The phrase “deliberative polls” most often refers to political philosopher James Fishkin’s conception of a multi-stage opinion poll that incorporates systematic deliberation on policy issues. As such, Fishkin has trademarked the phrase “deliberative polling.” It can refer to other methods of introducing a deliberative component to public opinion research, some of which are touched on below.
Deliberative polling addresses the relationship between public opinion and policy outcomes. Many political philosophers believe that in a representative democracy, policy should be responsive to considered public preferences. Many empiricists conclude that policy is responsive to public opinion as expressed in conventional surveys – but whether such opinion is “considered” is another matter. Survey researchers agree that citizens do not typically form considered opinions on policy issues. As Anthony Downs (1957) has argued, citizens tend to practice rational ignorance: they expend little effort informing and refining political opinions that, individually, will have vanishingly little impact on outcomes. In consequence, conventional opinion surveys often tap what Philip Converse (1964) called “nonattitudes”: opinions made up on the spot. As such, not only are individuals’ responses often unstable, but the aggregate results can be distressingly inconsistent (for instance, influenced by substantively trivial changes in wording) and, arguably, irresponsible.
Deliberative polling seeks, in Fishkin’s words, to establish “the conclusions people would come to, were they better informed on the issues and had the opportunity and motivation to examine those issues seriously” (Fishkin 1997, 162). Fishkin proposes that the results have a “recommending force”: they indicate to political leaders and the public at large what the public’s view of an issue would be if citizens were to form considered opinions.
Deliberative polling begins with a baseline survey administered to a random sample of the relevant public. Many of these respondents – typically several hundred people – then convene to learn about the specified issues, through carefully vetted briefing materials, and to discuss the issues in moderated discussion groups. At the end, a post-deliberation survey measures possible changes in attitude induced by the process. For instance, the (US) National Issues Convention in January 1996 convened 460 participants (about half of the initial respondents) for three days of discussion on three broad issues: the US economy, the US role in the world, and the state of the American family. Participants discussed the briefing materials in small groups of about 15 apiece, and then posed questions to diverse panels of experts. The final survey elicited, inter alia, substantially less support for a “flat” (rather than progressive) tax, and considerably more support for current US levels of foreign aid, compared to the baseline. Follow-up surveys indicate that such changes in aggregate opinion tend to weaken, yet persist, well after the initial study.
Deliberative polls under Fishkin’s trademark have been conducted in the United States and elsewhere. They were originally conducted under the auspices of the Center for Deliberative Polling, University of Texas at Austin (often in cooperation with other organizations), and since 2003 by the Center for Deliberative Democracy at Stanford University. Most deliberative polls focus on substantially narrower issues than the National Issues Convention described above. Various polls have considered foreign policy priorities, public utility policies, criminal justice, and adoption of the euro.
Deliberative polls can be conjectured to improve opinion quality in various ways. Factual knowledge questions often evince substantial information gains. Some studies show marked gains in opinion constraint – for instance, larger correlations among responses to arguably related policy questions – although there is little evidence that deliberative polls actually narrow prior differences in constraint. Deliberative polls could evoke more stable expressed opinions (e.g., larger correlations over time), but available data do not permit stringent tests. Some evidence indicates that deliberative polls may reduce the likelihood of preference cycles (in which distinct majorities prefer outcome A to B, B to C, and C to A). More subtly, deliberative polls can be hoped to increase majorities’ willingness to accept the costs and risks of their preferred policies. It is unclear how such willingness could be measured systematically.
Deliberative polling combines values of standard social scientific research and democratic theory, creatively but not without compromise. In comparison with conventional surveys, deliberative polls resist replication due to their scope and expense, especially in national studies entailing long travel. The attempt to present balanced information is inevitably problematic. The group dynamics are inherently unpredictable; the use of multiple small groups does provide for measurable variation. The survey instrument limits participants’ ability to express their views. Some critics argue that any change induced by deliberative polls amounts to a Hawthorne Effect – participants’ response to knowing they are being watched – with no normative relevance. Advocates reply that the Hawthorne Effect classically refers to improved performance; if people deliberate more effectively and responsibly because they are conscious of being watched, so much the better. Observers must decide for themselves whether the results merit “recommending force.”
Deliberative polling bears comparison with other modes of deliberative research. These include, among many others, the “Citizens’ Juries” designed by Ned Crosby, citizen panels convened by the Public Agenda Foundation, Alan F. Kay’s public interest polling, and Holland’s “General Social Debate” of the early 1980s. These methods vary among many dimensions, including the number of participants involved; the efforts to assure representative participation; the length of the program; the scope of the agenda, and whether participants can have any influence; the amount and form(s) of information provided; the opportunities to discuss the issues and to ask questions; and the means by which final opinions are assessed. Arguably, given Fishkin’s reasoning, various methods should tend to converge upon qualitatively similar outcomes. This provocative possibility has been little explored.
- Converse, P. (1964). The nature of belief systems in mass publics. In D. E. Apter (ed.), Ideology and discontent. New York: Free Press, pp. 206–261.
- Downs, A. (1957). An economic theory of democracy. New York: Harper.
- Fishkin, J. (1997). Voice of the people. New Haven, CT: Yale University Press.
- Lindeman, M. (2002). Opinion quality and policy preferences in deliberative research. Research in Micropolitics, 6, 195–221.
- McCombs, M., & Reynolds, A. (eds.) (1999). The poll with a human face: The National Issues Convention experiment in poll communication. Mahwah, NJ: Lawrence Erlbaum.
- Sturgis, P., Roberts, C., & Allum, N. (2005). A different take on the deliberative poll: Information, deliberation, and attitude poll. Public Opinion Quarterly, 69(1), 30 – 65.