Only 6 left for Leave

Research undertaken for Best for Britain and Hope not Hate shows how the move to Remain plays out by constituency. London4Europe Committee member Michael Romberg writes.

You will have seen the press report in the Observer of 12 August 2018 that a majority of constituencies would now support Remain. The information has been presented by Best for Britain in a map. You can click on your constituency to find the change in opinion since then. No seat moved from Remain to Leave, though support for Leave strengthened in some places.

This table shows the ten Greater London constituencies that moved from Leave to Remain on the Best for Britain figures:

That means that only the following Greater London constituencies are still for Leave on the Best for Britain figures:

Bexleyheath & Crayford

Dagenham & Rainham

Hornchurch & Upminster

Old Bexley & Sidcup

Orpington

Romford

You can find more information on the 2016 referendum including estimates of constituency votes on Wikipedia.

A note on data quality

I cannot find any data tables or full methods statement published by Best for Britain, Hope not Hate or their data analytical firm Focaldata. That means that one cannot assess the quality of the data.

The referendum was not counted by constituency in London. Some returning officers published constituency or ward figures and others did not. Therefore most of the constituency figures for 2016 are estimates. The estimates used for 2016 are mostly those by Hanretty, an academic. You can read his article about the difficulties of creating the figures here.

To check data quality he looked at a selection of seats that were co-terminous with local authority areas – which meant that his methodology could be compared with actual results. 40% were within one percentage point of the right answer. A further 40% were between one and three percentage points of the right answer (that means that if his method said the Leave vote was 60% then in 80% of cases the true figure would be in the range 57%-63%.)

As a rule of thumb, the headline numbers in a large, properly structured opinion poll come with a margin of error of +/- three percentage points. More detailed analyses, looking at how subsets of the sample voted, come with larger margins of error. So a poll with a headline number saying 55R:45L might just as well be 52R:48L or 58R:42L. Numbers at more detailed levels, such as constituencies, would come with a larger margin of error.

So just as the fairest way to describe where we are in the national opinion polls is that the country is still divided half and half (and we need to think of the reasons for that, in spite of all the bad news about Brexit) it is probably safest not to put too much weight on the constituency numbers.

Articles on this page reflect the views of the author and not necessarily those of London4Europe.