The best estimates of how Britain’s ethnic minorities voted in the 2010 election, taken from the Ethnic Minority British Election Study, are CON 16%, LAB 68%. Last month British Future released a report, based on Survation polling, that suggested that ethnic minority voters in 2015 split CON 33%, LAB 52%.
This would represent a huge turnaround – a doubling of Conservative support amongst ethnic minorities and a drop of sixteen points for Labour. However, it is quite difficult to believe, or to tally with the actual election results we saw. A sixteen point swing from Lab to Con would be stunning (to put it in context, the 1997 swing from Con to Lab was ten points), movements of that scale can happen (look at Scotland), but they are hardly commonplace. And ethnic minority voters are highly concentrated geographically, if there had been such an outlandish movement from Lab to Con amongst BME voters we should have expected to see seats with a high proportion of ethnic minority voters swing disproportionately towards the Conservatives – we didn’t (it was the exact opposite). We would have expected the more ethnically diverse London to swing more towards the Conservatives – it didn’t, it swung much more heavily to Labour.
So what explains the difference? I expect it’s simply down to comparing apples to oranges. Polling ethnic minority voters is a hard challenge. Ethnic minority Britons are likely to be younger, less affluent and often live more transient lifestyles – all things that make groups harder to poll. Recent immigrants (and even some longer term residents) may speak English as a second language or not at all, which poses a problem for surveys. And of course, if a poll under-represents less affluent, less established, integrated and English-speaking minority communities and over-represents those who have been here for generations it may well misrepresent their voting intentions.
The 2010 Ethnic Minority BES (EMBES) was conducted using a proper stratified random sample. For reasons of cost it was limited to areas of comparatively high ethnic minority concentration, but the cut off was very low (it excluded areas with less than 2% ethnic minorities, where 12% of ethnic minorities in Britain lived). They contacted 31,000 randomly selected addresses in order to find 4224 eligible respondents and amongst them managed a response rate of 66%. Doorstep translation cards were available in Punjabi, Urdu, Hindi, Gujarati and Bengali, and other household members were allowed to act as translators if the person selected couldn’t speak English (there were translated versions of the survey available to assist). The British Future poll was done using an internet panel so would have covered ethnic minority voters living in all parts of Britain and all ethnic minority groups (the EMBES concentrated on Black African and Caribbean, Pakistani, Indian and Bangladeshi) but only those integrated enough into British society to have joined internet panels doing surveys in English.
Any difference between the two polls therefore may just as likely be from the radically different ways the polls were conducted as from an actual shift in voting behaviour. What we need in order to be confident is to compare like-with-like, data from 2010 and 2015 that was collected in the same way. Over on the YouGov site they have some analysis by Rob Ford, Laurence Janta-Lipinski and Maria Sobolewska comparing YouGov’s ethnic minority vote in 2010 and 2015. As with Survation’s internet poll, YouGov’s internet poll found higher levels of Conservative support amongst ethnic minorities than in the EMBES… but crucially, if you compared their 2010 figures to their 2015 figures there was only modest movement towards the Tories, it suggests a swing from Lab to Con amongst ethnic minority voters of 2 percent, rather than 16 percent.
We can do a similar thing with Ipsos-MORI’s data – after every election they publish a big aggregate of their data from the campaign, so we can compare their breakdown amongst ethnic minority voters in 2015 with that in 2010. Similar to the comparison in YouGov data, it shows some movement towards the Conservatives amongst BME voters, but only a modest one – MORI data suggests a 1 percent swing.
So the Conservatives do seem to be making some progress amongst ethnic minority voters… but it’s probably only a modest advance, as yet the huge Labour advantage amongst BME voters remains almost as large as it was at previous elections.