What We Know about Americans’ Voting Behavior
by Stephen Earl Bennett


I am fascinated by some of the observations made about how people vote. One, for example, illustrates a misunderstanding of voting behavior that crops up in comments about elections.

It goes something like this: Assume that one of the major political parties has drifted too far from the ideological center to be successful in national presidential elections. Were that party to shift toward the center – so the argument goes – it would lose as many voters, presumably those on the ideological fringes, as it would gain, again, presumably those at or around the center. This analysis assumes that large portions of the electorate are ideologically consistent and motivated.

Statements like this would not fare well in the voting behavior literature, which is huge and has had much to say about Americans’ voting behavior.

The Columbia school of voting studies

The empirical study of U.S. voting behavior goes back to the 1940s. The first studies, utilizing the then-new technique of survey research, were conducted by Columbia University sociologists headed by Paul Lazarsfeld. He and his co-authors produced two books.

The first, The People’s Choice, appeared in 1944. Based on at least seven interviews with residents of every fourth house in Erie County, which is in the north-central region of Ohio, Lazarsfeld and his co-authors made several claims that were considered path-breaking at the time. The first was that most voters had already made a choice before the campaign began. Second, most people got their campaign-related information from opinion leaders who obtained information about the campaign by reading newspapers and then passed it along in subsequent conversations. Third, socioeconomic status, and to a lesser degree, religion, affected how people voted. Blue-collar workers voted Democratic, while the middle-class and the well-to-do backed the GOP. Catholics supported FDR and the Democrats; Protestants backed Wendell Willkie and the Republicans. Erie County, Ohio, is north of the Mason-Dixon line.

Lazarsfeld and his colleagues conducted a second study of presidential voting in 1948. This time the researchers conducted a four-wave survey in Elmira, New York, a city of roughly 50,000 with a diverse ethnic population. The research led to Voting, published in 1954.

Although considered to be a more sophisticated sociological analysis of how the ordinary voter reached a decision, or decided whether to go to the polls in the first place, Voting’s portrait of the American voter was not drastically different from that of The People’s Choice. Voting’s main findings were that (1) the family is the main source of partisanship; (2) the vote decision occurs within a politically homogeneous group; and (3) the mass media are much more likely to reinforce pre-existing views than to change opinions.

The People’s Choice
and Voting are known as the sociological perspective on voting behavior. Lazarsfeld and his colleagues are labeled the Columbia School of voting behavior research. Unfortunately, Lazarsfeld and his colleague would produce no further studies of voting in presidential elections.

The Michigan school of voting studies

The Columbia School was soon out-stripped in terms of impact on the field by scholars at the University of Michigan. Headed by Angus Campbell, the Michigan researchers would publish several articles and four books on voting behavior in the 1950s and 1960s: The People Elect a President (1952), The Voter Decides (1954), The American Voter (1960), and Elections and the Political Order (1966). Members of the Michigan group would continue to conduct research on American voting behavior over several four-year election cycles in the future. Even so, The American Voter would have the most important impact on our understanding of voting behavior.

The Michigan School’s analyses differed from those by the Columbia School in terms of methodology and substance. Unlike Lazarsfeld and his associates, the Michigan studies were limited to a pre-election survey and a post-election wave. Also unlike the Columbia group, the Michigan team conducted nation-wide surveys, which would become the usual method by which data on ordinary people’s voting were gathered.

The Michigan School’s theoretical contributions to the field are known as the psychological approach to voting behavior. Campbell and his associates asserted that the typical voter’s choice once inside the voting booth rested on long- and short-term psychological factors.

The long-term factor shaping the voting decision is party identification, which is simply an individual’s psychological identification with one of the major political parties. Almost everyone manifested some degree of psychological affiliation with a party, a disposition that was formed early in life, shaped largely by one’s parents. For most people, most of the time, partisan affiliation persists over the life-time.

Several factors act as short-term forces influencing the voter’s decision: opinions about policy issues, assessments of the candidates, views of important social groups, and assessments of the political parties “as managers of government.” Although not totally determined by partisanship, each of these short-term factors is influenced by party identification. Furthermore, short-term forces tend to be changeable from one election to the next.

The Michigan team sketched a relatively dim portrait of the ordinary citizen. He or she was said to be largely indifferent to public affairs, relatively poorly informed, expressing inconsistent political opinions, and almost entirely non-ideological. Moreover, the less partisan the individual, the more he or she was to be politically apathetic, ignorant, and incoherent in political outlook. In an essay published in 1964, Philip Converse, one of the Michigan team, extended and elaborated upon these findings.

Challenges to the Michigan school

Challenges to the Michigan school’s analyses were not long in coming. In 1966, for example, V. O. Key, Jr.’s The Responsible Electorate drew upon Gallup polls from 1936 to 1960. Key divided the electorate into three categories: (1) “new” voters, who had not previously cast ballots; (2) “stand-patters,” who voted for the same party across elections; and (3) “switchers,” who voted for one party in election #1, but changed to the opposition in election #2. Arguing that “voters are not fools,” he believed the electorate was “an echo chamber,” resonating with the quality of information stemming from politicians. If pols fed Jane and John Q. Public pap, Ms. and Mr. Public spouted pap. At bottom, however, Key contended that if people thought the incumbent party had performed well, it was supported at the polls. If not, the opposition deserved a chance to govern.

Key’s concepts needed sharpening, and his data wouldn’t hold the weight he put on them. His thesis would be revised in 1981 by Morris Fiorina (Retrospective Voting in American National Elections), who developed the notion of “retrospective voting” that also drew from Anthony Downs’s An Economic Theory of Democracy (1957). Fiorina accepted the finding that the public tended to be indifferently informed, but argued that, nevertheless, people were able to assess how well things had been going during a party’s tenure in office. Building on Key’s assertions, Fiorina contended that if voters judged a party’s performance poorly, they turned to the opposition on Election Day. Should the retrospective judgment be positive, the incumbent party would be retained in power.

Other researchers attacked the Michigan school’s findings on different grounds. In The Changing American Voter (rev. ed., 1976), for example, Norman Nie, Sidney Verba, and John Petrocik argued that the Michigan team’s assessments of the American voter were applicable to the relatively quiescent 1950s, but out-of-date for the increasingly ideologically-charged 1960s. Nie and his associates claimed that evidence from 1964 and thereafter showed Americans’ political opinions were much more consistent, and voters displayed greater rationality inside the voting booth.

Nie and his colleagues’ findings quickly foundered on the shoals of methodology. Several researchers asserted that changes in the way the Michigan school asked questions on its surveys under-girded Nie and his associates’ findings. Once changes in question-wording were taken into account, Nie and his co-authors’ critics claimed, the post-1950s American voter looked pretty much like the one described in The American Voter.

A combination of more recent elections and especially the advent of more sophisticated statistical techniques, led many scholars to test the Michigan school’s arguments about the causes of Americans’ voting behavior. Some regard The American Voter Revisited (2008), by Michael Lewis-Beck and three others as the epitome of such research. Other studies sometimes employed data gathering techniques in addition to surveys. The general result was support for The American Voter.


Where are we? Although the corpus of subsequent research on American voting behavior is huge, the Michigan School’s characterization of American voters and of voting behavior has stood the test of time.

Although The American Voter appeared over half-a-century ago, the portrait its authors drew of the American electorate holds up pretty well. Not only is the typical citizen only marginally interested in and attentive to public affairs, he or she is poorly informed, expresses mostly inconsistent political opinions, and is mostly non-ideological. When Jane or John Q. Public enters the polling booth, she or he is influenced mostly by a psychological identification with one of the major political parties, and to a lesser extent, by how she or he feels about the candidates, the parties’ records, and a few important issues, especially those impinging on pocketbooks.

Assumptions that ideology plays an important role in most people’s minds are mistaken.

Before closing, we need to ask if the Michigan School’s portrait of the American voter can accommodate Donald Trump’s emergence as the Republican Party’s presidential nominee in 2016. Yes, if we paid attention to recent polls showing large portions of Republican Party identifiers have been discontented with the GOP’s leaders. A Rasmussen Reports poll of likely voters in mid-2016, for example, found that 73% of likely Republican voters said their party’s leaders had lost touch with GOP voters, and just 20% believed Republican leaders were doing a good job representing the party’s electorate. By contrast 62% of likely Democratic voters thought their party’s leaders had done a good job representing its voters, while only 28% alleged the party’s leaders had lost touch with Democratic voters.

Other polls from the years immediately before the GOP’s 2016 national nominating convention should have alerted us to the possibility that Trump would emerge victorious, without having to jettison the Michigan model of voting.

Stephen Earl Bennett is Emeritus Professor of Political Science at the University of Cincinnati, where he was employed between 1969 and 1999. His Ph.D. is from the University of Illinois at Urbana/Champaign. From 2003 to 2008, he was Adjunct Professor of Political Science at the University of Southern Indiana. Before pursuing the Ph.D., he taught Social Studies for two years at Nicolet High School in a suburb of Milwaukee. Bennett and his wife, who is president of the University of Southern Indiana, live in Evansville, Indiana.

Guest Voice
Click here for reuse options!
Copyright 2017 The Moderate Voice