Who’s in the Room? How Well Did Two National Minipublics Capture the Diversity of Attitudes in the Swedish Public?

Minipublics derive their legitimacy and interest from their claim to reflect broader society and to include a diversity of relevant perspectives in their deliberations. Minipublics are recruited through a combination of random selection and stratification to minimise selection biases that otherwise would undermine the inclusiveness, balance and perceived legitimacy of minipublics and their outputs. We test whether members of Sweden’s first two national-level deliberative minipublics differ from the general public in terms of attitudes. The Food Agency citizen panel on sustainable, healthy and affordable diets was recruited using geographical and demographic stratification variables. The Swedish climate assembly had a higher monetary incentive and included stratification by demographics, geography voting intentions for the eight political parties, as well as degree of climate worry. We test whether minpublic answers to pre-deliberation differ from answers to the same questions asked in national representative surveys. We use principal component analysis to identify and compare key attitudinal dimensions and compare the distributions of the first principle component. Members of the food panel answered significantly differently from the public for 9 out of 16 questions. This was also reflected in significantly different scores on the first principle component (accounting for 42% of variation in responses in the national survey). Food panel members expressed greater support for government interventions and concern for climate than the national sample. Recruitment quotas for political support, climate concern and demographics were almost completely met for the climate assembly. Unsurprisingly then, members were more similar to their respective national sample than in the food panel. However, significant differences remained for 3 of 26 analysed questions. Members were more in favour of citizen engagement on climate policy, felt greater personal agency for climate, and expected employment opportunities to result from the green transition. There was no significant difference in the first principle component (accounting for 35% of national variation), although the minipublic members did not represent the full range of values recorded in the national survey. A discursive fringe of the Swedish population was apparently not represented in the assembly. Stratification including an attitudinal and a political support variable, and higher incentives recruited a more representative minpublic for the climate assembly than was achieved for the food panel. Nevertheless, the most extreme skepticism towards climate policy and action was not captured in either minipublic. Democratic lotteries with stratification can improve the diversity and representativeness of a minipublic. However persistent selection biases as illustrated in both cases has important implications for inclusiveness of perspectives in deliberation, for democratic legitimacy, and for the ability of assemblies to capture and engage with the full diversity of discourses amongst the public.