You’ve never seen it. You can’t touch it. But everyone knows it exists. The very definition of a social construction. One of the biggest weaknesses the study of public opinion has suffered from over the years has been the conflation of this essential constructivism with a positivist attempt to render the ideational material, to observe the inherently unobservable. Public opinion doesn’t exist in the positivist sense. Consequently it cannot be studied using tools designed for observation of things which do exist.
Opinion polls are just such tools. They were first used on a mass scale by George Gallup to (successfully) predict the outcome of the 1936 US Presidential Election. Since then polling has evolved into an entire industry, but the basic method remains the same. A survey is taken of the voting intentions of a representative sample of the population, and the principles of statistical inference used to estimate the reliability of the results as an estimate of the entire population’s future behaviour. The problem with polling, however, is that its use has been extended beyond the measurement of ‘real’ phenomena, such as voting, to gauge more ethereal concepts such as attitudes and values and opinions. The problem with this extension is that it moves away from studying something that exists ‘out there’ towards studying something that doesn’t. And since it’s not possible to study something that doesn’t exist, what polls looking at questions other than voting intention actually do is create, rather than observe, a distribution of views in response to a question. They thus fail the test of scientific validity, that regardless of what method of observation you use, you must begin by trying to observe what you think you are observing.
Here’s how it works. A pollster asks a member of the public a question on the political issue of the day. They get a response. They assume the respondent’s opinion matches the response given. They fail to account for the extent to which the respondent actually had a view on the subject before being prompted to express one. Most people, asked a straightforward question about an issue they have seen on the news or in the papers, will be able to give an answer. This does not mean that they had an opinion before they were asked for one. What polls tell us is not what the public is thinking about an issue, but how it would respond if asked the same question about that issue en masse. Fine for predicting voting behaviour; the question is going to be asked on election day. Not fine for other issues; what does it matter if 55% of the public would agree, if asked, with statement A, if they are never going to be asked? It’s like saying 100% of the public would be killed, if they were all dropped from 30,000 feet onto a pointed stick. They’re not going to be. So who cares?
This is why I don’t care what you think.
Picture credit: Wikipedia user “RadioFan”, http://en.wikipedia.org/wiki/User:RadioFan