October, 2012
Volume 22, Issue 3



Simulating the Marketplace of Ideas: Fully Considered Healthcare Policy Preferences and Deliberation Effects on Opinion Change

Jin Woo Kim
Ph.D. Student, Annenberg School for Communication, University of Pennsylvania

A key characteristic of deliberative democracy is the requirement for the citizenry to think about the potential advantages and disadvantages of a political choice, as they form a public opinion. Clearly, those who do not know much about politics in the traditional sense – i.e., those who score low on quizzes tapping neutral facts – are not likely to make political decisions in such a fashion. But do informed citizens, at least, give careful consideration to various viewpoints and competing arguments regarding the issue at hand? If not, do the opinions of the "informed" citizenry differ from the opinions of the "considered" citizenry? And if there are differences, would actual group deliberation guide participants to adopt the collective opinions that the fully considered public is predicted to hold? While these questions raise important issues regarding the civic competence of the knowledgeable in light of deliberative theory, the extant research provides little evidence leaning either way.

To address these questions, I drew on The Healthcare Dialogue Project, a multi-wave study involving several online deliberations and surveys. Participants were drawn from a representative sample of adult American citizens (aged 18 or older) maintained by Knowledge Networks. The surveys included open ended-questionnaires that asked the participants to mention arguments for and against a healthcare policy proposal (nationalized universal healthcare), as well as more conventional measures of political information. This allowed me to examine how political information is related to the numbers of pro and con arguments given by the respondents. I then used the estimated effects of pro and con considerations on policy preferences to statistically impute the collective opinions that the public would form, if they all had "fully considered" the merits and perils of the policy. I calculated each respondent's fully considered preferences by replacing the observed values of pro and con arguments with their highest values. I also imputed "fully informed" preferences in a similar way, this time using political information, instead of considered arguments. Then I compared "fully considered" and "fully informed" opinions, which respectively corrected for narrow and imbalanced considerations and political ignorance. Finally, I pitted the actual collective opinions emerging from group deliberation against the two statistically imputed opinions.

My analyses show that well informed citizens tend to take a slanted set of arguments into consideration. In some cases, well informed respondents were no more likely than their uninformed counterparts to mention arguments that are uncongenial to their view or partisan identity, although they still knew much more congenial arguments than the less informed. As a result, "informed" preferences diverge from "considered" preferences in important ways; "fully informed" healthcare policy opinions are expected to be more polarized along the partisan line, and be more conservative as a whole than the observed opinions. On the other hand, "fully considered" preferences are predicted to be more liberal, and show "divergence with polarization." Finally, I also found that actual deliberation moves policy opinions towards statistically imputed fully considered preferences, but not towards fully informed preferences. Taken as a whole, these findings suggest the need to better conceptualize and measure both what we mean by informed opinion and knowledgeable citizens. They also suggest a troubling tension between the "well informed" and the "well considered" public.

Note: Gathering of the data reported here was supported by grants to Drs. Vincent Price and Joseph N. Cappella from The National Science Foundation (Grant EIA-0306801) and the Annenberg Public Policy Center of the University of Pennsylvania. Views expressed are those of the author alone and do not necessarily reflect opinions of the sponsoring agencies.