Helping Onward understand BAME voters better

Authors: Justin Ibbett, Nik Siersted

The Conservatives have known for a while about their need to truly understand ethnic minorities. A recent event held by the party, featuring think tank Onward, might just represent a turning point.

Onward consists of a number of prominent Conservatives such as Will Tanner, Theresa May’s Former Deputy Head of Policy, and Kate Fall, former Deputy Chief of Staff to David Cameron. At the event, the think tank outlined the Conservative Party’s need to appeal to ethnic minority voters. A recent poll conducted by Ipsos MORI found that only 19% of BAME voters showed support for the Conservative Party, in contrast to 73% who did so for Labour. This is not a new issue. The same poll conducted in 2010 showed only slightly less dramatic figures.

Here at focaldata, using the latest data science techniques, we helped Onward to translate traditional polling analysis into actionable insights in two main ways.

Constituency-level modelling

Firstly, by modelling ethnic minority vote share in each individual constituency, we were able to show exactly which seats were at potential risk, depending on changes in the ethnic composition of the population or variation of BAME votes between the parties. Come election time, the knowledge of where BAME anti-Conservative feeling is most present will be invaluable, as door to door campaigners will be best informed as to where to concentrate their efforts.

Deconstructing the BAME label

In addition, our models were able to unpack ethnic minority opinion from the general BAME label, as used in Ipsos MORI polls, into its constituent groups. This increased the level of granularity, considering the opinion of mixed race, Asian Indian, Asian Chinese, African-Caribbean and many other ethnic minorities separately, allowing for a much deeper understanding of public opinion – one which reflects the disparity of opinion across ethnic minorities. For instance, the campaigns and policies that motivate the British Chinese will have differences from those that motivate British African-Caribbeans.

MRP’s impact on campaign effectiveness

We use a particular statistical technique called multi-level regression with post-stratification (MRP). This involves using regular national surveys to infer opinion in small demographic or geographic groups. It also uses old data or similar surveys to improve its accuracy.

In our view, the granular element of this technique provides a plethora of new insights that in future will come to be considered vital across the political and corporate spectrum.

Gone are the days of companies, or parties alike, developing a marketing program or campaign aimed at the “average customer”. We therefore also work with advertising agencies, helping them to understand public opinion geographically, and big brands in getting more value from their research data.

At focaldata, we’re always happy to see campaigns use data science to improve their understanding of public opinion. As with Onward, we look forward to helping more political clients use our granular data analytics to answer two critical questions needed to run an effective campaign: who do I target and how do I engage them?

Get in touch to know more about how MRP helps campaigns at on, follow us on Twitter at@FocalDataHQ or @JustinIbbett, or catch up with our latest article on MRP-driven insight.