My doctoral research has afforded me an opportunity to study in depth how political knowledge conditions the decision-making process of voters. I want to further develop this research by focusing on the relationship between political knowledge and party identification, turnout, and vote choice. I aim to address new challenges facing western democracies. These democracies have witnessed a steady decline of electoral participation (Blais 2010) and their citizens are less likely to identify with political parties than they did in the past (Dalton 1984; Dassonneville and Hooghe 2016).  Finally, past research has found that traditional strategies focused on civic education do not lead to sizable increases in political knowledge among the public, which suggests that there is a need to identify new strategies to enhance citizens’ political competence. I will examine how political knowledge relates to turnout as well as partisan dealignment. I will also investigate what information helps citizens identify and vote for the party that best corresponds to their political interests and preferences.


The Conditional Impact of Political Knowledge on Turnout

Political scientists generally conceive of models of turnout as different from models of vote choice and study these two questions separately. It is assumed that voters decide first whether they will vote and only subsequently decide whom to vote for (Campbell et al. 1960, Miller and Shanks 1996, Blais 2000, Gidengil et al. 2012). This premise has been challenged in recent research. Clarke et al. (2009) argue that voters decide to vote rather than abstain for the same reasons that explain for which candidate they vote – namely, that they like a given party more than its competitors. If voters have no preference for either party, they have no reason to go out and vote.

If this argument is correct, the results observed in my dissertation regarding vote choice may have an incidence on turnout as well. If the determinants of preferences for a given party vary across knowledge levels in the case of vote choice, it may also be the case that voters at different levels of political knowledge are motivated to vote rather than abstain for different reasons. Knowledgeable voters may be motivated to vote because they like a given candidate’s policy proposals, while less knowledgeable voters may be motivated to vote based on feelings of identification with candidates’ parties. Finding confirmation for this hypothesis would indicate that the determinants of vote choice also matter for participation, providing greater insights in the causes of abstention and participation. 

The analysis will test a turnout model on different subgroups of the electorate defined by their level of political knowledge. Building upon a model of political participation defended by Riker and Ordeshook (1968) the impact of three determinants of turnout will be taken into account: party preferences, civic duty, and the cost of voting. It is hypothesized that preferences for political parties will play a different role for individuals with varying levels of political knowledge in the propensity to vote or abstain. Knowledgeable voters should be motivated to vote when a party defends a policy they care for, while party identification should have a stronger impact among less knowledgeable voters. Furthermore, the cost of voting should play a weaker role in the decision of knowledgeable voters, since their higher level of political knowledge enables them to overcome these costs. However, feelings of civic duty are not expected to have a differentiated impact in the decision to vote or abstain since civic duty is a deeply held conviction rather than the result of abstract reasoning. Investigating these hypotheses will provide researchers and social entrepreneurs alike with tools to combat declining turnout in advanced democracies. 


Political Knowledge: Cause or Consequence of Political Dealignment?

Political dealignment refers to a structural and over-time process in which the ties between citizens and political parties are eroding (Dalton 1984, Dalton and Wattenberg 2002). This process has given rise to an increasingly large group of citizens who do not identify with any political party. As a result of this change, there is an increased volatility in election results, more split-ticket voting, and voters decide later which party to vote for during elections (Dalton and Wattenberg 2002, McAllister 2002). Furthermore, citizens are more likely to express their political demands through alternative methods, such as protests, rallies, and participation in social movements instead of traditional partisan politics (Dalton and Welzel 2014).

One explanation for this phenomenon attributes responsibility to increasing education and political knowledge. The argument holds that partisan attachments are used by citizens as a heuristic, helping them choose for whom to vote when they lack knowledge on a given issue (Dalton and Wattenberg 2002). According to Dalton (1984), increasing cognitive mobilization has armed citizens with the necessary skills and resources to make their vote choices independently of partisan attachments. This functional view of partisanship implies that knowledgeable citizens are not in need of a partisan shortcut and thus derive less benefit from identifying with political parties. On the other hand, for less knowledgeable citizens, identifying with a party is still a useful heuristic.

This explanation raises more questions than it answers. Many studies have found that partisans have higher levels of political knowledge than nonpartisans, rather than the opposite (Campbell et al. 1960, Dettrey and Palmer 2013). Given this situation, the idea that high levels of political knowledge lead to dealignment is counter-intuitive. How can increased levels of political knowledge lead to dealignment if partisans are more knowledgeable about politics than nonpartisans? By means of a panel study, I aim to disentangle the nature of the relationship between political knowledge and partisanship. The study will consider the possibility that this relationship is bidirectional. Up to this point, studies have relied on the assumption that political knowledge determines partisan identification. However, the reverse is also possible. Individuals who strongly identify with a political party may feel like they have a greater stake in politics and thus devote more attention to political debates and elections. This finding would challenge the assumption that the relationship between these two variables is unidirectional and explain why political knowledge is positively correlated with party identification rather than the opposite.

 


The Knowledge Gap and the Representation Deficit

Due to structural disadvantages and limited opportunities, individuals who belong to minority groups exhibit lower levels of political knowledge (Delli Carpini and Keeter 1996). Such groups include women, visible minorities, linguistic minorities, the less affluent, youth, and the less educated. Since the doctoral dissertation has shown that individuals with higher levels of political knowledge vote more ideologically, it is likely that voters who belong to the groups identified above face additional hurdles identifying the electoral candidate that best represents their ideological preferences. This may lead to a systematic underrepresentation of these voters and a lack of willingness on the part of governments to promote their interests over those of constituencies with higher rates of ideological voting.

I propose to examine more closely this knowledge gap. First, I aim to quantify the importance of the knowledge gap between marginalized voters and the rest of the population. Second, I want to identify how the decision-making process leading to the vote differs for these various groups. Finally, I want to ascertain whether electoral results would change if there were no gap in political knowledge and which parties gain and lose votes due to this gap. I hypothesize that the knowledge gap makes it difficult for marginalized voters to elect candidates who best defend their interests, explaining why there is limited incentive for elected officials to do so. Confirming this hypothesis would show the importance of improving political knowledge among marginalized voters.


Experimentally Randomizing Party Identification and Political Knowledge

Experiments rely on randomization to identify causal relationships. However, some concepts can be difficult to manipulate experimentally. This is the case of long-term and long-held individual characteristics, such as party identification and political knowledge. Until there are recognized ways for political scientists to randomly induce party identification and knowledge in experiment participants, our ability to make causal statements about these two key concepts is limited. I thus suggest investigating new ways to randomly manipulate these attributes in experiment participants.

The strategy I would recommend in the case of party identification relies on previous work from the discipline of psychology. Social identity theory has established that even minimal prompts can lead individuals to identify with a particular group and modify their behavior accordingly (Tajfel 1978, Tajfel and Turner 1979, Sherif 1954, 1958). This is true even in experiments in which psychologists alter perceptions of group membership to study in-group bias. In future research, I would like to rely on the foundations of social identity theory and its experimental practices to develop a process to randomly allocate party identification during experiments.

Another strategy would be used to experimentally induce political knowledge. According to most models, external benefits increase political knowledge (Converse 1964, Luskin 1990, Lupia and McCubbins 1994). This means that individuals are more motivated to learn about politics if they get something in return for acquiring this knowledge. I thus propose to include this factor in an experiment on political knowledge. This random variation can be leveraged to make causal inferences regarding how this higher knowledge determines the behavior of the treatment group in comparison to the control group. 


Can Voters Learn What They Need to Know

The main thrust of the literature seeks ways to improve political knowledge among the general public in order to help them identify and vote for the party that best corresponds to their political preferences. Previous work has produced pessimistic findings, as civic education and increasing media coverage do not appear to broadly increase political knowledge in the electorate (Hooghe and Dassonneville 2011; Prior 2007; Delli Carpini and Keeter 1996; Galston 2001; 2004). A second strategy to address this problem consists of identifying specific information that can help voters cast their vote for the party that best corresponds to their preferences. The argument is that voters may be able to identify and vote for the party that best defends their interest even with modest levels of information (Sniderman et al. 1993; Popkin 1994; Lupia 1994; 2016). Such informational shortcuts are known as heuristics.

In order to identify useful heuristics, I propose an experimental research design based on my previous experience working on political knowledge in an experimental framework. The experiment I propose includes one control group and three treatment conditions. In the control group, the level of political knowledge of participants and their voting intentions are measured. This allows researchers to measure gaps in voting intention due to differences in political knowledge. Participants in the treatment groups would follow the same process, but would also be given specific information before volunteering their voting intentions. By varying the content of the information provided to the participants, it is possible to contrast and compare the effectiveness of distinct pieces of information in helping voters identify which party best defends their interests. By comparing the gap in voting intentions across knowledge levels in each of these treatment groups to the same gap in voting intentions in the control condition, it will be possible to evaluate which type of information is most beneficial to voters trying to identify the party that best represents their preferences and interests. 


Using Campaign Data to Understand the Nature of Party Identification

The concept of party identification is strongly contested in political science. For some, it is a psychological and social attachment that is developed early in life and enduring throughout the life cycle (Campbell et al., 1960, Green et al., 2004). For others, it is a running tally of short-term evaluations which changes depending on the actions and performance of political candidates and parties (Fiorina 1981, Clarke and McCutcheon, 2009). Supporting literature exists for both interpretations.

I believe that recent data acquired during the 2015 Canadian election can contribute to this ongoing discussion. The election was very long, stretching from August to October. The length of the campaign can be leveraged to conduct studies that would be difficult to perform in shorter campaigns. The Local Parliament Project surveyed Canadians during this campaign. This study is the largest study ever conducted during a Canadian election. More than 35, 000 Canadians participated in the study, providing a daily sample of over 600 hundred individuals. This gives the study a formidable sample size that can be leveraged to consider variations over time in ways that cannot be mirrored by traditional surveys.

The length of the campaign, the variability in voting intentions over the course of the campaign, and the high sample size can be used to conduct innovative research on party identification. If the supporters of the social-psychological model of party identification are correct, then party identification should remain constant throughout the campaign, even as the campaign unfolds and voting intentions fluctuate. On the other hand, if the running tally conception of party identification is correct, then party identification will vary along with campaign events and voting intentions. This research project contributes to the literature on the nature of party identification.