At the end of February, Anita Alvarez’s tough year got even worse when her strongest rival in the state’s attorney’s race, Kim Foxx, ran the board on editorial endorsements, picking up nods from the Tribune (which had endorsed Alvarez in her previous two campaigns), the Sun-Times, and the Daily Herald.
Alvarez has been the subject of ongoing, astonishingly persistent protests in the wake of the Laquan McDonald shooting. The endorsements are signs of disenchantment in establishment circles as well, not to mention Foxx’s backing from Toni Preckwinkle and the Cook County Democratic organization, with the long list of prominent politicians that follows with it. (For a good roundup of why that might have happened, read this Eric Zorn column.)
But they’re also … newspaper endorsements. They might matter as a weather vane, but could they actually help Foxx get more votes?
Perhaps surprising in these benighted times for print journalism, there’s evidence in recent studies that they could.
The first comes from a 2008 study by Chun Fang Chiang of National Taiwan University and Brian Knight of Brown University, and their findings make a certain intuitive sense:
Voters are rational and, when evaluating endorsements, attempt to filter out any such bias on the part of the media. The key insight of the model is that, if voters do filter out media bias, then endorsements for the Democratic candidate, say, from a left-leaning newspaper are less credible and should thus have less influence than a similar endorsement from a neutral or a right-leaning source.
Chiang and Knight looked at the endorsements of 166 newspapers in 2000 and 212 in 2004, and estimated how the Democratic vote share would have changed if all of them had endorsed either the Republican or Democratic presidential candidate in those years. Overall, they found the net effect would have been four percentage points. It’s not a ton, but it’s enough to be significant in a close race.
A more recent study from 2012 by Kyle Dropp (then a Ph.D. student at Stanford) and Christopher Warshaw of MIT actually found somewhat different results. Helpfully, it specifically addresses primary voting:
In the primary election experiments, an endorsement from newspapers that are ideologically similar to the voter increases support for the candidate by about 5 percentage points, while an endorsement from newspapers that are ideologically distant from the voter decreases candidate support by approximately 10 percentage points.
Which is a lot; as the authors point out, endorsements may be more influential in a primary “where the spacial positions of candidates appears to be similar.”
What can we conclude? First, that endorsements do actually seem to matter and might be most important in closely contested primary races. Second, the endorsements could genuinely help Foxx, perhaps most of all the Tribune’s. The Tribune editorial board’s centrist-Republican lean might have suggested a preference for Alvarez, so its endorsement of Foxx could, in Chiang and Knight’s model, have more influence.
Dropp and Warshaw’s model suggests something slightly different. The activists who have protested Alvarez are, it is fair to say, “ideologically distant” from the Tribune editorial page. Would that make them less likely to vote for Kim Foxx? Perhaps it’s also fair to apply some common sense in this instance—there’s nothing in the editorial that makes Foxx out to be less appealing from a liberal/progressive standpoint, and it certainly seems unlikely that any taint from the fact of the endorsement alone would dissuade the passionate opposition towards Alvarez.
It will be a fascinating test. Alvarez just squeaked past Tom Allen in her first primary, in 2008, winning by a mere 1.24 percent, but she has carried the incumbent’s advantage ever since in a town “where politics just about always beats policy,” as Roosevelt University’s Paul Green told Chicago. An early February poll put her seven points ahead of Foxx—but with 26 percent undecided. And a seven-point lead is in the ballpark of what newspaper endorsements could close.