Research on the last two contests finds that, under specific circumstances, endorsements can significantly increase a candidate’s chances of winning.
By Tom Jacobs
(Photo: Brendan Smialowski/AFP/Getty Images)
The Cincinnati Enquirer has now joined the Dallas Morning Newsin breaking with its tradition of supporting Republican candidates for president and instead backing Hillary Clinton. Given Ohio’s status as a swing state, the key question becomes: Do such surprise endorsements make a difference?
A study published earlier this year, which analyzed the 2008 and 2012 elections, suggests the answer is yes — so long as they don’t conflict with the tone of the paper’s news coverage.
“Not all unexpected endorsements have the same effect,” a research team led by Northwestern University economist Agustin Casas writes in the journal Economic Inquiry. “Endorsements which are consistent with respect to the newspaper’s discourse, and which come as a surprise compared to the newspaper’s endorsement history, have a large and potentially decisive effect in tied contests.”
Casas and his colleagues Yarine Fawaz and Andre Trindade compared the dates of newspaper endorsements during the last two presidential elections with data from the Intrade prediction market — specifically, “the daily trading prices of the contract ‘Obama to win the election.’”
They found endorsements do indeed improve a candidate’s odds of winning, especially if the race is close. In a state where the contest was tied, for example, “a Democratic endorsement from a large newspaper increases Obama’s probability (of winning) from 50 percent to 58 percent.”
To determine the specific elements that make an endorsement particularly influential, the researchers examined which candidates the newspaper has backed in the past, noting any abrupt shifts from one party to the other.
“Endorsements which are consistent with respect to the newspaper’s discourse, and which come as a surprise compared to the newspaper’s endorsement history, have a large and potentially decisive effect in tied contests.”
They also used data from an earlier study that rated the overall ideological tone of each paper’s news coverage. This was done by calculating the number of times the publication employed such Republican-friendly phrases as “death tax,” as opposed to descriptors favored by Democrats such as “estate tax.”
The researchers found “inconsistent endorsements” — those that are at odds with the rhetorical style of the news columns — have no measurable effect. Such statements of support “are unexpected,” the researchers write, “but may be more confusing than informative for their readers.”
In contrast, the endorsements that have the greatest impact (if the trading markets are a reliable indicator) are those that are consistent with the language used on the news pages, but surprising in that they represent a shift from the endorsements of past years.
So if the news columns of a paper reflect — however subtly — a liberal bias, and the traditionally Republican editorial page endorses a Democrat, the paper is, in effect, speaking with one voice.
This research suggests that internal consistency — along with the jolt that breaking with precedent produces—can have a real effect on how a newspaper’s readers vote.