Last week an Independent article claimed
The article began
It is “very likely” that the UK voted for Brexit because of illegal overspending by the Vote Leave campaign, according to an Oxford professor’s evidence to the High Court.
Professor Philip Howard, director of the Oxford Internet Institute, at the university, said: “My professional opinion is that it is very likely that the excessive spending by Vote Leave altered the result of the referendum.
“A swing of just 634,751 people would have been enough to secure victory for Remain.
“Given the scale of the online advertising achieved with the excess spending, combined with conservative estimates on voter modelling, I estimate that Vote Leave converted the voting intentions of over 800,000 voters in the final days of the campaign as a result of the overspend.”
Is the estimate conservative? Anthony Masters, a Statistical Ambassador for the Royal Statistical Society, questions the statistics in the Spectator. The 800,000 was based upon 80 million Facebook users, 10% of whom clicked in on the advert. Of those clicking, 10% changed their minds.
Masters gave some amplification on in a follow-up blog post Did Vote Leave’s overspending cause their victory?
The reasons for doubting the “conservative” figures are multiple, including
– There were not 80 million voters on Facebook. Of the 46 million voters, at most only 25.6 million had Facebook accounts.
– Click through rate for ads is far less than 10%. In UK in 2016 it was estimated at 0.5%.
– Advertising is not the source of campaigning. It is not even viewed as the primary source, merely bolstering other parts of a campaign through awareness and presence.
– 10% of those reading the advert changing their minds is unlikely. Evidence is far less.
Anthony Masters concludes the Spectator piece by using Professor Howard’s own published criteria.
Prof Howard’s 2005 book, New Media Campaigns and the Managed Citizen, also argues that we should apply a different calculation to that submitted to the High Court. His book says to apply a one per cent click-through rate, where 10 per cent “believe” what they read; and of that 10 per cent act. This ‘belief’ stage appears to have been omitted in the High Court submission’s final calculation. Using these rates, this calculation turns 25.6 million people into 2,560 changed votes – hardly enough to have swung the referendum for Leave, given that their margin of victory was over a million votes. If we share a belief in accuracy, this erroneous claim should have limited reach.
There is further evidence that runs contrary to Prof Howard’s claims.
1. The Polls
To evaluate the statistical evidence for a conjecture – particularly for a contentious and opinionated issue like Brexit – I believe one needs to look at the wider context. If a Facebook campaign swung the Referendum campaign in the final few days from Remain to Leave, then there should be evidence of a swing in the polls. In the blog article Masters raised three graphs based on the polls that contradict this swing. It would appear that through the four weeks of the official campaign the Remain / Leave split was fairly consistent on a poll of polls basis. From analysis by pollster YouGov, the Leave share peaked on 13th June – ten days before the referendum. The third graphic, from a statistical analysis from the LSE, provides the clearest evidence.
The peak was just three days before the murder of MP Jo Cox by Tommy Mair. Jo Cox was a Remain campaigner, whilst it was reported that the attacker shouted slogans like “Britain First”. The shift in the polls could have been influenced by the glowing tributes to the murdered MP, alongside the speculation of the vile motives a clear Vote Leave supporter. That Jo Cox’s murder should have had no influence, especially when campaigning was suspended as a result of the murder, does not seem credible.
On Twitter, Anthony Masters also pointed to a question in Lord Ashcroft’s poll carried out on the day of the referendum – How the United Kingdom voted on Thursday… and why to a graphic that looked at when people had decided which way to vote. At most 16% of leave voters made up their minds in the last few days, slightly less than the 18% who voted remain.
The same poll looked at the demographics.
This clearly shows the split by age group. The younger a voter the more likely they were to vote Remain. It is not a minor relationship. 73% of 18-24s voted for Remain, whilst 40% of the 65% voted. Similarly, the younger a person the greater the time spent on social media such as Facebook.
2. Voting by area
Another, aspect is to look at the geographical analysis. Using Chris Hanretty’s estimates of the EU referendum results by constituency, I concluded that the most pro-Remain areas were the centre of major cities and in the University Cities of Oxford, Cambridge and Bristol. This is where the most vocal people reside.
The most pro-Leave areas were in the minor towns such are Stoke and Boston. Greater Manchester provided a good snapshot of the National picture. Of the 22 constituencies is estimated that just 3 has a majority remain vote. The central to the City of Manchester. The constituencies on the periphery voted to Leave, the strongest being on the east of Manchester and a few miles from the city centre. Manchester Central contains many of the modern flats and converted warehouses of Manchester. Manchester Withington has a preponderance of media types working at Media City for the BBC and ITV, along with education sector professionals.
These are the people who are not just geographically marginalised, but often feel politically marginalised as well.
Concluding comments
Overall, Professor Howard’s claims of late Facebook posts swinging the Referendum result are not credible at all. They are about as crackpot (and contradict) as the claims of Russian influence on the Brexit result.
To really understand the issues one needs to look at the data from different perspectives and the wider context. But the more dogmatic Remainers appear to be using their power and influence – backed by scanty assertions -to thrust their dogmas onto everyone else. This is undermining academic credibility, and the democratic process. By using the courts to pursue their dogmas, it also threatens to pull the legal system into the fray, potentially undermining the respect for the rule of law for those on the periphery.
ManicBeancounter 