General Election 2019 Stat Checks: Week Three

The campaign continued, as did the questionable claims.

In the third week of campaigning for the 2019 General Election, there were many questionable claims by politicians, journalists and campaigners.

The Invisible Cut

Claim: Labour cuts Conservatives’ poll lead to eight points.
Rating: False.
Reasoning: On the Independent website, the article cites a poll by BMG Research, which was sponsored by the news organisation:

Labour has cut the Conservatives’ lead to just eight points in the second week of the general election campaign, according to a poll for The Independent.

The internet panel poll was conducted between 12th and 15th November, and surveyed 1,506 adults in Great Britain.

The article and its headline are comparing that poll to “other polls in recent days” by different polling companies — rather than the same polling series. There are systematic differences between different polling companies, meaning their vote intention estimates should not be compared like this.

In the prior vote intention poll by BMG Research (on 5th to 8th November), the lead was also eight points — it was not “cut”, and is false to claim so.

A £100bn tax cut?

Claim: The Conservative government has given away £100bn in tax cuts.
Rating: Likely to be misleading.
Reasoning: Labour’s Shadow Chancellor John McDonnell claimed that:

£100bn has been given away in tax cuts.

This figure was supposedly received by “corporations and the wealthy”. In functional and accounting terms, a tax cut does not mean money is given away — merely less is paid.

As Channel 4’s Fact Check have identified, this figure is not current, despite the wording: it is based on the period from 2010/11 to 2023/24, from the OBR Policy Measures database.

“Over £86bn” of the total amount is from corporation tax. That statistic also shows the gross change: not accounting for tax rises that affect corporations. This is not the net change: which is £7.2bn over that same period.

Highest in Europe?

Claim: Labour’s proposed corporation tax rate would be “the highest in Europe”.
Rating: False.
Reasoning: At the Confederation of British Industry’s annual conference, the Prime Minister Boris Johnson claimed:

The alternative is Jeremy Corbyn — who would whack it straight back up to the highest levels in Europe.

This is false. Labour’s proposed main corporation tax rate would be 26% — as it was in 2011. The OECD currently lists Belgium (29%), France (32%), Greece (28%), and Portugal (30%) as having higher main rates.

As the Institute for Fiscal Studies identifies, Labour are implying their government will raise the highest corporation taxes (as a share of GDP) among the G7 countries:

This is a different measure to talking about the main rate. (Image: IFS)

Columns of debt

Claim: Labour borrowed £500bn in the last 33 years.
Rating: False.
Reasoning: A Labour councillor shared a graph, which showed:

Labour [Total borrowing of all Labour governments in the last 33 years]: £500bn
Conservatives [Total borrowing of this Conservative government in the last 9 years]: £770bn

Clearly, a group called ‘Collective Voice’ originally created this graph. (Image:

The figures here do not line up with sums of public sector net borrowing (excluding public sector banks), according to the Office for National Statistics. In April 1997 to April 2010, it was: £524.5bn. In May 2010 to September 2019: it was £855.5bn.

That might not be the right way, to look at it either — given debt can be repaid. Let’s look at public sector debt expressed as a proportion of national GDP:

The debt reference value is 60%. (Image: ONS)

There is context needed too. The Labour government mostly ran deficits under 3% of GDP during its time in office. When the financial crisis happened, this deficit increased to 10% in 2009/10 — both as GDP shrunk and because of counter-cyclical and structural borrowing.

From this high point, the government has been slowly reducing the annual deficit — meaning that the deficit has only been under 3% in the last three financial years.

The labelling is strange too. In the last 33 years, there has only been a Labour government for 13 of them.

Self-selecting social surveys

Claim: Other polls shows Jeremy Corbyn as the clear winner of the ITV debate.
Rating: False.
Reasoning: These are not ‘polls’ — conducted by social research companies — but self-selecting clickable surveys on social media.

Martin Lewis is not a polling company.

Self-selecting social media surveys are not reliable measures of public opinion. Due to self-selection bias, we cannot make any inferences from these results to the general population.

Twitter surveys have no controls, no weightings and no credibility. (This will be the subject of a longer post next week.)

Cynics and undecided voters

Claim: Labour’s rise in 2017 came from undecided voters.
Rating: Needs clarification.
Reasoning: At the Left Foot Forward blog, a writer states:

In fact, those votes were there the entire time, but the polls didn’t show them. Why? They came from ‘don’t know’ respondents.

The British Election Study analysis of the BES internet panel found people saying Don’t Know switching to Labour is only a partial explanation.

Some voters also switched from other parties to Labour during the campaign.

The Labour campaign pulled back voters, including from those undecided at the start. (Image: BES)

There are other issues with the article. The author refers to a ‘sample population’, which is an oxymoron: we draw samples to make inferences about populations. Samples of 2,000 people as erroneously described as “small”.

The author writes:

I assume that the proportion of Don’t Knowers is going up on this increasingly divided and shivering isle; we live in the age of voter volatility.

We can look at the data tables. In YouGov’s polling, the Don’t Know rate for its survey on 16th to 17th May 2017 was 16%. For its poll on 17th to 18th November 2019, that rate was 17%.

For this company, the rate is pretty similar at similar times.

Additionally, the article falsely claims that: “Polls don’t show the data on these undecided voters”. Headline vote intention estimates are typically presented excluding undecided voters (or imputed).

In this house, we make proportionate bar graphs

Claim: In 2017, the Conservatives had less than 50% of the total vote in Chippenham, and the Liberal Democrats had more than half of the Conservative total votes.
Rating: False.
Reasoning: In the 2017 General Election in Chippenham, the Conservative candidate received 54.7% of the vote. The Liberal Democrats received 25.6%, and Labour got 19.7% of votes cast.

This was created in R Studio Cloud, using the ggplot2 package. (Image: R Pubs)

According to the Chippenham Liberal Democrats, that looks like:

This is a plain misrepresentation. On their graph, the Liberal Democrat bar is closer in height than to the Conservative one (a difference of 29.1 points) than the Labour one (6.0 points, due to rounding).

Registrations and applications

Claim: More than 7% of the electorate have registered to vote since the General Election was called.
Rating: False.
Reasoning: Eoin Clarke, a Labour campaigner, claimed that:

2,298,013 have now registered to vote this election… That’s more than 7% of the electorate & enough to skew the polls.

The application figures refer to 29th October to 21st November. In December 2018, there were 45.8m UK Parliamentary electors. This means 2.3m applications expressed as a proportion of the electorate is 4.8% — not “more than 7%”.

Some applicants may already be registered, or are changing address details. From December 2016 to December 2017, the electorate is increased by 380,000. Between 18th April and 22nd May 2017, there were over 2.9m applications to register to vote.

A large pike occurred at the end of the pre-election registration period. (Image: GOV.UK)

Opinion polls ask about vote registration, rendering Dr Clarke’s assertion that registrations “skew the polls” nonsensical.

Earners in the top 5%

Claim: If you earn £80,000 a year, you are in the top 5% of earners.
Rating: Partially true.
Reasoning: An audience on Question Time said that he was “nowhere near in the top 5%, let me tell you, I’m not even in the top 50%”, despite having an income of over £80,000. The BBC’s Reality Check team looked at this matter.

If we only look at income taxpayers (those above the personal allowance threshold), then the HMRC Survey of Personal Income for 2016/17 estimates:

Median income taxpayer: £23,600
95th percentile: £75,300

To reiterate, this is only looking at income taxpayers — those employees with incomes above the personal allowance threshold.

Both the HMRC estimates and the ONS Annual Survey of Hours and Earnings (ASHE, with covers employees) show that the employee income distribution has a long right tail, with few jobs having a high income. In the jargon, the distribution is ‘right-skewed’.

This is the proportion of jobs within 20p of the stated rate. (Source: ONS)

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