In times of economic and political turmoil, people increasingly turn their attention to how their societies are governed. As a result, it seems as though matters relating to elections and electoral processes are permanently in the public spotlight. Globally, over 80 national elections or civil ballots are scheduled to take place in 2011; this month alone, there are to be seven national elections or referenda (in Benin, Cape Verde, Chad, Ireland, Kosovo, Switzerland, and Uganda). (The year’s busiest month will be April, during which eleven separate national ballots are already scheduled, with a constitutional referendum also now likely to be held in Egypt.)
Perhaps surprisingly, the practice of democracy has been influenced in several respects by systematically conducted empirical research. For example, much of the text formatting and layout of ballot papers is subject to evidence-based design principles. Research showing that multi-page ballots lead to higher levels of voter error has led to ballot papers being produced so that all candidates appear on the same page (increasing its size if necessary). Other research has led to the introduction of graphical elements, such as candidate photographs and political party emblems. Overall, studies spanning several disciplines (such as psychology, optometry, and ergonomics) have helped influence what goes on inside the polling station. However, research has now begun to show that the polling station itself is of relevance, and in a particular way that will undoubtedly be seen as controversial.
In many countries, voting takes place in approved venues provided by public or voluntary bodies. Examples include schools, community centres, police stations, and churches. Research suggests such venues include several subtle cues that draw attention to electoral issues and that prime voters to cast their ballot in particular ways. One concerning example is the presence of religious or ecclesiastical décor, as might be expected if a polling centre is located in a church or a religiously-oriented school (the case for the majority of stations in Ireland, where a national parliamentary election takes place this Friday, 25 February).
Based on data from elections in South Carolina, researchers at California State University have established that votes cast in religious buildings are more skewed in favour of conservative electoral candidates than votes cast in secular locations. The effect is quite large: while 32% of secular location votes went to conservative candidates in the 2004 gubernatorial election, 41% of votes cast in churches did so. Due to the standardized way in which voters are assigned to polling stations in South Carolina, there is no particular basis to believe that conservative voters are any more likely to end up voting in churches. Further, the analysis covered over 400 separate stations, implying that the dataset used was statistically robust.
The researchers followed this up by examining the results of a different vote (again in South Carolina), where voters took part in a 2006 referendum to amend the state constitution. Once again, there was a significant conservative skew in church-located voting, as judged by votes cast in favour of an amendment that introduced a religiously influenced definition of marriage. On this occasion the difference appeared small (83% vs. 81.5%), but as it was based on an overall high rate of votes cast in favour (leading to what statisticians call a “ceiling effect”) and as the sample of polling stations studied was very high (covering over 1,400 stations), the findings were highly statistically significant.
While offering food for thought, such studies can be difficult to interpret or generalize. For example, voters in South Carolina may have a baseline level of religiosity that is easily provoked by subtle visual cues at election time. In addition, South Carolina elections may be religiously charged in particular ways that are not representative of other places. However, wherever they are held, elections during recessions often produce choices between conservatism and liberalism. Furthermore, across the world, many lobby groups explicitly attempt to link electoral choices to religious agendas in order to advocate for particular political outcomes (see, for example, the way the Irish Labour Party has been characterized as a pro-abortion movement). And in most countries, the majority of the electorate usually professes (to census takers) to be, at least nominally, religious.
During elections, the electoral process itself is given much attention. As well as the evidence on ballot design, research from behavioural sciences may also be relevant. For example, when considering possible reforms, it may be wise to reflect further about where—as well as how—elections are held, and to present scientifically conducted research as the basis for modifications.