The punchline of a Deliberative Poll is a change in policy attitudes, a change in answers to the question: what is to be done? We studied fifty-eight indices of policy attitudes in nine national Deliberative Polls conducted between 1995 and 2004. Four of these Deliberative Polls were American, four were British and one was Australian. Topics varied from US foreign policy (2003) and the US general election (2004) to Britain’s future in Europe (1995) and the British General Election (1997). All samples were national with sample sizes ranging from 238 to 347. Seven of the Deliberative Polls were face to face and two were conducted online.
The first point to note is that there is a lot of change in policy attitudes. Seventy- two per cent of the fifty-eight indices show statistically significant net change comparing the answers on first contact with the answers at the conclusion of deliberations. The magnitudes of the changes are also large (Luskin et al. 2007b).5
A second point is that there is clearly an effect of salience. The more salient the issue to begin with, the less likely the net change. If respondents have already processed an issue, even with fairly imperfect and unbalanced deliberation in their daily lives, they are less likely to change their views. They may, in other words, have already arrived at fairly firm views. If we proxy salience by the time one knowledge scores, then there is a strong negative relationship (Luskin et al. 2007b).6
While there is a great deal of net attitude change, deliberation has a value even when there is no change. If the public thinks X should be done, but has not thought about the issue much, has not tested its views in comparison to alternative policies and the reasons for them, then there is an issue about how seriously, from the standpoint of normative legitimacy, one should take those views. They reflect very little thought and little consideration of opposing possibilities. On the other hand, if those views survive a serious deliberation unchanged, then they take on the added legitimacy of the public’s considered judgements. Those views have been tested in a context of opposing arguments with good information. Hence, regardless of change, the conclusions at the end of a well constituted Deliberative Poll offer a representation of what the public would think if it were thinking about the issue under transparently good conditions. It is those judgements, change or no change, that should have a recommending force to policy makers and representatives and those concerned with the public dialogue. While some of those views may correspond with the views in conventional polls, one can never be sure, unless they have been tested by adverse arguments, unless they have been tested by serious deliberation about policy alternatives.
It is worth adding that some Deliberative Polls have occurred in contexts soon before a vote where the results were widely publicized by broadcast media partners. In those cases - the British General Election, the Australian referendum on the Republic, the Danish referendum on the euro - there were significant changes in voting intention. Hence the opinion changes are not only significant; they can also be politically consequential.