We always include information items and these routinely show statistically significant gains, as in the first Deliberative Poll in Britain (Luskin et al. 2002), the National Issues Convention in the US, and almost all others. Sometimes the changes are large. Consider the Deliberative Poll in Northern Ireland on education policy. Even though the sample was drawn from parents only (because that is what the government authorities regarded as directly and legally relevant to education policy), there were massive gains in knowledge. On average, the sample answered only 22 per cent of the information questions about Northern Ireland’s education system correctly before deliberation, but answered 50 per cent correctly after deliberation. For example, the percentage knowing that schools receive more funding for older pupils increased from 21 per cent to 79 per cent, and the percentage knowing that the new entitlement curriculum requires that ‘every school provide all fourteen-year-olds with a choice of at least twenty-four subjects’ increased from 21 per cent to 74 per cent (Fishkin et al. 2007).12
We have found a consistent pattern in that those who gain information are also those who change their views (Luskin et al. 2002).13 The information driven model, which we found in the first Deliberative Poll, has defined a pattern for others.