Opining About the Perils of Opinions: The Opinion Poll Model of Policy Making

29 Jun

Public opinion is central to the democratic political process and it has never been more important that today, when opinion poll numbers are constantly cited in the media to buffet policy choices. It behooves us hence to foremost understand opinions and the process of opinion change, and then to think critically about whether the causal mechanism driving ‘opinion change’ are commensurate with the expressed ideals of ‘democracy’.

What is the value of an ill-considered opinion from a person with limited knowledge of the facts? Close to none, one would expect. But apparently, it is worth much more in a policy debate on the Hill if sound bytes by politicians quoting poll numbers to buffet the validity of their issue positions are anything to go by.

Courtesy significant advances in sampling methodology, communication technology, and computational technology, one can now conduct a nationwide Opinion Poll cheaply (relatively) and quickly. Every major media company, from New York Times to Fox News, now publishes stories about the ‘findings’ from the polls with unerring frequency and drops these numbers casually on near about every policy issue, let alone questions like, ‘What should Paris Hilton eat for breakfast?’

Given the important role that media has in ‘framing’ the issue (Iyengar), and the fecundity of the polls, news media now often cites figures from opinion polls as part of a story on an issue and asks politicians to defend their policy choices (in six seconds or less) given the poll numbers. Correspondingly politicians increasingly cite poll numbers on issues as corroboratory evidence for or against a policy direction.

Where do opinions come from?

In a culture that values ‘individual expression’ above everything else, it isn’t surprising that people offer opinions on issues they know little to nothing about an issue. Funnily, and as has been extensively documented in Political Science literature, people not only offer opinions about what they know nothing about, they also offer opinions about non-existent (phantom) issues. (Lippman, 1993 and others) Krosnick et al. have posited a more benevolent interpretation of ‘phantom opinions’ arguing that these opinions originate from ‘violation of communication norms’. Even if Krosnick is right, there is wide agreement within the field that the general public which makes it to the voting booth and gleefully casts its vote (a behavior strongly based on overall opinion) is deeply ignorant about most issues.

Leaving aside ‘phantom opinions’, let us try to understand where opinions come from. “Every opinion is a marriage of information and values-information to generate a mental picture of what is at stake and values to make a judgment about it” (Zaller, 1991). It is important to notice how Zaller uses the term ‘information’ which he describes later in the paper as whatever political information a person consumes via media or other ways. By limiting himself to political information, Zaller mistakenly assumes that political opinion making sits in an isolated bunker – only affected by relevant political information – in people’s minds. Neuman in his book, ‘Common Knowledge’ has persuasively argued to the contrary. Leaving Neuman’s objections aside, it is easy to surmise that information has generally little to do with facts of the case. Secondly, we have yet to tackle how much of the opinion is driven by ‘values’ and how much of it is driven by ‘information’ but it seems intuitive that the mix would vary depending on a variety of factors ranging from the issue at hand (opinion on a value issue like abortion would inarguably have higher percentage of ‘value’ as compared to one on economic policy), need for cognition (people with higher need for cognition would use more ‘information’), cognitive ability, amount of information, etc.

Normative Questions

Since the publication of American Voter (Cambell et al.), and Converse’s later explications (1964, 1970), which described the average American voter as apathetic, and largely ignorant about major issues, political theorists have grappled with the threat that an uninformed voter poses to the claims of normative superiority of democracy. If democracy was to be claimed as a ‘normatively superior system’ in itself, without resorting to claims about its superiority as an instrumental good that provided ‘better governance’, it was important for the political theorists to argue that voters voted their interests – a claim which was no longer possible in lieu of evidence that pointed to widespread ignorance. The more severe threat that the theorists are rightly concerned about, is whether the democratic system can continue to deliver its benefits if the voters ceased to vote ‘their interests’ given a lot of benefits in the system are predicated on that assumption. While the conjecture is open to empirical analysis, we can theoretically analyze the value of an ‘opinion’ in an ideal democratic model.

The value of an expressed opinion in a democracy is directly proportional to its ability to tap into a voter’s ‘real interests’, best understood as ‘interests’, as understood by the voter, under ‘full knowledge.’

Ideal Opinions, Opinion Aggregation

Value of opinion cannot be pried apart from the system in which it is used. The composition of ‘ideal opinion’ would vary according to the system. Since we are talking about a democracy, let’s analyze its composition here.

The value of an opinion in a discussion is if it reveals a hitherto unknown piece of information. In a poll where you are asked to furnish your ‘considered preferences’ that benefit is lost to some degree. One can argue that there is indeed some knowledge hidden in the choice but since a poll weighs considered choices equally with ill-considered ones, and because we don’t know what led to the final choice, it is impossible to argue whether democratic majorities do bring forth collective knowledge. The only condition in which the scenario would hold is when a majority of the voters vote for the ‘right’ preference.

Let us now assume ‘full-information’ and the only variable as ‘value system’. If there is a ‘common’ universally accepted value system, then there is little value in soliciting opinions from everybody. It is when you start thinking about ‘averaging’ across value systems, a tenuous concept at best, that you need to think about soliciting opinions from others. Polling populace on a fixed choice of politicians is an impoverished way to go about tapping into ‘people’s will’.

Perhaps the way to tackle the problem is by actually breaking this down into a two-step problem – information maximization and ‘averaging’ over values. I believe we have the scientific community to address the former, but I do not have the answer to the latter but perhaps informed deliberation – which involves ‘public exchange of reasons’ – about consequences would be one way to approximate that.