In 1947, the American public was canvassed for its opinion on the 'Metallic Metals Act'. As is usual with opinion polls, there was a sizeable minority of 'don't knows', with about 70% of people offering a definite response in favour or against. What was unusual was that its subject was a fabrication. There was no 'Metallic Metals Act' (a metallic metal being like an ironic irony, or a tautological tautology). Of course, opinion surveys aren't bogus. They do very effectively canvas responses to simple poll questions. If they select a representative sample at random, and appropriately weight the results, the responses will closely match those for the public as a whole. And when the question is something straightforward like 'how do you intend to vote in the upcoming election?', quantitative surveys have some predictive ability, notwithstanding the periodic difficulties which we will come back to.

On the other hand, 'public opinion' is a mirage. The very phrase implies a general social will, and the methodology implies that it is composed of individuated, definite, latent 'opinions' which will then determine behaviour. The mirage of public opinion is invoked typically to provide a pseudo-mandate for policies that elites have already decided to implement. In this respect, polls often conceal and suppress far more than they reveal. Pierre Bourdieu, whom I cited in a previous post, put it like this:

Public opinion is an "artifact, pure and simple, the function of which is to dissemble that the state of opinion at any given moment is a system of forces and tensions and that nothing is more inadequate for representing the state of opinion than a percentage".

The practise of opinion polling, says Bourdieu, rests on a number of false assumptions. In the first place, it assumes that those answering have the same range of competences as those who interpret the polls - politicos, commentators, etc. This is not usually the case. The example given at the start of this article shows that people often feel pressured in such interview situations to have an 'opinion' on something they haven't really thought about. In that instance, they will try to base their opinion on whatever information they have gained from the very mass media that is polling them. The large numbers of "don't knows" and "not applicables" offered by respondents in polls show that many people have a realistic sense that they simply can't answer the question as put to them, but these responses are usually set aside or just factored out of surveys to produce this artifact known as "public opinion".

Secondly, Bourdieu says, it assumes that all opinions have some sort of equivalent value and are formed in a similar, individualised way. But not all opinions have the same weight or conviction behind them, and they are formed in circumstances that don't mimic the isolated, individualised survey interviewing situation. Most people form their opinions on something by talking it through with others. The context in which such discussion takes place means that actual opinion is usually formed through group experiences, such as those arranged by class, race, religion, nationality and gender. In this sense, the collective logically precedes the individual: rather than some general public will being derived form individual units of 'opinion', individual ideas are usually derived from collective experience and ideas. But the practise of opinion polling does not reproduce this aspect of the way ideas are formed. These are some of the reasons why there are consistent problems with even the more reliably predictive election polling. Pollsters tend to blame discrepancies on voters misrepresenting their views. But it is just as plausible that the individualised context in which polling questions are put forces people to rely on their impressions of successful media campaigns, whereas their actual decisions as to how they will vote will take place after argument, reflection and a certain amount of discounting for media hype. The brief 2010 'Cleggasm' was arguably a pure spectacle phenomenon with little real traction in the population, but pollsters have tended to blame the fake Lib Dem surge on voters deliberately concealing their true opinions ('shy Labour voter' etc).

This is - of course - not to say that all such surveys are useless. I do recommend a careful, scrupulous, discriminating use of opinion surveys to assess the prevailing view on various political issues. But this should be done bearing in mind what polls are for. They polling industry is not an extension of the democratic process, but a quite deliberate attempt to subvert democratic reasoning and debate, to 'settle' important issues before they have been discussed, to bully opponents and put them on the back foot etc. Polls are taken to provide elites with raw materials the better to manipulate us with, not so that they can respond to our genuine concerns. Further, taken collectively they tend to disclose a set of apparently contradictory dispositions, ideas, and values, suggesting that the pseudo-finality of "a percentage" is indeed hopelessly inadequate as a register of the real ideas that people have. So, if majorities appear to support you on an issue, this may mean nothing more than that they are open to your argument, and that there might be an overlap between their priorities and yours. Similarly, if a poll says the majority opposes your position, it does not mean that they really do, and that if they really do, they are particularly settled in their view, or set much store by it, or will allow it to dictate their behaviour. There are undoubtedly issues on which clear majorities have a clear view one way or the other, and the polling evidence on such matters should not be ignored or discounted. But there is no 'public opinion', and any idea that opinion polls could by themselves provide a 'mandate' for certain policies has to be dispensed with. Polling is a weapon in struggle, and it is a weapon that is largely in the hands of the ruling class.