Ain't No Such Thing As 'Swing'!

A regular feature of the BBC election night broadcast for many years was Peter Snow and his ‘swingometer’. The swingometer was supposed to show how the total number of votes for each party, as they were being counted, represented a national ‘swing’, either towards Labour or towards the Conservatives.
This became more complicated when other parties had to be brought into the equation: how much of a ‘swing’ towards, or away from, the LibDems was involved in an overall ‘swing’ towards, or away from, Labour or the Conservatives?
There has also, for many years, been a concept in the academic study of Politics known as the ‘swing voter’. Since most people in the country support one particular party and continue to do so, through thick and thin, for most of their adult life, there needed to be an explanation for why one one party sometimes won an election and sometimes another party did. This was solved by introducing concept of the ‘swing voter’, an imaginary person who does not fit the usual pattern of supporting only one political party, but instead flits about like a butterfly – settling on one party for one election and then settling on another for the next election. These ‘swing voters’ were thus the key to understanding why one party would win over another.
Both of these theories, however, are highly suspect and do not stand up to the evidence. Even the most cursory look at the election results from one constituency to the next reveals that there is nothing uniform or consistent about voting patterns in different constituencies. In one constituency, the ‘swing’ may be from Conservative to Labour, while, in the next constituency at the same election, the ‘swing’ is from Labour to Conservative.
What purports to be a national ‘swing’ is nothing but an averaging out of lots of little ‘swings’ that may be moving in different directions all across the country. But it is also more complicated than that.
If we look more closely at a single constituency, we may see that the Tories increased their majority over Labour by a certain amount, and this appears as a ‘swing’ from Labour to Conservative. However, a third party –Greens or Plaid Cymru, for instance – may have gained more votes than they had previously, at the expense of Labour. The Tory vote might even have gone down from the previous election, but because votes for another party lowered the total Labour vote, even a reduced Tory vote could result in an increased Tory majority.
This is not just a hypothetical example, but happened over and over again in constituencies around the country in 2015, as it has in previous elections. Because the Tory majority was increased, it appears as a ‘swing’ from Labour to Tory. And yet the reality is, in some of these cases, more like a ‘swing’ from Labour to Green, or Labour to Plaid Cymru, or to whoever.
There is also another challenge when it comes to the concept of ‘swing’. Even at the constituency level, what appears to be a ‘swing’ from one party to another may be the result of who votes as much as it is a result of which party they actually vote for. Since it is a secret ballot, we can never know for sure how any one person has voted from one election to the next. All we have is the aggregated result of many people voting at a particular election, in a particular constituency (or ward).
In 2017, more than 14 million registered voters did not vote. That is significantly more than voted for either Tory or Labour candidates in that election. At the constituency level, this translates to a huge pool of voters who do not vote in any particular election.
Is a ‘swing’ from Labour to Conservative in a particular constituency the result of ‘swing voters’ changing their allegiance from Labour to Conservative; or is it the result of Labour voters ‘swinging’ towards the Green and thus giving the Conservatives a larger majority; or is it the result of more Tory voters deciding to vote in that election than they did in the previous election; or perhaps fewer Labour voters deciding to vote?
There are many factors involved when it comes to why people vote the way they do – or why they vote at all. How this translates into a majority for a particular candidate in a particular constituency is also unique to that constituency. Every constituency has its own story to tell when it comes to explaining a particular election result. And these cannot be simply added together into any kind of uniform pattern across the country.
This is not the only reason, but certainly one of the reasons why the pollsters got it wrong at the 2017 general election and why they got it wrong at the EU referendum. And it could well be that they have got it wrong in 2019 as well…