Credit: K. LAMARQUE/REUTERS/CORBIS

“Could a seemingly innocuous factor, the type of polling location where people happen to be assigned to vote, actually influence how voters cast their ballots?” Jonah Berger and colleagues asked themselves this question, and went on to answer it with two types of study (J. Berger et al. Proc. Natl Acad. Sci. USA doi:10.1073/pnas.0711988105; 2008).

The first was an analysis of results from a general election held in Arizona in 2000, the ballot for which included a proposition to raise state sales tax from 5.0% to 5.6%, to increase education spending. Polling stations included churches, schools, community centres and government buildings.

Berger et al. predicted that voting in a school would produce more support for the proposition than voting in other places. Indeed it did, but not by much compared with other documented effects on voter choice such as order on the ballot paper. Nonetheless, the effect persisted through tests for various other confounding factors (for example, the possibility of a consistently different level of voter turnout at school polling locations).

The second study was a carefully run online experiment that also involved a proposed tax increase to fund schools. The 'voting environment' was manipulated by exposing participants to typical images of schools or control images. The upshot was the same, with the school images prompting greater (and apparently unconscious) support for the initiative than, for example, an image of an office.

All in all, the authors conclude that what they call contextual priming of polling location affects how people vote. They reasonably wonder whether such factors could, for example, influence voting in a church on such matters as gay marriage and stem-cell research.

But here's a thought. In the event of science spending being on the political agenda, why not offer the lab as a polling station? But maybe dim that fluorescent lighting, and persuade all those bearded fellows in white coats to take the day off — or not, as the case may be.