Article by Annabelle Lever: “…Creating a citizens’ assembly that truly reflects society as a whole isn’t so simple, however. In particular, only a very small percentage of those invited to participate actually agree to do so. According to a 2017 study published European Journal of Political Research, the precise percentage depends on how large, complex and time-consuming the process is likely to be. It ranges from 4% for larger, more onerous assemblies to 30% in a couple of exceptional cases, and averaging out at 15% across all countries and all forms of assembly. As a consequence, the formal equality of opportunity that unweighted lotteries promise tends to result in assemblies skewed to the socially advantaged, the partisan, and those most confident in their practical and cognitive abilities, whatever the reality.
To create an assembly that is more descriptively representative of the population – or one that looks more like us – several approaches are used. One is to have an initial phase of unweighted selection followed by a second phase that uses weighted lotteries. Another is to use stratified sampling or forms of stratification from the beginning.
For the Climate Assembly UK, organisers sent out 20% of its 30,000 letters of invitation to people randomly selected from the lowest-income postcodes, and then used random stratified sampling by computer to select 110 participants from all the people who were over 16 and free on the relevant dates.
Because citizen assemblies are very small compared to the population as a whole – France’s Convention for the Climate was made up of just 150 people – the descriptively representative character of the assembly can occur on only a few dimensions. Organisers must therefore decide what population characteristics the assembly should embody and in what proportion. Randomisation thus does not preclude difficult moral, political and scientific choices about the assembly to be constructed, any more than it precludes voluntariness or self-selection…(More)”.
