Postal Ballots and Survey Bias

There was no “dodgy postal vote spike” in the 2019 Election.

After the 2019 General Election, claims spread that postal voting had surged.

These claims are false. Electoral Commission statistics shows postal voting fell between 2017 and 2019.

The claim originated from an Ashcroft-badged post-election survey. These surveys overestimate postal voting. This systematic error arises due to the survey’s design.

A survey estimate of 38%

On Christmas Eve, Prof Grayling (NCH) wrote there was a “dodgy postal vote spike”. About 3,000 Twitter users shared that post:

Another Twitter post on the same day asserted:

The postal vote was 37%. In previous elections the postal vote never got out of the teens!

The claim comes from the Lord Ashcroft Polls-badged post-election survey¹. In his initial analysis, Lord Ashcroft wrote:

We found 38% saying they had voted by post.

That paragraph was later removed.

Actual counts from the Electoral Commission

The question is: what proportion of voters cast their ballots by post?

How common is postal voting? (Image: Ian Britton/Flickr)

This is different to the number of issued postal ballot papers, as a percentage of the electorate. Some registered electors do not cast a ballot. In 2019, 1.3m UK postal ballots were not returned.

That figure was 18.0% across the United Kingdom in the 2017 General Election. This appears to be basis for claiming “the postal vote never got out of the teens”. That does not answer our question.

To find our answer, we use two measures from Electoral Commission reports:

  • Number of ballot papers returned by postal voters which were included in the count of ballot papers.
  • Ballots at count.
The 2017 figure is the sum of England, Wales and Scotland. (Image: Google Sheets)

The Ashcroft-badged surveys are of Great British voters. For 2017, I calculated the GB statistics using figures for England, Wales and Scotland.

The Ashcroft estimate for postal voting was 36% in 2017. The Electoral Commission data suggests 22.1% of GB counted ballots were postal votes.

For the 2019 General Election, the Ashcroft estimate was higher: 38%. Actual counts find that measure was 21.4% — lower than in 2017.

Across the UK, there were 6.99m counted postal ballots were in 2017 (21.6%). The volumes reduced in the 2019 Election: to 6.72m (20.9%).

Statistical bias

Surveys provide estimates, subject to many sources of potential error.

Those errors can be systematic, leading to consistent overestimation or underestimation. Bias is another name for these systematic errors. This statistical term does not mean or imply political bias.

Why does this bias arise?

The Ashcroft-badged surveys seek a sample of self-reported voters, not all adults. We need to look at the fieldwork dates of these surveys:

  • 2015: Tuesday 5th — Thursday 7th May 2015;
  • 2017: Tuesday 6th — Friday 9th June 2017;
  • 2019: Wednesday 11th — Thursday 12th December 2019.

Readers should recognise these elections were on the respective Thursdays. The fieldwork begins before polling day.

The first question screens for voters. In the 2017 survey, it was:

Q.1 Have you already cast your vote in the General Election which will be held on Thursday 7th May?

Did you vote in the General Election held today across the United Kingdom?

The response options were:

  • Yes, by post;
  • Yes, in person;
  • Yes, by proxy;
  • No.

Only people who say ‘yes’ are part of the final sample. The survey is of voters. Before polling day, the only way to say ‘yes’ is to say you voted by post.

There was a change in methods. The 2015 and 2017 post-election surveys had mixed modes: telephony and internet panels. In 2019, survey responses were all online, with no phone component. People signed up for internet panels may be more likely to vote by post than non-panellists. Also, electors may claim to vote when they have not done so.

This graph for the BES Internet Panel in 2015. (Image: British Election Study)

As a result, the Ashcroft-badged surveys overestimate postal voting in Great Britain. Actual counts show postal voting fell from 2017 to 2019. There was no “dodgy postal voter spike”.

Electoral Commission data comes from their General Election reports. These reports cover 2015, 2017, and 2019. The Ashcroft estimates are from the three post-election surveys: 2015, 2017, and 2019. A Google Sheet page shows the calculations.

¹Lord Ashcroft Polls is not a polling company. It does not have independent capacity to survey the public.

This blog looks at the use of statistics in Britain and beyond. It is written by RSS Statistical Ambassador and Chartered Statistician @anthonybmasters.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store