The records of the 1948 election voting study in Elmira, New York, include the documentation underlying the book Voting: A Study of Opinion Formation in a Presidential Campaign, co-authored by Bernard R. Berelson, Paul F. Lazarsfeld, and William N. McPhee. The authors positioned their study as one of a series conducted by various scholars in the context of different elections from 1940 to 1952. Their overall aim was to attain a "better understanding of the processes of democratic elections," particularly in terms of how citizens make up their minds as to how to vote (Voting, vii). The 1948 presidential election pitted the Democratic Party candidate and incumbent U.S. President Harry S. Truman against the Republican Party nominee and New York Governor Thomas E. Dewey. But the primary and general campaigns were also notable for the range of other viable candidates or potential candidates from across the political spectrum, from Henry A. Wallace on the left to Strom Thurmond on the right. Further, this was the first presidential election since the end of World War II, at a time of great changes domestically and internationally, and the first since the early 1930s without Franklin D. Roosevelt at the head of the Democratic ticket. Throughout the election season, Truman was viewed as such a vulnerable candidate that the mass media conveyed the expectation that Dewey would defeat him, yet Truman won, elected by 2.5 million votes. It was in this volatile political environment that Bernard Berelson and his team launched their study of voters and voting.

Berelson acquired his Ph.D. in 1941 from the Graduate Library School at the University of Chicago. His thesis focused on the mass media's effect on the formation of public political opinion. In 1944, he joined the Bureau of Applied Social Research at Columbia University, where its founder, Paul Lazarsfeld, was also interested in the power of mass media and its influence on public opinion. In 1944, Lazarsfeld and Berelson's work The People's Choice was published, a study of the 1940 election, using Erie County, Ohio, as the site of analysis. Berelson returned to Chicago in 1946, but kept in contact with Lazarsfeld, eventually planning a study of the upcoming 1948 election.

Berelson and Lazarsfeld chose Elmira (including the surrounding unincorporated towns) as the site of their 1948 study because it represented a "normal" community from which generalizations might be drawn. Specifically, Elmira was of moderate population size; independent but not isolated from any metropolitan district; socially and economically stable; had good media exposure with a typical educational and cultural environment; had a balanced labor and industrial situation; was politically competitive; and had a typical ethnic composition. They divided a map of Elmira into 816 segments, numbering them in a serpentine fashion, and selected every third segment, yielding 271 segments to visit. Within the segments, all households (about 6,000) were noted with the final sample of 1,267 households chosen from those. Interviews were then attempted, with a final sample size of respondents totaling 1,029.

Berelson and Lazarsfeld decided to employ the so-called "panel method" in Elmira by which the same individuals were interviewed at multiple points over time in order to assess the dynamics of their decision making process. Accordingly, the study team implemented four "waves" of questionnaires in 1948: in June, before the party nominees were decided at the July conventions; in August, after the conventions and before the campaigns' fall push; in October, as the election drew near; and in November, right after the election. (The collection also holds questionnaires from 1950, but Voting does not seem to refer to these and it is unclear what their relationship to the initial study is.) Not all members of the panel participated in all four waves for various reasons, but 746 (72%) did. 907 (88%) participated in three of the four waves. Only 43 (4%) participated in June and then dropped out.

Subsequent to the fieldwork, William N. McPhee, who was a sociologist and invented a computer simulation model for voting, joined the Elmira study team. McPhee used a computer to help analyze the results of the survey and he wrote several chapters of Voting, which was published in 1954, six years after the election.

(The principal sources for this note were Voting: A Study of Opinion Formation in a Presidential Campaign, documents in the collection and Wikipedia.)