The act of voting is a fundamentally important exercise in all democracies but explaining why individuals cooperate on a large scale when it is individually costly to do so has proven to be a challenging theoretical and empirical puzzle. The limited explanatory power of extant explanations of turnout has prompted political scientists to look at non-traditional predictors of turnout. The recent finding that more than half of the variation in voter turnout could be attributed to genetic factors (Fowler, Baker & Dawes 2008) holds the potential of profoundly changing the way one views voting behavior. However, the specific genes associated with voter turnout or the precise causal pathway linking genes to voting behavior remains unknown. This research project investigates hundreds of thousands of variants across the entire human genome to look for which ones are related to voter turnout. Identifying these variants provides a better understanding of how genes influence voter turnout and will ultimately help to develop and improve theories of political behavior.
The specific questions this research addresses are: 1) which specific genetic variants are associated with voter turnout; and 2) what is the causal pathway linking these genes to voting behavior? In order to answer the first question, the investigator will perform a genome-wide association study of voter turnout. This specifically entails looking for significant associations between each of 592,652 distinct genetic variants spread across the human genome and validated as well as self-reported voter turnout. The analysis will be based on a sample of approximately 3,700 subjects. To answer to the second question, the data set also contains information on several individual attributes known to be highly predictive of turnout such as income, education, personality traits and cognitive ability. Therefore, the investigator formally tests whether specific genetic variants indirectly influence turnout via one of these attributes. Many genetic variants have been found to be associated with traits related to voter turnout giving us an idea of which pathways to investigate. For example, several genetic variants are known to be associated with personality traits and recent research in political science has shown personality traits influence turnout (Mondak, Hibbing, Canache, Seligson & Anderson 2010, Gerber, Huber, Raso & Ha 2008, Gerber, Huber, Doherty & Dowling 2009, Mondak & Halperin 2008). Researching the functional role of genetic variants found to be associated with turnout helps uncover causal pathways.
The discovered link between genes and politics presents the possibility of fundamentally changing how we view voting behavior. However, until one better understands the mechanism linking genes and voting behavior one cannot suggest what types of policies the government could put in place to promote higher turnout or the strategies political actors should adopt in order to mobilize their supporters. This research is necessary to move this nascent research into something useful for policy makers and political actors much in the way genome-wide association studies have aided researchers in better detecting, treating, and preventing diseases.
Political scientists are interested in studying genotypic data to better understand the biological mechanisms that underlie behaviors of interest. The hope is that knowledge of such biological mechanisms will aid in developing or refining theories. Along these lines, previous work demonstrated that voter turnout has a significant genetic basis. Two genetic variants known to play an important role in the serotonin and dopamine neurotransmission systems have also been shown to be associated with voting. However, these two variants are a tiny subset of the likely many associated with turnout. Rather than focus on a handful of candidate genes, genome-wide association studies (GWAS) investigate hundreds of thousands of variants across the entire human genome. Due to advances in technology and constantly falling costs associated with genotyping, researchers are currently conducting genome-wide studies investigating over 1,000,000 variants. This project conducted the results from the first genome-wide association study of voter turnout. We analyzed 510,579 variants genotyped for over 6,500 Minnesota citizens who were eligible to vote in the 2010 midterm election. Our analysis failed to uncover any associations that achieve genome-wide levels of significance after correcting for multiple testing. However, we found several suggestive results that were significant using slightly less stringent criteria. We attempted to replicate several of these associations in two distinct samples from Sweden and California respectively, though none of these associations were significant after correcting for multiple testing in either of these replication samples. While thus far there have been no published GWAS of political behavior, political scientists are beginning to move towards this type of analysis by partnering with researchers in health and medicine who have collected this data for other purposes. In addition, several large national studies will soon release genome-wide marker data to researchers that will enable GWAS of political traits. This project is instructive for future work in that it suggests that the study of complex behaviors that are likely to be influenced by a large number of common genetic variants with small effects will require large samples to detect these effects given the need for multiple-testing correction. Therefore, social scientists should follow the model currently practiced in medical genetics of creating large-scale consortia made up of many different studies in order to achieve the power necessary to detect small effects.
