate the hidden biases of modern polling?