27 April 2017

Statistics and Other Fibs

The election – just too complex.

by J.R.Thomas

Eyes down, look in.  The electoral campaign is underway and the waterboarding of the public by polls and electoral surveys has begun; by the early part of June you won’t believe a word from any of them.  Which is a pity, because the polls are mostly surprisingly accurate.  There are reasons why they may not be at a micro level of course; the person dragged away from watching East Enders by the call from the eager young pollster might tell naughty little lies to tease the pollster, or because they irritably think such information should be kept private; or the pollee in the street might think that other voters behaviour might be influenced by one contending party appearing to be stronger or weaker; polls can indeed encourage changes in voter behaviour by encouraging tactical voting (though most voters won’t admit it, they probably vote as much to keep disliked candidates out as to put favoured ones in).  On the latter point, it is thought that the early success of Bernie Sanders running for the Democrat party Presidential nomination last year grew so rapidly because his rising in the published polls convinced potential voters who were not overly enamoured of Hillary that he had a chance of winning.  So they switched to Bernie.  No doubt the Liberal Democrats will be trying to engineer something of the same sort this time.

Often this does not matter much.  The voters may fib a bit and try to mislead, but all this fibbing and misleading tends to cancel out, leaving the polls reflecting a reasonably true position.  One great exception was 1970, when there seemed to be a genuine change of mind on the part of some voters after very poor employment figures a couple of days before election day, creating a bit of a Heath surge and causing some Labour supporters to stay at home.  But it was not the polls that were wrong – they just did not poll late enough to pick up that change of sentiment and intent.

So you can be reasonably sure that if Mrs May and her Tories are on 45% and Mr Corbyn’s Labour Allsorts on 24% then, if we were voting for a President, Mrs May would have it in the bag. But it is not as simple as that, at all.  Our constituency system guarantees a level of complexity that can turn clear predictions into misleading forecasts and allows elections safe in bags to escape at the last minute.  Just look at Devon and Cornwall. West Country voters have a long history of radicalism which has evolved into a tendency to vote tactically. Historically, the Liberals tended to get the votes of those who could not countenance a Tory MP, or as it may be a Labour MP, but in either case would put up with a Liberal one.  That sort of held together until 2015 when the West Country finally followed its long changing demographics (all those reasonably well off retirees) and went Conservative.  But will the old ways reassert themselves this time with the effects of Brexit intruding into many voters’ calculations?

There will be voters this time who think that it is vital that Britain Brexits, or Bremains, and will answer “Conservative” or “Labour” or Liberal D” to the pollster because, fundamentally, that is what they are.  But in the booth on 8th June they might think Brexiting or Bremaining is a matter of such import that they must hold their nose and vote this one time for the local Corbynista because he and they are a Bremainer and it is vital to Britain’s future.  Or contrariwise and vice versa.  (In fact the danger is perhaps more significant in the vice versa than the other way round – in those northern seats with large “Leave” majorities, the “Remain” supporting Labour MP’s could be in serious danger – though whether the Conservatives or Lib Dems or UKIP will benefit is not so clear.  Maybe the Labour incumbent will slip through anyway as the vote divides in every direction….)

Will UKIPpers abandon the purple and move red to let Labour in? Or decide that true blue Mrs May’s Tories are not so bad after all? (Those of our readers with long memories will remember Basildon, a relatively safe Labour seat, going Conservative in the Thatcher days, won by one Harvey Proctor whose views were not that remote from modern UKIP.)  Will SNP voters think that, given a majority of Scots are still against independence in spite of Ms Sturgeon’s great efforts to persuade them otherwise,  the best result might be an SNP government in Edinburgh, but a Tory/Labour/LibDem one in Westminster; and if so, which?

Which is all to say that there are so many considerations this time, that who knows what voters will think, and do, in early June.  The national polls are not going to reflect that very well; they are just going to record an aggregation of what people are telling them, and not pick up all those local strategies.  For this reason local constituency polls will be much more interesting and better predictors, so the keen student of political prediction may wish to study key local polls to get a true feel as to what might happen.

Certainly Mrs May is taking a risk, maybe more than she might think.  One issue elections are liable to stray badly off message.  Think of Ted Heath in February 1974.  “Who runs Britain?” Ted demanded, “the Government or the miners?”  There could only be one answer thought Ted and the cabinet.  Alas, the voters had not read the briefing paper and put Harold Wilson back in, to do a deal with the miners and get the electricity back on.  The issue that matters so much to the politicians – or the one who called the election – may not matter that much to the electors who might care more about taxing the rich or an independent Scotland or staying in Europe.

And Mrs May is also kicking off at a self-imposed disadvantage.  We have alluded in these pages before to the Boundary Commission and its slow progress to redrawing a number of constituency boundaries. This, when concluded, is likely to give the Conservatives around an extra twenty seats, simply because there are a number of overlarge Conservative safe seats which will be divided, and a number of undersized Labour seats which will be merged.  Slow (painstaking would be politer) the Commission may be, but they will report in the autumn of 2018 and the proposals will be implemented in 2019.  For a politician of Mrs May’s natural caution, it is a brave thing to start a potential twenty seats down.  Mrs Clinton could warn on what happens next – you win the numbers game but don’t get the coconut.  This could be a real risk if there is a lot of tactical voting by Remainers, for example, with some prime Tory seats falling to the Lib Dems and Labour seeing its majorities fall in those safe northern seats, but not enough to lose them.

Even if the public don’t believe the opinion polls, the politicians and all those clever graduates who do the electoral grunt work know how to read them.  They will be watching very carefully what the pollsters report, particularly about which issues matter to the electors and what is happening in key constituencies.  The Shaw Sheet will not be predicting the outcome of the election (yet) but we suspect that Theresa and the more popular members of her cabinet (not you, Mr Hammond) will be spending some time in marginal Labour and SNP seats that voted to Leave, and that that chap who leads the Liberal Democrats (must write his name on a post-it note) will be in south-eastern seats with lots of Remainers, and that Mr Corbyn (probably on his own) should be (should be, but we hesitate to predict anything about Mr C) in seats where Labour could see the Tory vote split and Labour come through the gap. Eyes down…

If you enjoyed this article please share it using the buttons above.

Please click here if you would like a weekly email on publication of the ShawSheet

Follow the Shaw Sheet on

It's FREE!

Already get the weekly email?  Please tell your friends what you like best. Just click the X at the top right and use the social media buttons found on every page.

New to our News?

Click to help keep Shaw Sheet free by signing up.Large 600x271 stamp prompting the reader to join the subscription list