Explaining MRP

We have arrived at the new constituency estimates for Remain/Leave voters using a statistical method called Multi-level regression with poststratification (MRP), which has been undertaken for HOPE not hate and Best for Britain by Focaldata.

MRP is a statistical technique for estimating public opinion in small geographic areas or sub-groups using national opinion surveys. It originated in America, Columbia University to be precise, and was used by academics to estimate state-level opinion cheaply, given the expense of doing polls throughout the country.

MRP has two main elements. The first is to use a survey to build a multi-level regression model that predicts opinion (or any quantity of interest) from certain variables, normally demographics. The second is to weight (post-stratify) your results by the relevant population frequency, to get population level (or constituency level) estimates.

At the end of this process you get more accurate, more granular (thus more actionable) estimates of public opinion than traditional polling. There are however significant technical challenges to implementing it effectively. These include large data requirements, dedicated cloud computing resources, and an understanding of Bayesian statistics. It is far from a simple endeavour. 

Among the key features are:

• Estimates of opinion at local geographic levels, e.g. constituencies
• Estimates of opinion of precise demographic groups, e.g. young, women who are Asian, own their own house and drive a bicycle to work
• More accurate estimates of opinion generally because MRP doesn’t exacerbate problems of unrepresentative samples 
• Use of prior knowledge and old polls to improve accuracy and potentially cut sample size
• Quicker decisions as new data automatically flows through models to campaigners

Between 28 June and 6 July and 26 and 31 July, YouGov polled 15,340 people on behalf of HOPE not hate and Best for Britain with a range of questions about Brexit and other political and cultural issues and over the next few weeks we will be releasing more data.

However, the key value in MRP is that it will help us be far more effective in our local campaigning. MRP helps us better understand who we need to speak to on any given issue, how we should speak to them and on what issues.

The data will guide our door-to-door campaigning, but also but direct our online campaigning as we will be able to customise our messages to specific audiences far more effectively.