It’s all about location. But Panagopoulos’ street signs were pushing a nonpartisan civic message, not a candidate-specific one. In 2008, Ohio State’s Todd Makse (now at Dickinson) and the University of Colorado’s Anand Edward Sokhey set out to see if they could identify a link between candidate signs and support. They went to Franklin County, a celebrated Ohio swing county that includes Columbus, and drove the streets of 30 precincts on two weekends, one just after the conventions and one just before the election. Each time they saw a candidate sign displayed at a single-family home, the academics marked it with a GIS locator and then linked each address to its residents’ voting records and sent a survey to each household, asking the resident “most responsible” for the sign to complete it.
In 2008, 14 percent of homes displayed a sign for a candidate, and 65 percent of them were for Democrats. A few traits were consistent across party lines: Young people, generally less politically active than older folks, were more likely to display a sign. (In the survey, young people indicated that signs were one of the forms of activism with which they were most comfortable.) The act of putting a sign in a window seems as contagious as leaving one of its panes broken: Those whose neighbors had signs were more likely to have one themselves, regardless of whether they supported the same party or different ones. Those whose houses were exposed to greater levels of car traffic—Makse and Sokhey coded every thoroughfare in their precincts into one of six categories, from “dead end” and “main artery”—were more likely to put out signs, suggesting that the motivation to put one out was more “strategic” (to reach as many people as possible) than “expressive” (a need to be public about one’s allegiances).
Voters decorate their lawns. In 2006, Auburn agricultural economist David N. Laband, who had previously conducted studies showing that people who wore campaign buttons and checked the Federal Election Commission contribution box on their tax returns were more likely to vote, started looking at signs. Along with colleagues Ram Pandit and John P. Sophocleus, they mapped Auburn, Ala., neighborhoods south of Interstate 85 and marked which houses displayed an American flag on Memorial Day and July 4. Once football season kicked off in the fall, they documented expressions of support for the Auburn Tigers with a flag, sign, pom-pom sticker, or an inflated figure of Aubie, the school’s mascot. Just before the elections they went back to the same neighborhood with an eye out for candidate signs. Seventeen percent of houses had a flag, 7 percent had football paraphernalia, and 12 percent had a political sign.
After the November 2006 elections, Laband matched the addresses up against the voting histories of the people who lived there. Households that displayed either an American flag, football insignia, or campaign sign were 2.4 times more likely to have a resident who voted in the elections than houses which had none of the three. While campaign signs were the most strongly predictive of having cast a vote, just sporting an American flag made a household twice as likely to have a voter, and even Auburn football gear made it 1.6 times more likely. (The authors don’t account for whether they’re really measuring a mediating variable, like whether Auburn fanhood is a proxy for having a college degree.) The same sense of expressiveness that inspired people to publicly project their patriotism or fandom seemed to be driving them to vote.
Signs do not necessarily translate into votes. But just because some people are the expressive kind doesn’t mean they’re always in a mood to express themselves. In 2010, before the midterm elections, Makse and Sokhey returned to Franklin County and found only 6 percent of homes had signs, down from the 12 percent they’d observed in 2006. The share touting Democrats had fallen to 57 percent. But the drop in signs did not translate into a steep drop in votes. The authors describe one of their precincts, 1028F on Columbus’s east side, as a “racially diverse, middle-class neighborhood”; more than three-quarters of its votes went to Obama in the 2008 general election, and 19 percent of its households had a sign for Obama or another Democrat. When the authors returned in 2010, shifts in the political climate were evident in local yards. Only 3 percent of homes had a Democrat’s sign this time, one of the lowest shares among 30 thirty areas that Makse and Sokhey tracked. But the party’s votes didn’t fall dramatically. The precinct gave nearly the same share of the vote to the Democrat at the top of its ticket in 2010 (Gov. Ted Strickland) as it did to Obama two years earlier— and experienced less of a turnout drop-off than other parts of the state.
Why was there a dip in sign display, but not in actual support? Makse and Sokhey couldn’t clearly attribute causality, so it remains unclear whether residents were less interested in displaying a sign in 2010 or just that Democratic campaigns that year (with less funding and fewer volunteers) weren’t as capable of producing signs and getting them to supporters. But their findings should give pause to those who spend primary day seeking clues to the outcome by counting logos in the highway median. Indeed what Shaheen observed on her drive from the Seacoast to Manchester may have been less a leading indicator of grass-roots enthusiasm than a lagging indicator of the climate. As New Hampshire Public Radio’s Josh Rogers noted, the winter’s mild weather may be have its own chilling effect on sign displays: It’s harder to drive their wooden stakes into the bare ground than into bountiful snowdrifts.