It's shaping up to be a rough Senate cycle for Democrats, with forecasters giving Republicans a 68 percent chance of taking over. But with only a week to go, Democrats do appear to have one thing going for them: newspaper endorsements.
The pattern isn't universal (the Denver Post endorsed Republican Cory Gardner for Colorado's Senate seat), but two struggling Democrats have been lent a hand by the local press in recent weeks. Rep. Bruce Braley (D-IA), who current forecasts suggest has about a 1-in-3 chance of winning, was endorsed by the Des Moines Register, while Kentucky Secretary of State Alison Lundergan Grimes, who's even further behind, was backed by both the Courier-Journal in Louisville and the Lexington Herald-Leader over Senate Republican leader Mitch McConnell.
The endorsements may not be enough to put Braley and Grimes over the top, but the political science literature suggests they'll likely matter a little bit. GWU's Danny Hayes has a great rundown over at the Monkey Cage.
Bush vs. Gore and Bush vs. Kerry
In one study, Brown's Brian Knight and National Taiwan University's Chun-Fang Chiang matched up daily tracking polls for the 2000 and 2004 elections with data on when newspapers issued endorsements, and on the ideological leanings of those newspapers. The tracking polls also asked which newspapers respondents read, which let Knight and Chiang see how peoples' views changed after their newspapers issued endorsements. The study is correlational, so it's possible that other factors changed respondents' minds, but Knight and Chiang argue that the most obvious alternative explanations don't check out, implying that their estimates really reflect the effect of endorsements.
They estimate that if, in the 2000 and 2004 elections, every newspaper had endorsed George W. Bush, Democrats' vote share among newspaper readers (about three-quarters of voters, they estimate) would have been 2.6 to 3.1 percent lower. That's not a whole lot, and the change they're modeling (literally every newspaper backing Bush) is beyond improbable, but in an election as close as 2000, even a tiny effect could have mattered.
The researchers also found that the results were bigger when endorsements were surprising, given newspapers' usual slants. In 2004, they estimate, the Chicago Sun Times' endorsement of John Kerry and the Denver Post's endorsement of Bush encouraged 3 percent of each newspaper's readers to switch, whereas barely anyone was swayed by the New York Times' endorsement of Kerry.
In 2012, Dartmouth's Kyle Dropp and MIT's Chris Warshaw conducted a survey experiment where 5,500 respondents were asked for views on a hypothetical congressional race, either in the general election or primary. They were shown short descriptions of the two fake candidates, and asked which they'd be more inclined to support. But some respondents were also informed that their local newspaper had endorsed one or the other candidate.
"In the general election experiments, an endorsement from newspapers that are ideologically similar to the voter increases support for the candidate by approximately 5 points," Dropp and Warshaw concluded. In the primary election experiments, an endorsement from an ideologically similar paper increased support by about 5 points as well, but an endorsement from an ideologically dissimilar paper reduced support by about 10 points.
Additionally, Arizona State's Kim Fridkin Kahn and Patrick Kenney found substantial endorsement effects in the 1988, 1990, and 1992 Senate races, while Georgetown's Jonathan Ladd and Berkeley's Gabriel Lenz found that British newspapers' switch to Labour in 1997 played a role in Tony Blair's landslide victory that year.
As Hayes notes, endorsement effects will inevitably change, and perhaps diminish, as the print newspapers that issue them continue to lose subscribers, so we shouldn't take these effects as laws of nature. But for the time being, it seems plausible that Braley and Grimes will get a small bump from their local papers' backing.