This is the author's version of the work. Please cite as:

    Arzheimer, Kai and Jürgen W. Falter. "Voter Behaviour." Encyclopedia of Political Behaviour. Eds. Kaid, Lynda Lee and Christina Holtz-Bacha. London: Sage, 2008. .
    [BibTeX] [HTML]
    title = {Voter Behaviour},
    author = {Arzheimer, Kai and J{\"u}rgen W. Falter},
    booktitle = {Encyclopedia of Political Behaviour},
    publisher = {Sage},
    year = 2008,
    editor = {Kaid, Lynda Lee and Holtz-Bacha, Christina},
    keywords = {voting},
    address = {London},
    html = {}

Voter Behaviour

Between the early 1940s and the late 1960s, four basic models of voter behavior have been proposed on which almost all studies of electoral behavior draw. These models describe how humans react to environmental factors and choose between different courses of action. Homo sociologicus (more or less implicitly) forms the basis of the approaches to voting behavior laid out in the first three parts of this entry. In contrast rational voter theory explicitly invokes homo oeconomicus through deductive reasoning. A closer examination reveals, however, that these seemingly very different approaches are in fact complementary and can be regarded as aspects of an overarching model. In the past few years this line of reasoning has become increasingly present both in social-psychological as well as rational choice writings.

The Micro-Sociological Model

The micro-sociological model was developed in the early 1940s by Paul F. Lazarsfeld and his colleagues and its formulation in The People’s Choice first appeared at the end of World War II. A milestone of modern electoral research, itwas also criticized for its methodological and empirical deficiencies and these critiques informed the design of the follow-up study Voting.

The motivating question for Lazarsfeld and his colleagues can be found in the sub-title of The People’s Choice; how do voters develop concrete vote intentions over the course of an election? Lazarsfeld et al. investigated this question by conducting an intensive study of Erie County, Ohio during the 1940 presidential election. They interviewed a representative sample up to seven times over the course of the campaign with regard to vote intention, their evaluation of the candidates and assessment of the major political issues. By doing so the researchers sought to determine how each individual voter developed their political attitudes over time and the impact of the campaign on this process.

Lazarsfeld et al. rapidly determined that socio-structural variables, above all socio-economic status and religious affiliation, strongly influenced vote intention for both major American parties. Taken together with a voter‘s living situation (urban or rural), the researchers constructed an “Index of Political Predisposition” with an extremely accurate predictive capability. Blue-collar workers and Catholics disproportionately trended toward the Democrats while Protestants and middle class voters predominantly supported the Republicans, with the interaction of both variables strengthening these effects.

With muted reservations, the authors concluded that the political preferences of their respondents were largely socially determined. For many voters, party choice was fixed months before the election and new information was used selectively to reinforce rather than challenge or update prior opinions. These findings were far removed from the ideal of the responsible democratic citizen, painstakingly informing themselves about the various parties and candidates before coming to a decision based on sober reflection.

As they conceded, however, Lazarsfeld et al. could only tentatively explain why socio-structural variables influenced vote choice so strongly, despite the relative anonymity of individual members to these large, impersonal structures. The authors argued implicitly that socio-structural variables could be viewed as indicators of membership in a mostly homogenous social environment of friends, family, neighbors and colleagues with similar political views. This web of interactions is then capable of reinforcing wavering individual opinions through social pressure. In these circumstances so-called Opinion Leaders played an important role by intensively informing themselves about political events through the media and then passing their observations on to less interested or educated citizens. To describe this relationship Lazarsfeld et al. formed their famous „two-step-flow“ hypothesis of political communication.

The Columbia-Group’s emphasis on the immediate social environment disposed them to observe an interesting phenomenon: If a voter‘s social environment is not homogenous and they belong to multiple social groups with incompatible political norms, conflicting behavioral expectations (cross pressures) should develop. To explain non-voting or party-switching, two phenomena that electoral researchers have always been pre-occupied with, Lazarsfeld et al. were forced to rely above all on cross pressures in the immediate social environment.

The Macro-Sociological Model

In contrast to the Columbia study, the macro-sociological approach focuses its explanations on processes at the level of the entire society. In Germany this approach was initially forwarded by M. Rainer Lepsius who was primarily occupied with “social-moral milieus”, a key characteristic of German society in the Imperial and Weimar periods. Internationally Lepsius had little impact, while even within the German literature his approach was soon displaced by a competing macro-sociological model that argued from the outset with abstract categories, was tailored to explain a larger area (Western Europe) and was easily portable to other contexts. This model was the “cleavage” theory of Seymour Martin Lipset and Stein Rokkan, originally formulated in the comprehensive introductory chapter of their work Party Systems and Voter Alignments.

By “cleavage” Lipset and Rokkan mean a social “fault line”, a sustained social conflict pitting (at least) two large groups with conflicting social interests (and defined by their social characteristics) against one another. According to Lipset and Rokkan, European social conflicts can be systematically ordered and divided into four groups:

  1. Conflicts between the national Center and the subordinate Periphery,

  2. Conflicts between the State and the Catholic Church;

  3. Conflicts between Urban and Rural territories; and

  4. Conflicts between Labor and Capital.

These four conflicts ultimately go back to processes of modernization. The first two refer predominantly to the cultural sphere and hearken back to the development of modern nation-states, while the latter conflicts are above all economically motivated and consequences of the Industrial Revolution.

For Lipset and Rokkan, social conflicts become politically relevant if a specific set of conditions is fulfilled:

  1. The conflict must remain virulent over a long period and play a central role in the life of the affected individuals.

  2. Social mobility must be low, so that one typically remains a lifelong member of the relevant social group.

  3. Those affected by the conflict must have the motivation and opportunity to ensure that their interests are incorporated into formal associations.

  4. The leaders of these “pressure groups” must found their own party or agree to some form of coalition with a pre-existing party.

  5. This party must have an opportunity within the electoral system to cross the threshold of parliamentary representation.

Under these conditions social conflicts achieve a sort of political reification. The parties that develop are understood as the agents of social groups and are treated as such by group members. The format of the party system that develops, such as the number of parties or polarization between them is determined by the number of relevant social cleavages and whether these fault lines run parallel to or overlap with one another. So long as the system of social conflicts remains stable, for example when parties negotiate a lasting compromise that is also acceptable to their represented social groups, the party system will remain fundamentally stable.

Lipset and Rokkan’s unpacking of the relationship between social structures and the party system is highly internally consistent and constitutes a powerful analytical frame, in that prior findings on voting behavior are easily integrated into a cleavage theory. An obvious deficiency in their model, however, is the failure to consider the individual level and the role of communication. Lipset and Rokkan are silent on why individual voters usually behave empirically as elites expect them to.

It is possible to close the micro-level gap and integrate Lipset and Rokkan’s macro-sociological model with the complimentary micro-sociological findings of the Lazarsfeld-Group and Lepsius’ work at the meso-level. But even this combined approach has a major deficiency, as it only poorly explain moments of political change. For relatively short-term fluctuations in the relative strength of political parties, which lead to relatively frequent changes in the size and composition of governing parties or coalitions, the picture of an ideal-typical homo sociologicus blindly adhering to the norms of his reference groups is unsatisfying. The socio-psychological model represents a solution to this problem and its findings are highly complementary with the previous sociological models.

The Socio-Psychological Model

Ten years after The People’s Choice Angus Campbell and his associates at the Survey Research Center published their first major election study. The Voter Decides was distinct from the Lazarsfeld-Group’s work in two respects. First, Campbell and his co-authors conducted a random sample covering the entire United States, as opposed to prior regionally limited inquiries. Second, Campbell et al. initially explained voting behavior exclusively through psychological variables, specifically the evaluation of candidates, their positions on the major political issues and their “party identification”, the degree of attachment to a political party. Initially all three psychological variables were considered equally important. Sociological variables, of primary importance to Lazarsfeld et al., were held in The Voter Decides as exogenous and remained unconsidered.

Campbell et al. were initially and correctly criticized for almost fully ignoring the social context of vote choice and for relying on variables so temporally and empirically connected with the act of voting that the model risked tautology. Campbell et al. reacted to these critiques by extending their inquiries to include the broader social context, a social-psychological model of behavior that would become identified with the University of Michigan as the “Ann-Arbor-Model”. Put forward in the American Voter, this new model demonstrated its effectiveness by using surveys from the 1952 and 1956 elections. The response from the scientific community was overwhelming: The American Voter became one of the most influential monographs in the history of electoral research and the Ann-Arbor-Model dominated the study of voting behavior in western democracies for many years after its publication.

The American Voter deviated from the Michigan-Group’s past work in two respects. First, party identification was now taken as a long-term stable variable, causally prior to individual evaluations of candidates and political issues. Second, psychological variables were no longer taken as given, but rather were seen as influenced by a voter’s sociological background. This included the experiences of an individual’s reference groups with the various parties and an integral role for the strengthening or weakening of opinions through a voter‘s immediate social circumstances. The Ann-Arbor-Model can therefore be considered an extension of sociological theories of voting behavior.

It is often overlooked, however, that the American Voter describes a wide range of potential influencing factors, in later years were seen as alternatives to the social-psychological model. These factors include for example the institutional context, economic situation or personality structure of voters.

Using the famous image of the “causality funnel” Campbell et al. summarized the relationship of all these varied factors. Individual vote decision is understood as the result of a complex process principally traced far back into a voter’s past. At the moment of the vote decision itself, only the previously identified psychological variables are of interest. The further one moves back into a voter’s history, the more potential influences have to be taken into account to explain the final behavior. The causality funnel therefore expands into the past until a complexity of factors is reached that the researcher can no longer untangle.

Despite its evident advantages, a dispute arose over the next several years concerning the Ann-Arbor-Model’s portability outside of the American context. Especially problematic for the Michigan-Group was the central concept of party identification: The idea of a “psychological membership” seemed too dependent on the peculiarities of the American system, particularly the (relatively) stable two-party system, the organizational weakness of the major parties and the absence of historically-based ideological conflicts.

In an influential contribution Russell Dalton, Paul Allen Beck and Scott C. Flanagan showed that long-term stable party identification did not imply a “psychological party membership”. The characteristic coalitions between social groups on the one hand and ideological parties on the other in Europe, as set out by Lipset und Rokkan, could instead be considered the functional equivalent of party identificationdescribed in the American Voter.

The Rational Voter Model

The theory of rational voting goes back to Anthony Downs’ pioneering study An Economic Theory of Democracy. Building on prior work by Kenneth Arrow, Joseph Schumpeter, Herbert Simon and others, Downs applied the tenets of neo-classical economics to voting behavior and provided the impetus for a new and fruitful research agenda in political science.

Downs’ concept differed from previous approaches in two respects. First, Downs’ approach to electoral research was primarily theoretical: Downs engaged in no empirical studies, but rather limited himself to deductively deriving axiomatic propositions that could be empirically tested. Second, the approach founded by Downs was based much more strongly on formal modeling than earlier approaches. While the sociological and socio-psychological approaches tied into experiences from everyday political life and seemed intuitively plausible despite their abstractions, the rational choice approach initially struck many researchers as all too artificial and unrealistic. That the contributions of rational choice theorists were often presented through systems of equations only strengthened this impression.

The starting point for Downs is the assumption that politicians and voters behave as rational actors in a market, in which political power (in the form of votes) is exchanged for the realization of political objectives. The “rationality” of actors is therefore understood in a formal sense that has nothing to do with “reasonableness” in the commonly accepted sense, but rather is solely related to the decision between alternative actions.

According to the model, rational actors possess stable and transitive preferences, which gives them the ability to select from a set of alternatives to maximize their benefits. “Benefits” are not limited to economic gains for an actor, but rather any result that is in line with their preferences. “Stable” means nothing more than that the preferences of actors remain constant during the period in question; “transitive” means that there are no contradictory or cycling preferences. A rational actor that prefers a government formed by Party A to one formed by Party B, and prefers Party B to Party C, must therefore prefer Party A to Party C given a choice between the two.

If one assumes that political programs can be placed as positions on an ideological Left-Right continuum, a rational voter will choose the party that stands nearest to their “ideal point” on that continuum (the point where their benefits are maximized). At the same time, parties will formulate their political programs with an eye toward maximizing their vote total. As the preferences of actors are seen as stable, changes in behavior are only explained through structural changes, such through the entry of a Party D.

How the “benefits” of actors persist and how their preferences come about are not discussed in rational choice models. As the preferences of actors are as a rule constructed from observations of their behavior, the rational voter approach is fundamentally tautological: The rationality postulate is an axiom rather than an empirically tested hypothesis.

But this tautological structure is precisely the greatest strength of the rational choice approach. It makes it possible to connect psychological or sociological models to the rational choice approach by treating them as mechanisms for the construction of preference sets. Thus it is possible to hold exogenous the complex and often-idiosyncratic backgrounds of individual voters in order to focus on the influence the structural features of a situation have on decision-making.

Downs himself recognized some of the complications that arose from exporting market behavior to electoral research. The most famous of these problems is the so-called „paradox of voting“: Independent of the electoral system, in a mass-democracy with millions of voters the probability of any one voter casting the winning vote is impossibly small. There is therefore effectively no relationship between a voter’s behavior and the victory of their preferred party. While it is unlikely that an actor can draw some personal (instrumental) benefit from casting their ballot, participation certainly entails costs. A voter has to spend time and/or money to build a picture about the intentions of the various parties (information costs). Furthermore, the acts of voting or voter registration often take time, which means foregoing other material or immaterial benefits (opportunity costs).

The net benefit of voting is therefore always negative and rational individuals should not choose to vote. This conclusion contradicts actual voter turnout, which consistently reached 70-80% in many democratic states in the 20th Century. Many solutions have been suggested to solve the paradox of voting, though all have their complications.

The uncertainty inherent in the act of voting is not limited to whether one’s own behavior can affect the outcome. For example, a voter cannot be certain whether she will receive their desired outcome for the actual costs spent; her party could lose the election. Uncertainty also exists regarding a party’s future actions. Even when parties intend to honor promises made during the election, changes in the general political situation could cause them to depart from their programs.

Voters in mass-democracies find themselves thus in a “low-cost situation” (Kirchgässner). Rational choice explanations typically do not delve further into the intricacies of behavior in low-cost situations, as it is already highly irrational under these circumstances for rational actors to put effort into collecting information or engaging in a cost-benefit analysis. Instead moral and expressive patterns dominate behavior in these situations; decisions are made on the basis of everyday information, group norms or fundamental ideological beliefs, which function as information shortcuts (Popkin). Rational voters therefore often behave as the sociological or social-psychological models would predict. Such considerations were already evident in Downs’ approach and stand at the center of the research agenda for some newer theories of rational voting.

Kai Arzheimer / Jürgen W. Falter

See also: Lazarsfeld, Paul F., Party Identification,People’s Choice, The (book), Survey Research Center , Voting (book), Two – Step Flow Model of Communication

Further Readings and References

Berelson, B., Lazarsfeld, P. F., & McPhee, W. N. (1954). Voting. A Study of Opinion Formation in a Presidential Campaign. Chicago: Chicago University Press.

Brennan, G., & Lomasky, L. (1993). Democracy and Decision. The Pure Theory of Electoral Preference. Cambridge, New York: Cambridge University Press.

Campbell, A., Converse, P. E., Miller, W. E., & Stokes, D. E. (1960), The American Voter. New York: John Wiley.

Dalton, R. J., Beck, P. A., & Flanagan, S. C. (1984). Electoral Change in Advanced Industrial Democracies. In R. J. Dalton & S. C. Flanagan & P. A. Beck (Eds.), Electoral Change in Advanced Industrial Democracies: Realignment or Dealignment (pp. 3-22). Princeton: Princeton University Press.

Downs, A. (1957). An Economic Theory of Democracy. New York: Harper.

Kirchgässner, G. (1992). Towards a Theory of Low-cost Decisions. European Journal of Political Economy, 8, 305-320.

Lazarsfeld, P. F., Berelson, B., & Gaudet, H. (1944). The People’s Choice. How the Voter Makes up His Mind in a Presidential Campaign. Chicago: Columbia University Press.

Lipset, S. M. & Rokkan, S. (1967), Cleavage Structures, Party Systems, and Voter Alignments: An Introduction, in: Lipset, S. M. & Rokkan, S. (Eds.), Party Systems and Voter Alignments: Cross-National Perspectives (pp. 1-64). New York, London: Collier-Macmillan

Popkin, Samuel L. (1994). The Reasoning Voter. Communication and Persuasion in Presidential Campaigns (2 ed.). Chicago, London: University of Chicago Press.