21
2
THE SOCIOLOGICAL AND
SOCIAL-PSYCHOLOGICAL
APPROACHES
Vincent L. Hutchings and Hakeem J. Jefferson
Early proposed determinants of the vote choice
Why do citizens vote for one candidate rather than another? Various explanations have been
offered over the roughly 70-year history of voting behavior research, but two explanations in
particular have garnered the most attention and generated the most debate in the literature on
voting behavior. These explanations are known as the Columbia Model and the Michigan
Model, and describing these two theories – including their respective strengths and weaknesses
– is the subject of this chapter.
Before summarizing these two foundational theories, it is important to ﬁrst understand the
prevailing view about the origins of candidate preference prior to the scientiﬁc study of this
issue. Before the advent of voting behavior research, many believed that ordinary citizens, much
like elites, viewed the parties and candidates in ideological terms and made their political judgments
largely on this basis. At the time that research on voting behavior was ﬁrst developing,
and indeed long before this time, this explanation held considerable weight. Even today, variants
of this explanation remain in political science and such views are often implicit in standard
political reporting, which reﬂects a powerful, though empirically unsupported, view of the
ordinary citizen as ideologically committed and attentive to the particulars of the American
political system (Achen and Bartels 2016). Thanks in part to the pioneering scholarship of the
authors of the Columbia and Michigan Models and earlier work by journalist Walter Lippman,
we now know that voters are not typically motivated by ideology and are often ignorant of the
most basic facts of politics (Campbell et al. 1960; Lazarsfeld, Berelson, and Gaudet 1944; Lippmann
1922). Indeed, even those familiar with public policy debates are often unaware of the
position of the candidates and the parties on the issues (Converse 1964).
The Columbia Model – described as such because of the contributions of Paul Lazarsfeld,
Bernard Berelson, and other scholars at Columbia University – helped to explain how voters
make sense of the political world despite their general inattentiveness to politics. This groundbreaking
theory was ﬁrst articulated in The People’s Choice, the ﬁrst study to rely on representative
sample surveys to explain what factors motivated individual voting decisions (Lazarsfeld,
Berelson, and Gaudet 1944). The Columbia scholars examined survey respondents in Ohio
from May to November of the 1940 presidential election. They found that exposure to the
election campaign essentially had two effects: it reinforced the choices made by early deciders and
V. L. Hutchings and H. J. Jefferson
22
it motivated the latent predispositions of uncommitted voters. These predispositions were associated
with three types of social characteristics: (1) a citizen’s class status; (2) a citizen’s racial or
religious identiﬁcation; and (3) the region of the country a citizen lived in and whether they
resided in an urban or rural setting. The social characteristics that were activated by the campaign
were described as the Index of Political Predispositions (IPP) and this index predicted the 1940 vote
quite accurately. For example, the IPP predicted that rural, middle class Protestants would vote
for the Republican candidate and in fact this held true for about 74 percent of this group.
The Columbia scholars followed up on The People’s Choice with the equally impressive book
Voting (Berelson, Lazarsfeld, and McPhee 1954). As with The People’s Choice, Voting argues that
one’s social identiﬁcation, or political predispositions, largely accounts for how citizens think
and act with regard to politics. In Voting, Berelson and his co-authors surveyed the citizens of a
moderate-sized community, in Elmira, New York in 1948. In this study, Berelson and his colleagues
provide much greater evidence, than in the previous study, of the importance of the
social characteristics outlined above. Because their study focuses on only one community,
however, they could not fully explore the role that place of residence plays in shaping political
attitudes.
The Columbia Model was an important ﬁrst step to answering the question posed at the
beginning of this chapter and many of its insights regarding the political signiﬁcance of social
groups remains relevant today. Still, in spite of its strengths, the theory suffered from a very
serious weakness. A model of vote choice based on social-demographic categories does not
account very well for swings in election results over time. In other words, how could the Index
of Political Predispositions explain the vote for Truman in 1948 (a very close Democrat victory)
and the subsequent landslide victory of the Republican presidential nominee, Dwight Eisenhower,
in 1952? The social characteristics of the voters could not account for these shifting
preferences since the demographics changed little from 1948 to 1952. For this reason, an additional
explanation was needed to account for over-time change in voting preferences.
Although the Columbia school’s perspective on voting behavior proved inﬂuential, it was
insufﬁciently dynamic, thus giving rise to an alternative model of voting. This newer model,
proposed by scholars at the University of Michigan, offered a somewhat different explanation of
voting behavior and it eventually became the dominant model of voting in the scholarly literature.
Unlike the Columbia Model, which emphasized the social characteristics of voters, the
Michigan Model – as this approach came to be known – focused on the individual attitudes of
the voters as well as their identiﬁcation with one of the major political parties. Brieﬂy, the argument
of The American Voter (Campbell et al. 1960) can be summarized into three larger points.
First, the Michigan scholars argued that people identify psychologically with political parties just
as they do with religious, class, racial, and ethnic groups. Second, identiﬁcation with a political
party should be viewed as a long-term component of the political system and therefore this
identiﬁcation should persist even when people vote for a candidate representing the opposing
party. Lastly, party identiﬁcation is powerful not merely for its direct effects on vote choice, but
also for its indirect inﬂuence on attitudes associated with the vote. These attitudes include evaluations
of the major party candidates, the issues of the day, and the political parties.
The Michigan scholars offered a distinctive deﬁnition of party identiﬁcation. They deﬁned
it as a sense of psychological attachment to one of the major parties. Importantly, this did not
mean that one ofﬁcially belonged to the party in the sense that one held formal membership in
the party. By attachment they meant a sense of attraction (or repulsion) similar to the sense of
attachment one feels to one’s class, racial, or religious group. According to Campbell and his
co-authors, party identiﬁcation has two characteristics: direction and strength. By direction,
they simply meant whether one is a Democrat, Independent, or Republican. By strength, they
Sociological/social-psychological approaches
23
meant the intensity of one’s attachment to the parties. Thus, in this formulation, one’s attachment
to the party is either strong, not very strong, or one is an Independent who usually leans
toward one of the two parties.
It is important to note that party identiﬁcation could have been deﬁned as one’s prior voting
record, and in fact other researchers had previously described it this way. The Michigan scholars
rejected this deﬁnition because they did not want to confuse one’s attachment to the party with
how one votes at election time. They are not, strictly speaking, the same thing. Although not
common, Republican identiﬁers can, under some circumstances, vote for a Democrat, and vice
versa. If partisan attachment were not deﬁned as something other than the vote this could not
be recognized, let alone explained. In other words, the authors were interested in the extent to
which the one (attachment) explains the other (voting behavior) and they could not explore this
relationship unless the concepts were conceptually distinct.
As we have noted, the Michigan Model offered an explanation of vote choice based on a
psychologically oriented perspective. This model was different from previous models because it
did not locate explanations in the (relatively) immutable social characteristics of the voter, or in
their presumed ideological sophistication, but in a broader set of individual attitudes. It is in this
sense that The American Voter offers a “psychological” explanation of voting, although it does not
strictly rely upon models developed in the ﬁeld of psychology.
As indicated previously, the authors of The American Voter argued that attitudes about the
candidates, issues, and parties predicted the subsequent vote choice. More importantly, however,
levels of partisan identiﬁcation predicted these attitudes. In other words, to a large extent,
knowing which party a citizen identiﬁed with (and how strongly) determined how they would
evaluate the candidates, and what position they would take on the issues. Thus, for example,
citizens identifying with the Republican Party should also favorably evaluate Republican candidates
and generally agree with Republican leaders on the major issues.
An obvious criticism of this formulation is that policy positions and candidate evaluations can
also affect party identiﬁcation, just as party identiﬁcation can inﬂuence them. As plausible as this
may sound, the Michigan scholars mostly reject the alternative proposition that issues affect
partisanship. They adopt this view in part because of the relatively low levels of political knowledge
in the electorate and the general lack of ideological sophistication. In making their argument,
they also point out that party identiﬁcation typically develops in adolescence and tends to
be stable over time, while issue preferences and candidate evaluations usually arise much later
and are more ephemeral (see Carsey and Layman 2006 for more recent evidence largely consistent
with this view).
As evidence for the power of their theory, the authors of The American Voter examined
nationally representative survey data from the 1952 and 1956 presidential elections. Their results
are generally consistent with their expectations. They ﬁnd that their measure of partisanship is
strongly correlated with a variety of political attitudes about the candidates, parties, and current
issues. Without exception, the stronger one’s attachment to the parties is the more the respondent
tends to view the political world in ways consistent with that attachment. For example,
“strong Republicans” are much more likely to view Republican presidential nominee – and,
later president – Dwight Eisenhower favorably, agree with the Republicans on domestic and
foreign policy issues, and regard the Republican Party as a more effective manager of government
than Republicans with moderate attachment to their party, Independents, moderately
attached Democrats, and especially strong Democrats.
In their second test, Campbell and his co-authors argue that if party identiﬁcation affects
attitudes so deeply, the more partisan members of their sample should have more ideologically
consistent issue positions than less partisan respondents. In other words, respondents’ attitudes
V. L. Hutchings and H. J. Jefferson
24
about the candidates, issues, and parties should more consistently favor the Republican/
Democratic Party as the strength of partisan attachment increases. They ﬁnd that the evidence
supports these hypotheses. Finally, they show that their measure of partisanship is strongly
related to respondent vote choice in both 1952 and 1956.
According to the authors of The American Voter, issue preferences and ideological principles
are not the source of party identiﬁcation. Instead, partisanship is the result of pre-adult socialization.
Attachments to political parties, they argue, typically form in mid-to-late childhood largely
as a result of parental inﬂuence (similar to the acquisition of one’s religious afﬁliation). As evidence
for this claim, they use their survey data to show that respondents whose parents belong to
the same party were overwhelmingly likely to also identify with that party. If parents disagreed
with one another, the pattern was less consistent. An obvious problem with their evidence is
that it is based on respondent recall. It is possible that strong partisans recall their parents having
a similar party identiﬁcation even though this is not the case. This is called “projection,” and
there is certainly evidence in the literature that citizens tend to erroneously assume that others
(e.g., candidates, friends, relatives) have political views similar to their own. However, later
work by Jennings and Niemi (1968) examined high-school students and their parents’ partisanship
and found that there was considerable agreement.
Although the authors of The American Voter argue that partisanship is mostly stable over time
– at the individual level and in the aggregate – they do not argue that it never changes. Sometimes
it does and they identify two causes of these changes. The ﬁrst is what they call personal
forces or changes in an individual’s social environment. This was the least common explanation
for the roughly 20 percent of the respondents in their sample who indicated that they had
changed their party identiﬁcation.
They refer to the more common explanation as emanating from social forces. Unlike personal
forces that depend purely on personal circumstances, social forces involve experiences
shared across a large number of citizens. There are essentially two ways that social forces can
affect partisanship. The ﬁrst is through life-cycle changes. For example, young people are more
likely than older individuals to identify as Independents, whereas older people are more likely
to identify as partisans. The second way that social forces can affect partisanship is through
massive societal dislocations on the order of the Civil War or the Great Depression. From the
vantage point of the Michigan Model, issues can inﬂuence partisanship only when they achieve
this level of intensity.
In summary, the explanation for voting behavior offered by the Michigan Model can be
summarized into three larger points: (1) people identify with political parties just as they do with
their religious, class, racial, and ethnic group identity; (2) identiﬁcation with political parties
should be viewed as a long-term component of the political system that persists even when
people vote contrary to their identiﬁcation; and ﬁnally (3) partisanship is powerful not merely
for its direct effect on vote choice, but also for its indirect inﬂuence on attitudes associated with
the vote. The Michigan Model also holds that party identiﬁcation is typically formed before
citizens become adults (largely through parental inﬂuence) and that typically issues affect partisanship
only when they reach catastrophic proportions (e.g., during the Great Depression).
Partisanship and the normal vote
The concept of partisanship was reﬁned somewhat by Phil Converse, one of the co-authors of
The American Voter, some six years after the publication of this landmark book. In an article
entitled “The Concept of a Normal Vote,” Converse (1966) pointed out that there were
often large ﬂuctuations in presidential election outcomes from one year to the next with no
Sociological/social-psychological approaches
25
corresponding change in the overall distribution of party identiﬁcation. Converse interpreted
this to mean that the election outcomes could be characterized as ﬁtting into two categories: the
normal vote and short-term deviations from the normal vote. By the term “normal vote,” Converse
meant election results that simply reﬂected the voter’s stable commitment to either of the
two parties. In other words, since the majority of voters identiﬁed with the Democrats one
could expect the Democrats to win every presidential election if the electorate voted only on
the basis of their party attachment.
By “short-term deviations” from the normal vote, Converse meant that election outcomes
were sometimes inﬂuenced by ﬂeeting circumstances, such as scandals or international crises,
which did not substantially affect the overall balance of partisan support in the electorate. Sometimes
short-term forces worked to the beneﬁt of the Republicans (as in 1952 when General
Dwight Eisenhower, a war hero, was the Republican presidential nominee) and sometimes to
the beneﬁt of the Democrats (as in 1964 when the Democratic presidential nominee was the
former Vice-President of the recently assassinated President Kennedy). When these short-term
forces cancel out, or when they are essentially equal for both parties, Democrats were estimated
to receive 54 percent of the two-party vote.
Contemporary socio-psychological perspectives on voting
In the aftermath of its publication, The American Voter generated considerable controversy and
disagreement. In time, however, researchers focused less attention on the dominant role of partisan
identiﬁcation and more attention on the psychological underpinnings of the theory. In
particular, a number of scholars relied on social identity theory to help adjudicate long-standing
questions about partisanship and to reorient how we think about the concept. Green, Palmquist,
and Schickler (2002) were at the forefront of this effort, along with Greene (1999), and Huddy,
Mason, and Aarøe (2015).
In their 2002 book entitled Partisan Hearts and Minds, Green, Palmquist, and Schickler
develop an argument they had originally introduced years earlier in several inﬂuential articles.
In general, they argue that partisanship is a social identity and should be thought of in ways
consistent with our understanding of other salient social identities. Building on the arguments
put forth by the authors of The American Voter, Green and his co-authors also note that individuals
consider salient social groups when they think of the political parties and subsequently determine
their partisan loyalties by assessing which party most closely reﬂects their own unique set
of identities. Beyond this grounding of party identiﬁcation as a social identity, the authors
present evidence suggesting that partisanship demonstrates levels of stability akin to other social
identities. This stability is demonstrated at both the individual and the aggregate levels. On the
occasions when we observe shifts in partisan identiﬁcation, this is usually because individuals
perceive changes in the identities of those who belong to the party. For example, we may
expect shifts in partisan identiﬁcation if one of the two major parties were to become newly
associated with a different set of racial or ethnic groups. Interestingly, although their focus is
clearly on America, the authors demonstrate that partisanship in places like Great Britain and
Germany also operates similarly. The battles between Democrats and Republicans, or the more
complicated multi-party competition in other democracies around the world, are not merely a
ﬁght over ideas; they are also ﬁghts between members of distinct social groups engaged in the
same kind of conﬂict we observe between members of other social groups (for an opposing
view, see Abramowitz and Saunders 2006).
In a similar vein, Greene (1999) posits that we should consider party identiﬁcation through
the lens of social identity theory. That is, individuals derive a sense of their own self-worth from
V. L. Hutchings and H. J. Jefferson
26
their self-perceived membership in partisan groups. Consistent with early work on social identity
theory, Greene (1999: 394) suggests that partisan identiﬁers have an incentive to distinguish
their in-group from partisan out-groups, and engage in a style of social comparison that
“heighten[s] differences between the groups.” Using a psychological measure of group identiﬁcation,
Greene gauged the extent to which Democrats, Republicans, and partisan leaners
expressed a partisan social identity. Unsurprisingly, Greene ﬁnds that partisan social identity is
most pronounced for strong partisans relative to weak partisans and leaners. He also ﬁnds that
higher degrees of social identiﬁcation with a given party correspond with more negative perceptions
of the out-party relative to the in-party. This measure of partisan social identity was also
related to an engagement in partisan political behavior and an individual’s likelihood of
turning out.
Huddy and her colleagues (2015) have also sought to apply social identity theory to the issue
of partisan attachments. More speciﬁcally, they ask whether partisan identity is instrumental or
expressive. An instrumental understanding of partisan identity would suggest that individuals
collect some running tally of how well the parties have done and attend to questions of how
closely aligned the party’s goals and preferences are with one’s own. Focusing on the expressive
nature of partisan identity, however, builds on the idea that partisan identiﬁcation, like other
social identities, motivates individuals to behave in ways that protect the image of the group and
maintains its high regard relative to the out-group. To explore the alternative explanations of
partisan identity, the authors use campaign activity as their main outcome variable of interest.
Across different studies, respondents noted whether they planned to contribute money to presidential/congressional
candidates, and political organizations, and whether they planned to volunteer.
Whereas a measure of expressive partisan identity is related to the participatory behaviors
across the various studies, they ﬁnd that an ideological intensity measure (the authors’ measure
of instrumental partisanship) has a more limited inﬂuence.
In the past few years, scholars have reexamined the original perspectives outlined in the
Columbia and Michigan Models. In 2008, a new group of scholars – Michael Lewis-Beck,
William Jacoby, Helmut Norpoth, and Herbert Weisberg – sought to revisit the groundbreaking
arguments of The American Voter. Speciﬁcally, they sought to subject the theories introduced
in this volume to examination with more recent survey data (i.e., the 2000 and 2004 American
National Election Studies). This replication was a bold effort. If the groundbreaking results of
The American Voter are as enduring as the original authors believed, they should hold up even
when examined with survey evidence collected almost 50 years after the data used in the original
study. This updated test of the Michigan Model was explored in a book called The American
Voter Revisited.
Lewis-Beck and his co-authors designed their book to mirror the same set-up as the original,
with chapters mostly following the same pattern and addressing the same subject matter as the
original book. One interesting point highlighted in the replication is that, although the authors
of The American Voter emphasized individual attitudes, they did not neglect the role of socialdemographic
groups. Echoing results from Voting, as well as the original Michigan Model, the
authors of The American Voter Revisited ﬁnd that group members who identify strongly with the
group are more likely to vote in accordance with the group. These effects are particularly strong
for some voters (e.g., Jewish voters and, to a somewhat lesser extent, union members, Hispanics
and women) but much weaker for others (e.g., blacks and Catholics).
Lewis-Beck and his colleagues, again building on the model developed by the Michigan
scholars, argued that the inﬂuence that is exerted by social groups is done largely through
their impact on party identiﬁcation. In other words, just as the Columbia Model suggested,
social group attachments inevitably lead to partisan group attachments. However, once these
Sociological/social-psychological approaches
27
socio-demographic afﬁliations lead to the “appropriate” partisan identiﬁcation, it becomes
exceedingly difﬁcult to dislodge this attachment. And, typically partisan group attachments
exert more inﬂuence on political judgments than the social group memberships that give rise
to it. If interest group leaders try to change the party identiﬁcation of their members, for whatever
reasons, they are likely to have a difﬁcult time doing so – especially for older group
members.
In 2016, Christopher Achen and Larry Bartels also addressed the fundamental question of
what motivates voters. In their book, Democracy for Realists, they deconstruct what they refer to
as the “folk theory of democracy” – the notion that citizens evaluate political leaders on the basis
of ideology or issue preferences. The authors review a broad range of scholarship on this question,
both within and outside of the US. Consistent with the early Columbia and Michigan
Models, Achen and Bartels conclude that voters fail to meet even the more limited expectations
of the retrospective model of voting (i.e., rewarding and/or punishing elected ofﬁcials based on
assessments of their performance in ofﬁce rather than their adherence to abstract ideological
principles). Given their condemnation of the “folk theory of democracy,” the authors turn to a
more realistic basis for political judgment – salient social groups. Again echoing the pioneering
work of Lazarsfeld, Campbell, and their respective colleagues, Achen and Bartels conclude that
the vast majority of voters develop their attachment to the parties based on social group loyalties
(and pre-adult socialization from their parents). Similarly, they argue that the actions of political
ﬁgures are primarily interpreted through a partisan or social group lens.
Although there is considerable evidence in support of the foundational works in public
opinion, most of this evidence is derived from the US (although see Achen and Bartels 2016).
Research on both stable and emerging democracies suggest that the concept of partisanship does
not always perform as anticipated in other parts of the world. Holmberg (2007) provides an
excellent review of this literature. He notes, for example, that the concept of party identiﬁcation
translates better in some parts of Europe than others. More speciﬁcally, partisanship is more
volatile in the Netherlands than in Britain or Sweden. Moreover, the number of strong partisan
identiﬁers is mostly stable in the US, but has been declining in Sweden. In the case of emerging
democracies in Eastern Europe, Holmberg reports that partisan identiﬁers are also rare and –
more troubling for the standard Michigan Model – such attachments that are developing seem
mostly based on cognitive rather than affective factors.
Conclusion
Here, we have attempted to provide a broad overview of the early history of scholarship pertaining
to candidate choice and some of the more recent examinations of these theories both
domestically and internationally. In particular, we focused on two complementary, though distinct,
schools of thought that laid the foundation for much of our contemporary understanding
of how individuals come to make political decisions.
To recap, the Columbia Model, with its focus on the role of socio-demographic factors,
including social class and membership within salient social groups, brings to the fore a consideration
of how political decision-making is informed by one’s place in the social hierarchy and
one’s connection to others who share in their social identity. Focusing more directly on the
attitudes individuals hold toward political objects (e.g., candidates, parties, and issues), the
Michigan Model adopted a more social-psychological understanding of individuals’ decisionmaking
calculus. Describing partisan identiﬁcation as a psychological attachment to one of the
political parties, akin to the attachment one might have to one’s racial or religious group, the
Michigan Model placed at the center of our theorizing a concern with party identiﬁcation as a
V. L. Hutchings and H. J. Jefferson
28
key predictor of individual-level behavior. Remarking on its stability, the authors concede that
partisan identiﬁcation is not immune to change, but note that change requires either a shift in
an individual’s personal situation or the occurrence of a large-scale, salient social or political
event. With few exceptions, the claims put forth by the Michigan Model are replicated and
extended by the authors of The American Voter Revisited, published more than 50 years after the
original text.
Finally, as the other chapters of this handbook demonstrate, there is a great deal of new and
exciting work being done that both clariﬁes and complicates our understanding of the nature of
individuals’ political beliefs and behaviors. And while much of this work warrants careful consideration
by readers, we are particularly excited by scholarship that bridges literature in political
science and social psychology to interrogate partisan identiﬁcation as a meaningful social identity
that forms the foundation for citizens’ interactions in the political world. The chief tenets of
social identity theory, which center on the connection between one’s sense of self and her
membership in a salient social group, not only provide an interesting lens through which to
consider this important feature of the American political system. These tenets also allow us to
make sense of deep and persistent divides between individuals for whom a partisan identity is
about more than electoral choice, but occupies a central role in a broader set of social identities
that motivates behaviors across a range of domains. In fact, this way of thinking about partisan
identity takes us full circle to the Michigan Model’s early sense that there was an important
psychological underpinning to this politically meaningful construct.
In focusing on these two schools of thought and some of the subsequent scholarship they
have inspired, we hope we have provided the reader with a useful starting place for grappling
with the ideas and debates highlighted in other chapters of this handbook. And though many
scholars appreciate both the role of social groups and the centrality of partisanship in the American
political system, debates will undoubtedly continue in the literature about the relative
stability or dynamism of partisanship, the inﬂuence of context on political identity, the role of
social identities on political choice, and, in the case of new immigrants and emerging democracies,
the very roots of partisan attachment. Clearly, the work we have endeavored to describe
in the preceding pages represents the very beginning of what continues to be an exciting and
growing area of study in our discipline.
References
Abramowitz, A. I., and Saunders, K. L. (2006) “Exploring the Bases of Partisanship in the American Electorate:
Social Identity vs. Ideology,” Political Research Quarterly, vol. 59, no. 2, June, 175–187.
Achen, C. H., and Bartels, L. M. (2016) Democracy for Realists: Why Elections Do Not Produce Responsive
Government, Princeton: Princeton University Press.
Berelson, B. R., Lazarsfeld, P. F., and McPhee, W. N. (1954) Voting: A Study of Opinion Formation in a
Presidential Campaign, Chicago: University of Chicago Press.
Campbell, A., Converse, P. E., Miller, W. and Stokes, D. (1960) The American Voter, New York: Wiley.
Carsey, T. M., and Layman, G. C. (2006) “Changing Sides or Changing Minds? Party Identiﬁcation and
Policy Preferences in the American Electorate,” American Journal of Political Science, vol. 50 no. 2, April,
464–477.
Converse, P. E. (1964) “The Nature of Belief Systems in Mass Publics,” in Apter, D. E. (ed.) Ideology and
Discontent, New York: The Free Press: 206–261.
Converse, P. E. (1966) “The Concept of a Normal Vote,” in Campbell, A., Converse, P. E., Miller, W.,
and Stokes, D. (eds.) Elections and the Political Order, New York: Wiley: 9–39.
Green, D., Palmquist, B., and Schickler, E. (2002) Partisan Hearts and Minds: Political Parties and the Social
Identities of Voters, New Haven: Yale University Press.
Greene, S. (1999) “Understanding Party Identiﬁcation: A Social Identity Approach,” Political Psychology,
vol. 20, no. 2, June, 393–403.
Sociological/social-psychological approaches
29
Holmberg, S. (2007) “Partisanship Reconsidered,” in Dalton, R. J. and Klingemann, H. D. (eds.) The
Oxford Handbook of Political Behavior, New York: Oxford University Press: 557–570.
Huddy, L., Mason, L., and Aarøe, L. (2015) “Expressive Partisanship: Campaign Involvement, Political
Emotion, and Partisan Identity,” American Political Science Review, vol. 109, no. 1, February, 1–17.
Jennings, M. K. and Niemi, R. G. (1968) “The Transmission of Political Values from Parent to Child,”
American Political Science Review, vol. 62, no. 1, March, 169–184.
Lazarsfeld, P., Berelson, B., and Gaudet, H. (1944) The People’s Choice, New York: Duell, Sloane, and
Pearce.
Lewis-Beck, M., Jacoby, W. G., Norpoth, H., and Weisberg, H. F. (2008) The American Voter Revisited,
Ann Arbor: University of Michigan Press.
Lippmann, W. (1922) Public Opinion, New York: Free Press Paperbacks.