2021-007 | Download PDF
Campbell, W. Joseph. Lost in a Gallup: Polling Failure in U.S. Presidential Elections. Berkeley: University of California Press, 2020, 336 pp., $29.95 (hardcover). Reviewed by Michael J. Socolow, University of Maine (email@example.com)
It happened again in 2020. Just as in 2016, 2004, 1980, and most famously in 1948, prominent national polls published during a presidential campaign failed to precisely predict the ensuing outcome. These recurring election surprises regularly ignite public discussion about the accuracy of surveys, the ethical issues involved in publishing polls that might influence (and misinform) the American electorate, and whether pollsters are scientists or magicians.
W. Joseph Campbell’s Lost in a Gallup offers a readable and engaging selective history of polling failure. The book chronicles the role polls and pollsters play in structuring presidential campaign narratives. Campbell describes a repetitive pattern in U.S. media history where pollsters publish findings, the media then amplifies this data – that seemingly carries the patina of scientific certainty – and the relative positions of the presidential candidates become widely accepted. Then election day arrives, and voter behavior fails to correlate with pre-election polling. These surprising outcomes then incite a new cycle of criticism, reflection, and explanation. Lost in a Gallup closely analyzes these secondary debates, with a focus upon the tensions emerging between journalists and pollsters in the aftermath of failed predictions.
Lost in a Gallup is not concerned with detailing the minutiae of quantitative survey methodology across U.S. political history. Rather, the book is a reconstruction of the historical conflict “between anecdote-based reporting and data-based methods of information gathering; a tension between qualitative and quantitative methods of assessing public opinion” (p. 25). Continually throughout the book, famous political journalists, columnists, and broadcasters lament the growth in popularity of polling as reportage, and its encroachment on the space established for traditional “shoe-leather reporting” in newspapers and broadcasts.
Campbell’s study combines a review of these historical (and contemporary) tensions with a fascinating group biography of such historical figures in U.S. polling as George Gallup, Elmo Roper, Archibald Crossley, and Warren Mitofsky. The biographies are interwoven in a series of chronological case studies reviewing the ways polls influenced national media, and perhaps the U.S. public, during several (but not all) presidential contests over the last eight decades.
Campbell’s book is well-written, impressively researched, and detailed. In referencing primary source material from numerous archival collections, Lost in a Gallup contrasts public debates over polling with internal discussions within the polling community. In a sense, early pollsters like Roper and Gallup, and later ones like John Zogby and Mitofsky, were forced to become supportive competitors, as the industry they constructed needed to proclaim certainty that they suspected could prove elusive. Lost in a Gallup comes alive as Campbell relays the internal critiques, jealousies, evaluations, and self-assessments of the pollsters. The study reveals fascinating backstage communication hidden from both the public and media clients purchasing polls as a journalistic product. Even while publicly proclaiming methodological rigor, pollsters could be candid with each other about the spectrum of possible response and sampling errors that might bias their results.
Lost in a Gallup, in this sense, does a fine job of exposing the fact that these technicians – magically capable of measuring that amorphous and ethereal feeling known as “public opinion” – were neither as capable nor as skilled as they seemed. Skeptics always abounded – whether columnists like Mike Royko of the Chicago Daily News, Jimmy Breslin of Newsday, or scholars like Lindsay Rogers, who published a prescient and trenchant critique of polling in 1949. While Campbell chronicles the dialectic between pollsters and skeptics, it would appear that the historical trajectory shaping public discussion of quadrennial presidential elections has largely moved in one direction. In today’s media universe, there exist far fewer columnists with the influence of a Royko or Breslin, while social media and a mania for metrics and data have made the latest generation of pollsters – as embodied by FiveThirtyEight’s Nate Silver – as influential as any of the old-time newspaper columnists.
Lost in a Gallup occasionally covers well-trod ground. No history of polling failure can avoid either the infamous 1936 Literary Digest survey prophesizing Alf Landon’s victory, or the legendary “Dewey Defeats Truman” 1948 story, and both of these familiar cases are reviewed in detail. By now, the cycle Campbell describes seems routine. Polls issue a portrayal of a contest that may or may not accord with reality, and elections either ratify or demolish the findings. When things go wrong, pollsters retreat, assess the situation, and then return to repair the reputation damage. For some reason, the media, and much of the American public, seems to forgive and forget these errors, and trust the polls when the next election day nears. Lost in a Gallup provides a useful reminder for 2024 and beyond, and one hopes it’s referenced by journalists, critics, and scholars in the future. Should polls once again prepare Americans for an outcome that fails to materialize, we can’t say we weren’t warned.