Categories
External Events Politics

The untold side of voting behaviour

How is voting behaviour influenced by cognitive biases? In which ways does social interaction affect the beliefs of individuals? And what are the psychological determinants of support for either a liberal or conservative ideology? The event held on April 7, 2021 answered these and more questions, offering us some insight on the behavioural and psychological underpinnings of political preferences of voters.  

With Bocconi Professor Catherine De Vries as our moderator, we invited as our speakers Dan Braha, Full Professor at University of Massachusetts, and Yoel Inbar, Associate Professor of Psychology at the University of Toronto. Professor Braha has authored numerous articles and published three books aimed at inquiring the functioning of large-scale human-made systems, with a particular focus on dynamic social networks. Professor Inbar’s research is instead focused on the impact that psychological determinants have on social and moral beliefs. Most interestingly, the speakers offered us their insight from two different perspectives and methodologies, as Braha’s research is underpinned by a macro sociological approach, while Inbar’s studies are based on a more focused and psychological perspective.  

Voting contagion

Braha’s intervention focused on social contagion, introducing the role played by social influence on human behaviours and decisions. He explained how sources of social influence can be either external from our social circle (as determined by broader social factors) or internal (hence originating from the personal connections individuals entertain with family members and acquaintances). How these different sources of influence affect behaviour are linked to the role that human made networks play in the diffusion of information in complex systems. Braha explained that all networks share three commonalities: universalityout of equilibrium dynamics, and phase transitions. All these elements can be applied to political science analysis, allowing predictions of the behaviour and evolution of networks of voters.  

There is evidence that individuals in networks tend to imitate the choices of other agents because they trust the information on which the majority bases its decisions, even if this goes against evidence of one’s own private information. This behaviour of following the crowd can be explained in terms of social pressure to conform, determined by both a threshold and a network effect. The threshold effect refers to the fact that when a certain amount of people adopts a certain behaviour, other individuals experience social pressure to conform. The network effect is linked instead to the increased value which is attributed to a widespread behaviour in virtue of its diffusion. Connecting people in a network also allows them to influence each other’s behaviour and opinions in a dynamic way. The system can either stabilize into consensus or in a situation with a mix of individuals having different preferences. The patterns of social epidemics in networks can help in understanding a variety of social phenomena, including the affirmation of ideas and social norms, as well as the emergence of social unrest.  

Applying network dynamics to the US political context allows, for instance, to analyse the extent of the polarization of the political debate between Democrats and Republicans on Twitter and in the blogosphere, with the former being the most dichotomous of the two, presenting two different separate clusters of users based on political affiliation.  

What about voting?

The basic idea of Braha’s research is to establish a quantitative model able to predict voting behaviour in a network in which people’s decisions are determined by two forces: consistent partisan biases (prejudiced attitudes, party identification, religious background, and explosion to the mass media) and peer influence (friends, neighbours, and family). By looking at US data it is possible to characterize the probability distribution of voting behaviour for either the Democratic or the Republican Party in steady state. We can then change one of the two parameters to understands how the voting distribution is affected. The main finding is that, all things being equal, decreasing peer influence leads to a polarization of the system, while having both consistent partisan bias and peer influence maintains the system voting probability on a normal distribution for both Democrats and Republicans. 

How can peer influence effect be estimated empirically?  

In the US, Braha did so by focusing on the different counties within states. While it can be assumed that each state is exposed to the same consistent partisan biases, each county is instead affected by different degrees of peer influence, as determined by local socio-cultural structures. Hence, peer influence can be derived by observing how networks in different counties behave and evolve, while controlling for state level biases. Overall, the voter distribution at the level of the state is predicted to be gaussian (or normal), a result confirmed by empirical data. Repeating the estimation for all US states allows to derive a Social Influence Index which can be used analyse different electoral outcomes.

By looking at the evolution of the Social Influence Index over time, it is possible to see that, in the US, social influence became relevant for voting outcomes only after the landmark year 1984, an evolution defined as the “Herding Phase Transition”. This increased relevance in the network effect of social behaviour has been linked by Braha to the historical increase in US population density, which naturally affects the speed of propagation of social ideas. Indeed, there is a close link between population density and social influence, with the highest population density in the more populated US coasts, and lower in in the middle states. Hence, these less populated states have experienced a much slower circulation of liberal ideas, instead being bound to conservative values because of a lower degree of social propagation. Still today, the high correlation between conservative attitudes and low social influence remains in the central regions of the US, where communities purse familiarity and are more resistant to change. Moreover, there are geographical effects to social influence, as it is both spatially autocorrelated and clustered, with social influence in a state influencing the degree of social influence in nearby states. 

By characterizing each US election year as a vector of values of the Social Influence Index across states, different election years can be correlated in terms of high or low impact of social influence on electoral outcomes. Interestingly, Braha found that different years in which social influence was high correspond to important events in US political history: the shift to the Fourth Party System in the years 1896 to 1932, the Wall Street Crash of 1929, the realignment of the Fifth Party System to a Democratic majority with the 1932 election and the New Deal coalition, the Watergate scandal and Nixon’s resignation and, lastly, the current Sixth Party System, characterized by electoral dealignment, weakening of party loyalties, and the critical role of voters’ personal and social networks.  

In concluding the presentation of his research, Professor Braha underlined the possibility of applying this framework of study to other countries other than the US and stressed the need to evolve from “discrete choice” theory to a “network discrete choice” theory for explaining how individuals make decisions in the social and political domains, from the creation of investment portfolios to participation in grassroot movements.  

Disgust and politics

Professor Inbar presented his research on the influence of the psychological perception of disgust on conservative political attitudes of individuals. As defined by Miller (1998), disgust is “a sense of aversion to something perceived as dangerous because of its powers to contaminate, infect, or pollute by proximity, contact, or ingestion”. Inbar proceeded to explain that disgust is considered as a basic emotion which is found universally in all societies and is always triggered by some by common ‘core’ elicitors (as, for instance, rotten food), as disgust is an evolutionary response meant to reduce our contamination with pathogens which can possibility make us sick.

How is the emotion of disgust related to political preferences?  

Haidt, McCauley, and Rozin (1994) have constructed a “disgust scale”, which can be built by considering the answers of individuals to certain psychometric questions. Linking disgust and politics, on average, US voters who score higher in the disgust scale are also those who are characterized by more conservative political preferences. Inbar, Pizarro, and Bloom (2009) have found evidence of disgust sensitivity being correlated with conservatism in a sample of undergraduates and adults in the US, while Inbar, Pizarro, Iyer, and Haidt (2012) have confirmed the same findings for a larger sample also involving international respondents, controlling for different demographic characteristics as age, gender, religious affiliation, education, and ethnicity. Conservative ideology is then found to be a good predictor of conservative voting behaviour. Indeed, a study of the 2008 election showed that states with an average higher disgust sensitivity were more likely to prefer McCain instead of Obama.

What determines this link between disgust and conservative attitudes?

One explanation is offered by the “behavioural immune system” hypothesis, which states that disgust is intended to be a response to pathogen threats, but can also have behavioural adaptations, for which humans have the tendency to exclude individuals which display physical pathogen clues, as infection or wounds. However, disgust and rejection can also be elicited by the proximity of individuals which present themselves as disruptive of the idea of traditional society. As individuals display large variation in the calibration of their behavioural immune systems, they will also display different attitudes in terms of preference for traditional social attitudes. As Terrizzi et. al (2013) write, “social conservatism is in part an evolutionary evoked disease-avoidance strategy”.  

Conservatism is defined as the adherence to cultural and social traditions and is opposed to the advocation for change. As such, conservatism is linked to negativity towards potentially disruptive forces, such as immigrants and foreign groups, and to the upholding of culturally defined boundaries, such as those linked to rigid sex-related attitudes in relation to gay marriage and abortion. Inbar, Pizarro, and Bloom (2009) have found that for US voters the channel linking disgust sensitivity of individuals to political conservative preferences is specifically that of sex-related attitudes, while there is no significant association between disgust sensitivity and other political attitudes (as opinions on death penalty, gun control, or welfare). These findings have been confirmed by Crawford, Inbar, and Maloney (2014), which have found the degree of disgust sensitivity of US respondents to be linked to their attitude with respect to groups which challenge traditional sexual values and norms. Respondents with the higher disgust sensitivity were those who felt that moral standards were threatened by the group which the study identified as “sexual-morality threatening”, hence “sexually active youth, gays and lesbians, pro-gay activists, and feminists”. To conclude, Tybur, Inbar, Molho, and Guller (2015) have modelled the path between disgust sensitivity and ideology. They found that individuals who identify as conservative on the political and economic scale are those which display a high disgust for pathogen elements, which is then channelled through sexual disgust into the rejection of elements disruptive to traditional society.  

Thank you for reading, we hope to see you at our next event! 

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s