So, what do you think?
Public opinion research has a sordid history of failed efforts, followed by a period of scientific control and seriousness, and it is now degenerating into an unscientific commodity controlled by moneyed ideologues. What do people want? It’s a question that is essential for democracy. Represented officials need to know the will of the people in order to act on those preferences, or to convince the public that their preferred options are not ideal. Democracy can’t work when only the politically connected have a say in what policies are pursued–which collective goods will be produced and which bads alleviated. With public opinion, there is an air of democracy in the process. Elected officials must bow to public opinion, or else, presumably, the public will vote them out of office. Of course, if the public doesn’t know what the public will is, or is grossly misinformed about the issues the public cares about and their positions on important questions, then the public will have skewed issue salience and no sense of potential for political mobilization to generate collective goods or mitigate collective harms.
Bad public opinion research in the early and mid 20th century tended to be commodified. “Research” was conducted by “journalists” who sold the stories of what American’s think to publications, who in turn sold magazines and newspapers to people who wanted to know what other people think. The debacles of the Literary Digest’s prediction of an Alf Landon victory over FDR in 1936, and later the Chicago Time’s prediction of Dewey defeating Truman led to some circumspection, and scientific attention to the sampling process. Simple random samples with high response rates ensure that estimates from those samples will converge to the population parameter with repeated infinite sampling. Wow. How simple. You have to know that to pass most undergraduate statistics courses. Actually, “high” response rates aren’t enough. We hope that high response rates are enough. Random isn’t “people who are home at 7pm on Monday and answer their phone to answer questions from a stranger.”
Failed predictions in the mid-20th century led to decommodification of the survey research industry, and the most active research instead was taken up by social scientists at universities, particularly at Michigan, Berkeley, and Chicago, but also in many smaller research institutes. All of these had in common a fierce attention to detail and to scientific procedures for sampling and completion, and an increasing attention to question wording, reliability, and other technical issues in the field. University researchers also shared an abiding commitment to data quality for its own sake—their careers relied on the accuracy of these data, and the purpose of measuring public opinion was not to sell it to a newpaper, political group, or corporation. The data were for the scientific community, and were not for sale (though universities contributed to data sharing programs like ICPSR which furthered scientific progress).
With the advance of 24 hour television in the 1970s and 1980s, opinion polls once again became more lucrative for media organizations. At the same time, politicians came to see these methods as a means to assess chances and issues, and to try to manipulate public opinion by controlling the framing of issues. Jane Mansbridge provides a nice story regarding that during the ERA ratification campaign, when Gallup used question wording shifts to falsely suggest a surge in public opinion. The good liberal Gallup thought he was helping ERA, but instead wound up energizing the countermovement and decreasing the salience of survey research. More importantly, Gallup did what he did because he was a paid operative who was selling the opinions his customers wanted to hear. It was the beginning of the deluge of whore survey research. Public opinion as something that can be deliberately bought and sold, and manipulated. And, worse yet, at the same time we saw the beginning of the War on Education, which will serve to defund many minor survey research centers, reduce the ability of many colleges to support data sharing initiatives, and create a crisis for major research centers which took up the charge of conducting serious scientific studies of public opinion. To make matters worse, this period also saw an explosion of “Marketing Research” which tested the limits of interviewee cooperativeness and sold bogus products to companies hoping to capitalize on the identification of consumer preferences. Within 20 years after the election of Reagan, public opinion research will change radically, and will formally alter the scientific standards for data collection–in response to the preferences of powerful for-profit interests in the survey research and marketing fields. In the early 1980s, survey research textbooks would condemn to death any “study” which only received responses from 25% of the randomly targeted subjects, now a response rate of 25% is considered the “gold standard.” Jesus fuck…it gets worse…