Read on Twitter
Small, niche thread on (political) polling and whether we need to start having BAME/ non-white cross breaks as a standard in public polls, inspired by various conversations from @election_data, @MattSingh_ and @sundersays and others... (1/14)
In theory I’m sympathetic. We measure UK public opinion, and segment by a host of critical markers that drive voting patterns such as age, education, working status, and region. Showing how the UK votes and thinks by ethnicity should in theory be straightforward (2/14)
The U.K BAME population is c.15% (last census was 2011) today, bigger than the individual populations of Scotland, Wales, South west England, 18-24 age bucket, all of which are considered valid ways/buckets to analyse the UK. (3/14)
I’ve got 2 broad issues with analysing all public opinion polling by ethnicity. The first, is BAME a valid, coherent and fair way to group a portion of the electorate? Britain’s minorities vote in very different ways and they are starting to increasingly diverge over time (4/14)
Voters of Indian, Chinese, East African backgrounds+voters of Hindu+Sikh faiths increasingly vote as per white voters once statistically adjusting for other demographic factors. Many of the MRP estimation errors in 2019 were made in seats with high Indian % drifting right (5/14)
The other side of the BAME story is that voters of Pakistani, Bangladeshi, Carribean and mixed race backgrounds - even after adjusting for income, education, density of location are drifting rapidly to the left. (6/14)
Does it make sense in modern Britain to analyse voters by skin colour when non-white voters differ from each other significantly on many metrics - but crucially future political direction. All of this is related to my second issue with doing BAME specific cross breaks (7/14)
Because the typical UK poll of 2,000 people will still only include 300 BAME people, of whom 200-250 will express a voting intention. The BAME % is therefore likely to fluctuate depending on the composition from poll to poll. This could encourage sensationalist reporting (8/14)
I can see headlines of “shock” polling showing the Tories only at 10% amongst BAME voters in left papers, one day, and Tories making “shock” advances to get 1 in 3 BAME voters in right papers the next. The BAME cross break is vulnerable to being deployed in a culture war (9/14)
I can also imagine a world though where being able to analyse every issue via an ethnicity cross-break means that those who consume research for commercial and political purposes and are key decision makers are shown meta-facts and perspectives they might otherwise miss (10/14)
Many prescient political analyses though over the last 10 yrs have come from junking standard ways of analysing the electorate eg asymmetric age buckets, Britain’s synthetic “Govenrment Office Regions” and reconstituting the electorate into groups that are more coherent (12/14)
For example analysing the North as per a poll is problematic. Merseyside is socialist, Greater Manchester drifts left, the red wall has partially crumbled and Cumberland/County Durham has drifted culturally right, Yorkshire & Lancashire require detailed sub-level analysis (13/14)
If we start analysing all research via a BAME lens I hope it’s done robustly, reported responsibly, and doesn’t result in people/commentators missing the nuance that can easily be hidden underneath a simple but potentially misleading label (14/14)
