On many occasions I have argued that the representativity of political assemblies constituted ‘descriptively’ (i.e. by statistical sampling) only applies at the collective level, and that this requires members of such an assembly being limited in their function (in contrast to the mandate of elected members). This argument has failed to persuade some participants in this forum, so this post makes the point in a rather stark manner, in the hope that it will challenge my opponents to refute it or else accept it – ‘ to put up or shut up’ – as opposed to merely ignoring it. I’m puzzled as to the continuing necessity to labour this point, as its veracity derives from the meaning of the word ‘statistical’, nevertheless I will seek to hammer the nail in one more time.
Statistical sampling via random selection is widely used for proportionate opinion polling, but the problem with using random selection for relatively complicated issues like political representation (as opposed to preferences over different brands of washing powder) is that such surveys are inevitably of ‘raw’ (unconsidered) opinion. Nevertheless the representativity of the proportional sampling techniques used is hard to deny, hence James Fishkin’s attempt to seek to establish a ‘deliberative’ assembly using random sampling techniques, which combines representativity with informed deliberation in order to represent the ‘considered judgment’ of the whole population. However this requirement leads Fishkin to advocate a very thin form of the deliberative ideal, in which members effectively listen to balanced pro–anti arguments and then decide the outcome via secret ballot, as opposed to the rich active deliberation preferred by Habermasian deliberative theorists. Why should this be?
A proportionate opinion poll on political preferences might provide an outcome indicating that (say) 60% of those polled held a ‘conservative’ perspective and 40% a ‘liberal’ or ‘progressive’ perspective on a particular issue. (For the sake of this simple illustrative example no attempt is made to define these terms or to introduce the additional statistical complexity of the ‘don’t know’ votes.) Such a survey would be held to accurately reflect the raw (unconsidered) preferences (or, if you prefer, prejudices) of the population being sampled.
A Fishkin-style poll would seek to measure these preferences at the beginning and the end of the deliberative process in order to establish to what extent the raw preferences were altered by the information-exchange process. The final vote would be held to represent the considered opinion of the population being sampled (assuming that the information was well balanced and comprehensive).
But what would happen if individual members started behaving as advocates for their own views? Sampling theory would suggest that the deliberative microcosm would be constituted in a broadly similar way (60–40) to the general population, but there is a significant possibility, or even probability, that a small number of participants would have the persuasive and status powers to influence the debate in a manner that did not reflect the underlying preferences of the greater population (note that this is not possible in an opinion poll, where every ‘vote’ [box tick] has the same weight). Opinion polls are only representative at the collective (statistical) level, so allowing individual members of a statistically-representative assembly active roles would be as invalid as suggesting that the views of individual respondents (selected at random from the sample) in an opinion poll reflected the average views of the population. Any such behaviour in the opinion polling industry would result in instant dismissal.
The injustice of the active advocacy model would be even greater if those sampled were empowered to introduce proposals to the forum. Recalling the 60-40 conservative/liberal assumption in the target population, by nature liberals (progressives) are more disposed to introduce political innovations than their conservative opponents, who would, on the whole, prefer to keep things as they are. Hence one might anticipate that the progressives would tend to set the agenda of the assembly, even though their views were in a minority in both the assembly and the target population. It is hard to reconcile such an outcome with the norm of political equality assumed by democratic theory, except via the Whig assumption that deliberation has the automatic effect of turning conservatives into progressives (conservatism being the result of indoctrination by powerful elites, rather than a natural predisposition).
In sum, if we are going to take the word ‘statistical’ seriously then any attempt to assign an active role to individual members of an allotted assembly should be abandoned. The analogy between a proportionate opinion poll and an allotted assembly is an exact one, the only difference being the information factor. This is the reason that the most successful method of establishing deliberative democracy (the Stanford DP) incorporates such a weak form of deliberation – members, in effect, simply ‘weigh’ the arguments and then determine the outcome via the secret ballot.
Filed under: Sortition |