made, and the five least talkative members accounted for under 8 percent of the comments." 60
A final criticism of the deliberative opinion poll was that public opinion did not change all that much as a result. Individuals' opinions changed on many issues, although Fishkin and his colleagues have not yet separated how many of those changes were of degree (from agree to strongly agree, for example) and how many of the changes were of kind (from agree to disagree). Unfortunately, the way the deliberative opinion poll was taken does not allow us to determine how much of that change was due to the discussions in Austin and how much of it was a result of other factors. As Merkle points out, even when people do not engage in discussion or gain new information, their responses to the same survey questions often change over time. 61
Perhaps even more important, aggregate opinion did not change much after the deliberations. Of eighty-one opinion questions the interviewers asked NIC participants, in only seven cases did majority opinion change af-
ter the discussions. The National Issues Convention was a very expensive undertaking, and many critics wondered if the cost was worth the results. It seemed as though the "informed" citizens in Austin recommended more or less the same things that the "uninformed" citizens who had not taken part in the deliberations would have. 62 Perhaps the "top-of-the-head" answers citizens give to survey takers are in fact sufficient.
As Merkle points out, deliberative public opinion polls can still offer some intriguing possibilities. For example, people who participate or observe such polls may develop more reasoned opinions and be less susceptible to manipulation. Deliberative polls could also be helpful in cases where the public knows very little about an issue, as opposed to cases involving well-established public debates. These polls have also been used "as a way of getting citizens involved in framing the issues" or in cases where a researcher wants to combine the advantages of both focus group and survey research. 63 However, it seems unlikely that polls of this sort will replace more traditional survey research.
In this chapter, we have discussed the relationship between public opinion and democratic processes. Consensus about basic democratic values keeps our democratic system stable. However, important differences of opinion among major social groups and doubts about the ability of citizens with very limited interest in public issues to govern a nation raise questions about how public opinion should be integrated into this democracy. Should the public's power be limited to choosing between two major party candidates at elections held every few years, or should public opinion be a factor