Survey researchers/askers investigating public opinion have never agreed “`on what the public is and on what public opinion is'”. That’s a conclusion of the Second International Conference on Public Opinion Research, 1947 (Rogers, The Pollsters, p.11) and here’s a 2008 comment from The Sage Handbook of Public Opinion Research indicating that survey researchers remain in the same fix and that neither askers themselves nor their high-status publishers (in this instance, Sage, Macmillan & Free Press, and Princeton) have no shame in admitting it: “PUBLIC OPINION–A NEBULOUS CONCEPT In 1968, W. Phillips Davidson wrote the entry for `public opinion’ for the International Encyclopedia of the Social Sciences, concluding that ‘there is no generally accepted definition of the term (p. 188). Three years earlier Harwood Childs (Public Opinion, 1965) had described no less than 48 different definitions of public opinion in the relevant research literature. Since then, the definitional situation has certainly not become any clearer. . . . Public opinion continues to be one of the fuzziest terms in the social sciences” (p. 1).
Fuzzy?? Let’s be done with fuzzy. Advancement in social science and, thus, optimum problem solving and social betterment requires replacing fuzzy terms with operational definitions. Moreover, we need to remember that survey research cannot produce reliable information about public opinion. All you get from survey research/asking investigations of public opinion are what respondents say are their opinions and as we all know, and should never forget: what people say may, or may not, correspond to their actually held opinions.
Reliable information about opinions actually held by individuals and groups of people can be obtained by the use of at least two, preferably three or more, non-asking sources; such as observation, content analysis, and experiments. Triangulation is required. See also my discussion of what I call “proper” methods of data collection and proper research designs in Part 6 of my book, The Problem with Survey Research.