Gerard Westhoff critiques Labour’s misuse of statistics
Recent headlines have flaunted the claim that 75 per cent of Labour Party members oppose UK airstrikes in Syria. On the face of it, this statistic seems to show that Jeremy Corbyn’s anti-war stance is that of the people, and that his pro-airstrike colleagues are out of touch with the party membership and general population. But how accurate is this figure?
Leaving party politics to one side, lets look at the methodology behind this statistic, and any potential sources of bias.
Labour’s press release on the figures, states:
“A sample of this weekend’s consultation of Labour Party members, carried out in response to an email from Jeremy Corbyn, issued Friday 27th November, has shown that 75 per cent of Labour party members who have responded oppose UK bombing in Syria.
107,875 responses were received of which 64,771 were confirmed as full individual Labour Party members. The remainder included affiliated supporters and registered supporters.
Random sampling, of full individual Labour Party members who responded to the email, has shown:
75 per cent are against UK bombing in Syria
13 per cent are for UK bombing in Syria
11 per cent are undecided on the issue.”
The key points to focus on here are the phrases “random sampling” and “in response to an email from Jeremy Corbyn”.
Mr Corbyn emailed party members last week to ask for their views on whether parliament should vote for the airstrikes. However, this very email was the first source of bias in the poll, as it outlined his own personal view:
“I do not believe that the Prime Minister made a convincing case that British air strikes on Syria would strengthen our national security or reduce the threat from ISIS.”
The inclusion of his own opinion in the email is instantly likely to encourage those who agree with him to respond to his survey, and for those who are uncertain to be swayed towards his way of thinking. Another phenomena at work here is ‘conformation bias’ – meaning that those who are the most angry at the prospect of the UK bombing another Middle Eastern country, are the most likely to respond.
There are even claims that the email was not circulated amongst all Labour Party members, of which there are around 360,000, leading to suggestions that pro-Corbyn members were more likely to be asked to fill in the survey, meaning a representative sample of Labour views would not be obtained.
The next problem with this statistic is that the 75 per cent figure comes from a random sample of 1900 respondents, out of the 64,771 full party members who voiced an opinion. No details were given on the specifics of how the random sample was chosen – for all we know, the members of Corbyn’s team who read the replies could have selected the emails in such a way as to create a clear majority agreement for Corbyn’s anti-bombing stance (you always have to be inherently suspicious of a figure as round as 75 per cent).
Random sampling is a poor way of getting a general population consensus, let alone random sampling of an already conformation-biased data set. Instead, representative sampling, tailored to include a range of demographics – varying ages, ethnicity, income, gender etc. – would be the ideal way to get a comprehensive cross-section of party member views, and is what most large polling companies do. Why didn’t Corbyn outsource is query to an independent polling organisation? It leads to suspicions that he was simply looking for a belief reinforcing statistic to use as a threat against pro-airstrike Labour MPs.
The point I am making here is not a political one. I’m not trying to prove that Labour Party members are for or against bombing Syria. Politicians and businesses in the modern world are constantly throwing statistics at us, and often these are designed in such a way as to try and bamboozle us into believing anything. Whenever you read a statistic your first thoughts shouldn’t be shock or awe – they should be questions. What is the sample size? How is the sample representative? Are the pollsters biased?
I’ll leave you with Samuel Johnson’s view on statistics, “Round numbers are always false.”
Featured Image: Flickr