Converse's Uncomfortable Finding

A survey, a codebook, and a surprise

In 1956, interviewers from the University of Michigan fanned out across the country with a long questionnaire. They asked ordinary Americans what they liked and disliked about the two parties and the two presidential candidates, Dwight Eisenhower and Adlai Stevenson, and wrote the answers down in the respondents' own words. A political scientist named Philip Converse later sat down with those transcripts and asked a simple question: how many of these people think about politics in terms of a liberal-conservative spectrum?

His answer, published in 1964 as "The Nature of Belief Systems in Mass Publics," became one of the most cited and most resisted findings in the discipline. In the 1956 sample, only about 2.5 percent of voters reasoned as full "ideologues," people who used abstract concepts like liberalism and conservatism and applied them consistently. Roughly another 9 percent qualified as "near-ideologues," who used the terms but loosely. The great majority evaluated parties and candidates in other ways entirely: by what the parties did for groups they belonged to (good for farmers, for the working man, for business), by the "nature of the times" (prosperity or depression, war or peace), or by impressions of the candidates with no policy content at all.

It is easy to read this as condescension, and some did. Converse was not saying that ordinary voters were foolish. He was saying something more specific and, for the purposes of this series, more important: the left-right line, which intellectuals had adopted in the 1920s and 1930s, was simply not the instrument most citizens were using. Their views on different issues did not hang together the way the spectrum predicted. A respondent might favor federal aid to schools and oppose federal involvement in employment, or the reverse, and feel no tension, because the tension exists only if one first assumes the issues belong on a common axis.

Converse called this low "constraint": knowing one of a person's positions did not let you predict the others, except among the small, highly engaged minority. The line fit the elite. It did not fit the public.

Sixty years later

A reasonable objection is that 1956 was a long time ago. Education levels have risen, political information is everywhere, and the parties have become far more distinct. Surely citizens have caught up with the spectrum.

Donald Kinder and Nathan Kalmoe tested exactly this in Neither Liberal nor Conservative: Ideological Innocence in the American Public (University of Chicago Press, 2017). Using decades of survey data, they found that what Converse called ideological innocence persists. Most Americans still do not organize their opinions along a single dimension, and for most citizens the self-applied label "liberal" or "conservative" is a weak predictor of what they actually want government to do. The people for whom the labels work well remain the politically attentive few: the activists, donors, and heavy consumers of political news, who are also the people most likely to write about politics and therefore most likely to assume that everyone thinks as they do.

Verlan and Hyrum Lewis, in The Myth of Left and Right, treat this body of survey research as one of the three pillars of their argument: issue positions cluster only among the most engaged, which is exactly what one would expect if the clustering is produced by social reinforcement rather than by philosophy.

Why is the finding uncomfortable? Because the apparatus of American political commentary is built on the assumption it undermines. Pundits describe the electorate as "moving right" or "moving left." Campaigns position candidates at points on the line. Pollsters report the share of the public that calls itself conservative as though that told us something about policy. If Converse and his successors are correct, all of this describes a map that most of the territory has never been on.

It is uncomfortable in a second way as well. The citizens who do not reason ideologically are not thereby without views. They have views, often strong ones, on the specific matters that touch their lives: a factory closing, a school, a war, a price at the pump. What they lack is a reason to bundle those views in the particular way the two teams bundle them. When the election offers only the two bundles, those citizens must either pick the one that is less wrong for them or stay home. Neither choice measures what they believe.

Where the evidence pushes back

Critics of Converse have made several durable points. Some argue that his coding scheme undercounted ideological thinking by demanding a vocabulary rather than a structure; a voter can hold consistent positions without using the word "conservative." Others note that constraint rose after the 1960s as the parties diverged and gave voters clearer cues, and Matthew Levendusky's The Partisan Sort (2009) shows that voters who identify with a party now hold positions far more aligned with that party than their counterparts did in 1956. The fairest reading is that the public has learned which team holds which positions, and increasingly adopts them, without the underlying philosophical reasoning having spread much. That is consistent with Converse and with the social theory, but it does mean today's partisans are more predictable than their grandparents were.

Why this matters for the ballot

If most citizens are not on the line, then a ballot that offers only two points on the line is not representing them; it is asking them to translate themselves into a language they do not speak. Opening access for independents and third parties lets some of those untranslated views appear on the ballot in their own words.

— Michael W. Cunningham, Ph.D.

Cover image: "Ceiling of Library of Congress" via Wikimedia Commons. See the file page for author and license.