This joint academic project between Adam Berinsky and Eric Schickler is a rehabilitation of hundreds of public opinion polls conducted from the 1930s to the early 1950s, as archived by the Roper Center for Public Opinion Research.
The majority of the archived polls were conducted by Gallup’s American Institute of Public Opinion (AIPO), but the archive also includes numerous polls by the three other major survey organizations of this era: Roper’s eponymous firm, Hadley Cantril’s Office of Public Opinion Research (OPOR), and the National Opinion Research Council (NORC).
These survey data constitute a treasure trove of information on the attitudes of the American public during a critical era that stretched from the waning years of the Great Depression to the first stirrings of the Civil Rights Movement.
Yet despite its scholarly value, until recently this unique data source remained largely untapped, for two main reasons: early public opinion polls in the Roper archive were not only difficult to analyze, but the samples were also, for intentional and unintentional reasons, unrepresentative of the US adult population. Since pollsters were drawing conclusions about the American electorate, populations such as women, low-income, and Black Americans, particularly in the South, went unsampled.
The Berinsky & Schickler American Mass Public Study addresses this by creating several sets of respondent-specific weights designed to adjust for sampling and nonresponse bias. These weights are calibrated such that key demographic characteristics match their distribution in the target population.
These weights allow for these data to yield more accurate estimates of social and political attributes of interest in the US population. In Target Estimation and Adjustment Weighting for Survey Nonresponse and Sampling Bias (published by Cambridge University Press in 2020), Berinsky and his co-authors describe how to analyze the data now available through this project.
Berinsky is also the author of a 2006 paper, American Public Opinion in the 1930s and 1940s: The Analysis of Quota-Controlled Sample Survey Data (published in Public Opinion Quarterly), that describes strategies to properly analyze the public opinion data of the 1930s and 1940s.
For more information, please visit the Roper Center page on this project.