From an academic friend who knows a very great deal indeed about polls, voting, and public opinion (as in: he’s written books about them).
There are two ways to find out what people think, believe, want, and like: (1) Ask them, or (2) Observe their behavior.
If one had to choose the better way to understand some phenomenon in the social sciences, one would study behavior, not attitudes. Behavior talks: poll responses walk. And hard data on behavior are everywhere.
Historically, studies of public opinion were only about behavior. Even the study of “attitude” only emerged in the 1930s. Modern-type surveys did not even exist much before that. Public thinking was inferred from voting, newspaper stories, personal traits like religion or occupation and, of course, actual behavior. Chasing the tax collector out of town was a sure sign of unhappiness over taxes.
The academic Left has become infatuated with surveys since they control the questions and the interpretations, and can release the data as it suits them. Polls are putty in the hands of those wishing to make points not otherwise discernible. That the enterprise is draped in “science” and technical jargon settles the debate. So ignore actual tax avoidance and focus instead on what people say about paying taxes (they love the tax collector and crave more social welfare … yet all the while are cheating). Racial integration, for example, now becomes opinions about racial integration; and with the “right” data treatments, reality is no match for the skilled investigator. An ample literature exists demonstrating a weak link between attitudes and behavior. If you believe polls, nobody in America watches porn.
These observations are hardly novel. I’ve written about it all in my book Polling, Policy and Public Opinion. Interestingly, conservatives are regularly hoodwinked by poll results, though they insist the questions were loaded. They have it wrong. Verbal reality and behavioral reality are fundamentally different — witness liberal whites fleeing blacks when they move nearby.
Talk is cheap. I’ve tried to tell conservatives about this structural dishonesty of polling but they just don’t get it. They’re addicted to sound bites about liberal bias and lack any interest in technical details.
[Me] I don’t know that conservatives are any more averse to data, evidence, numbers, and science, than are liberals. I guess religionism throws a bigger wrench in the works among cons than among libs, giving a stronger bias towards magical thinking and reality-denial; but most people, of all political and confessional persuasions, seem to be able to ignore or reject even the “hardest” data if it makes them uncomfortable. I’d put myself at about the 99th percentile in data-orientedness, yet I catch myself reaching for the ignore/reject button sometimes. We are poor muddled creatures.