Election 2004: Barely registered: Is public opinion good enough for democracy?

Shaun Wilson, Australian National University

Adam J. Berinsky Silent Voices: Public Opinion and Political Participation in America, New Jersey, Princeton University Press, 2004 (224 pp). ISBN 0-69111-587-7 (hard cover) RRP $59.95.

Vincent L. Hutchings Public Opinion and Democratic Accountability: How Citizens Learn about Politics, Princeton, Princeton University Press, 2003 (171 pp). ISBN 0-69111-416-1 (hard cover) RRP $96.95.

Commentators have long been fixated on how to measure, aggregate, and interpret ‘public opinion’. Opinion pollsters in modern democracies claim to gauge public opinion and to tell governments how they should act. Yet, understandably, these claims are perennially questioned. Do polls measure our real interest in public policy and social problems as well as our capacity to agree or disagree? Is polling dangerous because it claims to inform about democracy without involving any flesh-and-blood democratic exchange?

Two new American studies offer intelligent and rigorous lessons on two problems in public opinion. In Public Opinion and Democratic Accountability Vincent Hutchings tries to discover whether public opinion is informed enough for democracy to work effectively. In Silent Voices Adam Berinsky tries to discover how our view of social attitudes and public opinion would change if we could include the views of those marginalised by the conventional ways of collecting survey information.

Hutchings attempts to provide a minimal vindication of the performance of contemporary (American) democracy. By minimal, I mean that Hutchings doesn’t test whether public life and public opinion meets the expectations that Jefferson, Rousseau, Habermas, or Dryzek hold for it. His target is worthy nonetheless. He challenges the pessimistic, generally elitist view of democracy that doubts the public is ever informed enough, and indeed even attentive enough, to perform the minimal role in its charge: to ensure that all democratic choices are informed.

Hutchings sets out three tests to assess whether voters exercise their choices in an informed way. First, how well does the media report on issues voters care about? Second, do voters take notice of this information when it really matters? And third, does having this information make a difference to voters’ decisions? He tests his argument using National Election Studies and Senate Study findings as well as detailed analysis of media coverage of major congressional votes during the 1980s and 1990s.

Hutchings believes that voters do take notice when politics concerns them.

On the first test, Hutchings finds that the media does cover Senate votes ‘adequately’ although voters receive more information about contentious rather than routine matters and when Senators vote against their party (pp. 30–31).

On the second test, Hutchings demonstrates that voters are motivated to take notice of political subjects that matter to them (he focused particularly on the Clarence Thomas Supreme Court confirmation vote, which produced novel political alliances). This is an important twist on the relevance of what political scientists call ‘issue salience’: following Hutchings, the media is especially attuned to report controversies that ignite public interest and give voters real information about the state of politics. And Hutchings also finds some voters—Christian conservatives and women—are highly responsive when their issues are raised in campaigns, suggesting elite efforts at controlling the political agenda warms the political temperature of major constituencies, while other voters (citing high-earning women and unionists) are partisan regardless of the prominence of their interests. We get a picture of a segmented electorate that is not as unresponsive as we are often led to believe, an electorate that can be successfully mobilised by elites, and that is minimally served by a media attuned to reporting salient political developments.

On the third test, Hutchings’ careful research tells us something about the level of ‘democratic responsiveness’ in contemporary politics. Responsiveness in broad terms is about the public’s capacity to respond meaningfully to political developments and vice versa. Hutchings believes that voters do take notice when politics concerns them, the media provides information relevant to political conflict, and that politicians respond to mobilised constituencies. His contribution adds a largely empirical volume to a wider controversy about democratic responsiveness in American political system (see, for instance, Manza, Cook & Page 2002). In effect, Hutchings’ view is one of a limited, but not to be dismissed, responsiveness, which we might position between scholars like Erikson, Mackuen, and Stimson (2002, pp. 33–53) who use aggregate opinion and policy measures to claim a high responsiveness between public opinion and politics, and scholars with Jacobs and Shapiro (2002, pp. 54–75) who believe politicians and the political system is losing its responsiveness to the public.

The level of political responsiveness is an empirical question that can’t avoid a confrontation with the measures routinely used to measure public opinion. Market research, private opinion polls, and social/political survey research have achieved the status of a precious commodity in the political marketplace but the quality and uses of opinion data continue to be controversial. Do respondents just tell pollsters what they think is socially desirable? Is the information gathered too contrived to reflect the complexity of a beliefs and policy preferences to substitute for deeper deliberation? And do biases that arise out of routine (but frequently dismissed) problems in surveying mean that results are plain misleading?

Large numbers of abstaining respondents are most likely when subjects get morally sticky.

The contrived nature of polling led sociologist Pierre Bourdieu to famously declare that public opinion does not exist, and that the ‘Don’t Know’ response is often the most useful information social scientists can find out about subjects. A thorough empirical investigation of the problem of respondents who tell pollsters they don’t know the answers, or slip out of the pollster’s sampling altogether, is offered by Adam Berinsky in Silent Voices. Berinsky believes that abstentions in survey responses really matter, especially when pollsters are dealing with sensitive subjects like race or sexuality (he calls this social complexity) or difficult to understand subjects like social welfare policy or budgets (cognitive complexity).

Large numbers of abstaining respondents are most likely when subjects get morally sticky or tough. Yet having a firm idea of the real state of public opinion is often most important on these difficult and controversial subjects. Berinsky says researchers should give these abstainers a ‘voice’ by seriously attempting to simulate their probable responses (based on measurable interests and opinions) so that we can see how public opinion might look had abstainers given pollsters answers. Berinsky’s strategy, then, uses statistical techniques to impute opinions to those who respond ‘Don’t Know’, largely by building a profile of the interests and attitudes of non-respondents and dealing with the serious problems of selection bias. Although Berinsky accepts that non-responses are legitimate (p. 48), he persists in arguing that critical information is often lost because of the social context of surveying: ‘Simply because respondents are unable easily to translate their thoughts and feelings into a summary judgment does not mean that those concerns are irrelevant’ (p. 49).

Like Hutchings, Berinsky focuses on salient and controversial issues that tell us a lot about the state of public opinion and political life: racial attitudes between 1972 and 1994; social welfare policy between 1972–1996; and a most well studied subject for opinion researchers: attitudes to the Vietnam War (which I’ll leave to one side). In interviews conducted for major studies like the National Election Study series, high levels of non-responses are not infrequent. Berinsky’s analysis of non-responses about school integration between whites and African-Americans finds that respondents with anti-integration feelings are more likely to use the ‘Don’t Know’ response to avoid expressing this opinion directly. Interestingly, more respondents choose non-responses now than they did in the past, probably because it is now more socially undesirable to express racist attitudes to an interviewer. In Berinsky’s terms, race has become more ‘socially complex’ and opposition to equality may be less acceptable in the current climate. Yet, this opposition lurks in the mud of numbers. Berinsky makes the following important assessment: ‘While Americans may express overwhelming support for the principles of integration, there appears to be a strong undercurrent of opposition to government intervention that is not visible in the marginals of opinion polls’ (p. 82).

Australians and Americans might also express the same ‘hidden hostilities’ to gay marriage, for instance, when they choose a non-response to a question about recognising same-sex relationships. In a June 2004 Newspoll, almost one in five respondents were uncommitted to supporting or opposing gay marriage. We might also want to know more about the 22 per cent of respondents to an August 2004 Newspoll who claimed to be ‘uncommitted’ when asked whether they agree with the Howard government’s action during the Tampa incident in 2001. Although public opinion is, on balance, unfavourable to the government’s response to Tampa (43 per cent), the high level of uncommitted respondents reveals all the problems that Berinsky brings out: a complex political problem, a normatively charged social issue, and (given recent revelations) more incentive to conceal support for the government’s actions.

These two books give us concrete evidence about things many of us claim we already know.

Americans might be hiding their soft racism from pollsters today, but Berinsky’s analysis of attitudes to the socially and cognitively complex question of social welfare policy shows something different. Berinsky finds that educated, higher income conservatives are more likely to express anti-welfare state opinion that colours the aggregate picture we have of support for the welfare state (p. 104). He reveals a more insidious relationship between biased polling and political exclusion: the strong voices of the middle class shape the results of opinion polling, which only serve to reinforce both the conventional wisdom about the anti-welfare instincts of the American public and, presumably, the mean spirited measures of the US welfare state (p. 104).

Berinsky’s findings on attitudes to welfare confirm the central normative claim of his study: ignoring non-responses has political consequences that go beyond the mere statistical problems of poor information. Yet he continues to affirm the place of opinion polling in public life, concluding that ‘polls may play a valuable role in the American political system by giving us a window onto the shape of the will of the people’ (p. 143).

These two books should interest us because they give us concrete evidence about things many of us claim we already know. Hutchings suggests that there is a minimum of democratic accountability in contemporary American politics: citizens can identify their interests and express them with the limited (and often biased) resources available to them. He does not apologise for the failings of American democracy; instead, he guards against the elitism on both left and right that disparages ordinary citizens by locating the fragile operations of a democratic public sphere (attentive citizens, relevant communication, and political responsiveness). It goes without saying contemporary democratic relations are degraded by media distortion, power, and control. Berinsky offers advice at another level, warning about our reliance on polls: frequently, polls fail in the task of reporting deep information about the state of public opinion, and polling biases may have an important impact on both our self-understanding and policy.

Election times are a ready reminder that polls are here to stay, and that they are a resource for the worst kinds of political cynicism: an artificially crafted responsiveness to the public for the purpose of winning elections, pitting majorities against outsiders (‘wedge politics’) and cultivating short-term thinking on policies. Yet our social and political compasses are probably more reliant than ever on polls. In recent months in Australia, polls have brought us solid information about more positive attitudes to asylum seekers and evidence of a huge shift in the public priorities for taxes and spending. The more liberal mood in Australian politics is a reminder that cynical governments and the worst kinds of poll-driven politics cannot outsmart the shifts and swings of public opinion in democratic societies, which of course depend on a deeper level of communication that cannot be replaced.


Erikson, R. S., Mackuen, M. B. & Stimson, J. A. 2002, ‘Public opinion and policy: causal flow in a macro system model’, in Navigating Public Opinion: Polls, Policy, and the Future of American Democracy, eds J. Manza, F. L. Cook & B. I. Page, Oxford University Press, Oxford, pp. 33-35.

Jacobs, L. & Shapiro, R. 2002, ‘Politics and policymaking in the real world’, in Navigating Public Opinion: Polls, Policy, and the Future of American Democracy, eds J. Manza, F. L. Cook & B. I. Page, Oxford University Press, Oxford, pp. 54–75.

Manza, J., Cook, F. L. & Page B. I. (eds) 2002, Navigating Public Opinion: Polls, Policy, and the Future of American Democracy, Oxford University Press, Oxford.

Shaun Wilson is Research Fellow at the Centre for Social Research in the Research School of Social Sciences at the Australian National University. He is a Principal Investigator of the Australian Survey of Social Attitudes, which will include the World Values Survey in 2005.