Would people believe a news story is accurate if it were published by a source that shares their worldview? Or would they believe claims that agree with their views, regardless of where they were published?

A 2018 survey by the Gallup and Knight Foundation concluded that Americans perceive news articles as biased (62% of news stories) or inaccurate (44% of stories) depending on whether they believe the news outlet shares their political affiliation.

But Associate Professor Mor Naaman and colleagues at the Jacobs Technion–Cornell Institute at Cornell Tech suggest that this survey and other recent studies on the phenomenon did not dig deep enough to find out whether individuals might be influenced by the political nature of the claims published in the news stories themselves.

A new report by Asst. Prof. Naaman and his colleagues shows that Americans are actually more likely to believe that a news story is accurate if the headline aligns with their political views — and that it does not matter whether the headline comes from a source that aligns with the reader’s views.

For instance, a left-leaning reader who sees the headline “Trump Lashes Out at Vanity Fair, One Day After It Lambastes His Restaurant” is more likely than a right-leaning person to believe the headline is true. For both these readers, it doesn’t matter whether the headline appears on Fox News or the New York Times, the researchers discovered.

The researchers conducted an online experiment with 400 Americans through Amazon’s Mechanical Turk, a crowdsourcing platform for surveys and other tasks.

The participants were shown a set of left-leaning, right-leaning, and nonpolitical headlines assigned randomly to either Fox News or the New York Times, and asked to evaluate whether the headlines were true or false. The researchers then asked the participants a series of questions to determine their political affiliation.

They found that participants overwhelmingly reported believing headlines that align with their political views, regardless of the source of the report. They also found strong evidence that those participating in the experiment were not always truthful about how they evaluated the headlines. For example, right-leaning readers would often say a left-leaning headline was false, even when they believed it was true.

When some of the participants were offered a small payment for “correctly” answering whether the headlines were true or false, they were less likely to respond in ways that aligned with their political leanings. The researchers found that right-leaning participants in particular rated more of the left-leaning headlines as true when they were offered the payment option.

The results provide insight and nuance into the question of trust in news. The findings suggest that the source of news may be less polarizing than previously thought. However, the experiment shows that people are likely to reject disagreeable information, even if they trust its source.

While the findings are preliminary, the researchers look forward to expanding their research with more headlines, more news sources, and a larger and more diverse group of participants.