There is no such thing as public opinion

"How much attention did you pay to this week’s Federal Budget?" For many respondents to this week’s Essential Research Poll, the answer was not much — 44 per cent said that they paid little or no attention to the budget. But in the same survey, 80 per cent were able to express an opinion about whether the budget would be good for them and people like them. At Larvatus Prodeo Mark Bahnisch writes:

I’ve always thought that almost all polling should ask questions about the degree to which respondents are interested in particular issues, and how strongly their stated preferences are held. We’d get a much more realistic picture of public opinion, that way. Of course, we’d also get one which would reduce a lot of the noise and fury of the media narrative to total irrelevance.

Mark’s got a point. In his 1992 book The Nature and Origins of Mass Opinion, political scientist John Zaller writes that "citizens do not typically carry around in their heads fixed attitudes on every issue on which a pollster may happen to inquire; rather, they construct ‘opinion statements’ on the fly as they confront each new issue." When people construct these opinion statements they "make greatest use of ideas that are, for one reason or another, most immediately salient to them — at the ‘top of the head’".

According to Zaller, most people pay little attention to politics. But nevertheless, they are able to provide the responses pollsters need in order to make newsworthy statements about what the public believe, what they want and what they are likely to do when the next election comes around. The reason they (or perhaps I should say ‘we’) are able to answer the questions is because they are exposed to a stream of news and information through the media. Zaller writes:

… most people on most issues are relatively uncritical about the ideas they internalize. In consequence, they fill up their minds with large stores of only partially consistent ideas, arguments and considerations. When asked a survey question, they call to mind as many of these ideas as are immediately accessible in memory and use them to make choices among the options offered to them. But they make these choices in great haste — typically on the basis of the one or perhaps two considerations that happen to be at the ‘top of the head’ at the moment of response (p 36).

Because of this, responses to similar sounding questions can be remarkably unstable. The ‘considerations’ respondents call to mind are often talking points in favour of one position of another — points made by subject experts, activists, politicians and commentators which are packaged into bite-sized pieces and reproduced in the media. When people have little interest in an issue, they make little effort to integrate these considerations. This is why Zaller argues that they are not ‘true attitudes’.

When responding to a pollster’s questions, changes in question wording or question order can affect which considerations come to mind. And at different times, different considerations are more likely to be ‘top of mind’. For example, it may be that ‘public opinion’ on climate change appeared to shift simply because talking points critical of climate science became much more available in the media.

The thing that really matters isn’t how people respond to question in a poll today. It’s how they will respond politically in the future. In his essay ‘Public opinion does not exist’, French theorist Pierre Bourdieu wrote that:

It is known that opinion polls are very bad a detecting the latent state of opinion and, more precisely, the movement of opinion. This is because the situation in which they grasp opinions is entirely artificial. In the situations in which opinion is constituted, and especially in crisis situations, people are faced with constituted opinions, opinions supported by groups, so that choosing between opinions clearly means choosing between groups. That is the principle of the politicizing effect of a crisis: one has to choose between groups which define themselves politically, and to take more and more positions on the basis of explicitly politically principles. In fact, what seems to me to be important is that opinion polls treat public opinion as a simple sum of individual opinions, collected in a situation which is ultimately that of the polling booth, where the individual secretively expresses an isolated opinion in isolation. In real situations, opinions are forces and relations between opinions are power relations between groups.

This is remarkably similar to Zaller’s position. In a paper on the concept of latent opinion, Zaller argues that skilled politicians will ignore opinion polls when they know currently unpopular measures will lead to popular results. The trick is to predict how voters will respond in a changed set of circumstances.

As Mark suggests, the media narrative built up around poll results may be nothing more than "a tale told by an idiot, full of sound and fury, and signifying nothing." ‘Public opinion’ is often little more than the echo of conflicts between rival elites.

This entry was posted in Uncategorized. Bookmark the permalink.
Notify of
Newest Most Voted
Inline Feedbacks
View all comments
Graham Young
11 years ago

Interesting Don, but from a political practitioner’s perspective I think it is a bit unconvincing. One simple point about the utility of polls. If your final sentence is correct, then a public opinion poll could be a handy score card of how the elites are tracking, so signifying much more than sound or fury.

But this assumes that the public are driven by politicians rather than politicians being driven by the public. While an important political skill is to know when to go against public opinion, another important political skill is to learn how to swim with it. The sensible use of polling for a politician is to work out what the public thinks and how to prioritise and argue so as to align the public with the politician’s point of view or at least craft a position which is acceptable to both.

In reality the public reacts not so much to intellectual cues which are thrown at them by politicians through the media, but to emotional cues which they absorb through their temperament, upbringing and environment. When I ask voters what they think it is certainly driven by the agenda or public discourse at the time, but it is also driven by deep narrative themes that have nothing to do with that.

So what they tell me is a mix of influences, of which “rival elites” is a very small percentage. Take climate change as an example, which appeared more or less out of nowhere in 2007. It spiked, partly in response to An Inconvenient Truth, but spiked higher in Australia than elsewhere. The reason, from what I can discern, is that we were experiencing a once in a hundred year drought.

My proof for this is the strong correlation between mentions of climate change and drought in the series of polls that I took in 2007 and 2008. Not definitive, but persuasive.

Climate change gave voters the illusion that they could do something about this drought, so it became a potent weapon for the political side aligned most strongly with it.

The dams are full to fullish, and climate change was slipping back down the rankings, even before Copenhagen and the subsequent government back-down putting us much closer to the position of most other countries where climate change has never been the focus of an election campaign.


[…] There is no such thing as public opinion ( […]

Graham Young
11 years ago

Yes, a lot of opinion polling is useless because it doesn’t understand its limitations, but that doesn’t mean there isn’t such a thing as public opinion. One of the reasons I do the polling that I do, and why it has pretty good predictive power, is that I am engaging with people who think more than the average about political issues, but who still have more characteristics of the average than than the elite. So they give you feedback that is similar to the feedback you will get from the whole community when they are put in a position of having to make a real, rather than hypothetical, decision.

When I saw your headline I thought perhaps it would have been an investigation of the way that we characterise the public as having an opinion when in fact what we are talking about is an aggregation of opinions that often have little more than the support of 50% of the population.

11 years ago

Hi Don and Graham

Very interesting. I think ‘public opinion’ is different to polling on specific issues. How so:

1. Who might I trust, not with an issue, but with the Prime Ministership;

2. Is there a compelling reason to change that might override the other issues, I would say WorkChoices and Climate Change in 2007;

3. If both parties are competent, then maybe I can afford to take a punt (2007);

4. Has one party disqualified itself by being massively incompetent or out of step (at various points Iraq and now fiscal debt did this for New Labour);

etc etc

I think the hardening of opinion at the ballot box is based on the biggest of big stuff, which is made up of many small decisions.

So, for example, Rudd’s ETS decision may not poll that badly in itself, but it feeds into a wider concern that the govt can not make up its mind about where it wants to take the country.

Qualitative polling seems more important than quantitative polls. Where do things like ‘ticker’ and ‘cost of living for working families’ come from. It’s in the things people say not what they say yes to.

11 years ago

I would also, wonder, Don, how much opinion at the ballot box on election day is more or less constructed on the fly. Obviously the hurdy-gurdy of an election campaign would imply less, but I think the two states may have more (in that they can both be hastily constructed) and less (in that the construction itself could differ dramatically) in common than posited.

Paul Frijters
Paul Frijters
11 years ago


I find the group angle raised here interesting because it suggests that there are latent groups to which politicians are trying to belong, but where the opinions pertaining to a group are shifting, creating the follow-or-lead problem for a politician. How would a poller go around ascertaining the implicit groups Bourdieux talks about?

Graham Young
11 years ago

Why people vote the way that they do is a complicated proposition. They may vote on the issues, or they may vote tactically. An example of the latter is the apocryphal Irishman who said “I don’t know who the government is, but whoever they are, I’m agin them.” In which case it doesn’t matter what the issues are.

It is true that particular parties “own” particular issues and that you are generally best to avoid an issue that is not owned by your side. A common mistake politicians make is to point out the errors their opponents have made in one of their areas of strength. The effect of this is a bit like criticising Don Bradman because he went out for a duck in his last innings. Rather than proving that he can’t hit a ball, it reminds the listener of all the balls that he has hit.

But another common mistake is to think that because an issue is in the nominal area of strength of their opponent, that the issue can’t be used. So, the Liberals are generally regarded as better managers than Labor. In which case, if they can turn a debate about health services into a debate about management instead, then, despite it being in an area of their opponents’ strength, they have a chance of winning the argument.

Which leads back to the role of qualitative research, which is to try to probe these nuances, giving the quantitative pollster a set of relevant, rather than irrelevant questions.

Voters who make up their mind at the last minute are generally likely to make-up their minds in line with the rest of the community, but not always so (look at the last Queensland election). They are not susceptible to nuances of policy debate, but they are influenced by blunt messages. So you can see a different pattern amongst them, depending on the messages in the last week of the campaign. But the messages they respond to are most likely going to be those that reinforce what they are already disposed to believe. Tell them something they are not disposed to believe and you hit a phenomenon called “cognitive dissonance”, a phenomenon which is dissonant with the claim that public opinion is just a creature of elite debate.

11 years ago

Ok, here’s the thing:

1. the Coalition must believe the public are willing to sacrifice some ‘goodies’ or they wouldn’t be cutting the Budget, they must think this is not mad politically.

2. Howard obviously never thought this after about 1998 as he loved to shovel the dough into hand outs …

What has changed. Well that’s the $64K question in this election. Has the recession (which many didn’t feel but most did in subtler ways thatn 1991) altered public tolerance for Govt making a tougher choice and cutting things they may benefit from.

This in my view is the public opinion question for 2010. The ETS and RRT policies from Labor complicate this analysis, but I am geneuinely of the view that teh Coalition will pretty much run a ‘no announcements’ campaign. In the past that would be suicidal political … is it still??

By ‘no announcement’, I don’t mean they won’t have policies or spend any money at all, but I think the Coalition will make very few promises aside from Budget cuts. They may reallocate some spening but not much.

Andrew Norton
11 years ago

Perhaps one way of looking at this is to say that there is public opinion on a wide variety of subjects, but one poll on its own cannot tell you whether it exists or what it is.

In a methodology I have tried to copy in a few articles, Murry Goot has for many years written articles which survey many polls on the same broad subject. If the same pattern of opinion appears despite differently worded polls over an extended period of time we can be reasonably confident that it is picking up on something real.


[…] "lot of opinion polling is useless because it doesn’t understand its limitations" writes Graham Young. One of the major limitations of polling is the tendency of respondents to answer questions about […]