Nick Gill and Tom Lord from the Sortition Foundation recently looked at the intersectionality of educational attainment and ethnicity in responses to our own citizens’ assembly recruitment processes.
But what does that actually mean?
In the world of democratic lotteries there is a common concern and ongoing debate about whether simply recruiting to targets, such as gender and age, will miss out on some important and relevant aspects of societal demographics. For example, what if we recruited only older women and only younger men? We would still match the demographic targets of having half women and half men and half younger and half older, but such a citizens' assembly would not look at all like the community it is meant to represent.
Luckily, in practice, these kind of sharp divisions never occur, due to the random nature of the processes involved.
However there are more subtle, and perhaps more important intersectionality concerns and Nick and Tom investigated two aspects of this:
- What is the relationship between educational attainment level and other measures of socio-economic status? We often use education as a measure of socio-economic status but there are others, most notably when working in the UK, the Index of Multiple Deprivation (IMD). We want to know how our choice of measure affects who we recruit.
- What is the connection between educational attainment level and ethnicity, for the people that we recruit? Is it the case that the set of respondents from marginalised ethnic groups is especially skewed towards being highly educated?
To answer these questions, we did some analysis of two recent assemblies for which we did the recruiting: the Citizens' Assembly on Democracy in the UK and Art for the People Citizens’ Assembly on Arts, Culture & Creativity, Coventry UK. Our analysis led us to a number of conclusions:
- First, our analysis confirmed something we already knew: the overall pool of respondents tends to over-represent people with higher educational attainment and under-represent people with lower educational attainment. This makes it all the more important that our selection process is stratified -- so we make sure that we select people with a range of educational attainments.
- Second, there is strong evidence that the over-representation of people with higher educational attainment persists even within the set of people who live in areas of low IMD (i.e. in more deprived areas). It is tempting to think that if we make sure some of our assembly members live in more deprived areas, then we have done enough to ensure diversity of socio-economic status. However this finding tells us that if we do this, then we run the risk of over-representing highly educated people -- stratifying according to IMD is not sufficient to eliminate this skewing.
- There is some evidence, and our results here are not definitive, that the skewing just mentioned happens independently of ethnicity -- so is not particularly better or worse in different ethnic groups. This finding was less clear-cut and is something we will need to investigate further.
One potential solution to achieving inclusion of marginalised voices is to use a hybrid recruitment strategy, the value of which has already been noted in the literature. The strategy would be to intentionally seek out representatives in identified, low-registration groups. Practically, we have started to investigate the possibility of combining our standard postal recruitment with some targeted door knocking. This approach has been used very successfully in Germany.
But this is not the end of the process. We will continue to investigate and innovate as we seek to support events to be all three of: diverse, representative of the population and inclusive.