Josh writes: One of the main things we learned from the recent UK General Election was that the forecasters got it catastrophically wrong – catastrophic in that the pollsters reputations are now in shreds. The collective narrative was that it had to be a hung parliament, nothing else was possible – even Nate Silver agreed so it had to be true.
Yet how wrong they were.
Not everyone was wrong – the Telegraph’s Dan Hodges got it about right, as did Janet Daley, also at the Telegraph, and I am sure there were others.
It is horriblly like that other consensus – the one that always has to be 97% and which we all know is also catastrophically wrong.
H/t to Paul Matthews whose excellent blog post about the election results has the reference to voters hiding in the deep ocean.
Silly – The polls (and predictions) were exactly correct. They were accurate to the hundredth decimal point. It was the voter’s (deniers) that screwed up and voted incorrectly. The parallel with climate disaster projections is astounding! GK
G. Karst
Actually, the polls WERE “correct” to within their measurement error. The polls showed that Labour and the Tories would have similar share of the national vote. In the event Labour won 31% and the Conservatives 37% of the national vote. These are very similar results to within the +/-3% margin of error. However, on this occasion the small difference provided a misleading indication of the coming election result.
The indication was misleading because the polls were indicating national share of the vote and the UK’s First Past The Post electoral system provides seats to the largest share of the vote in each individual constituency.
The effect of this was most clear in Scotland. The Scottish Nationalist Party (SNP) obtained 6% of the national vote but won 56 seats because that 6% was all concentrated in the 59 Scottish seats and, therefore, in most of those individual Scottish seats the SNP obtained more than 50% of the vote.
The Conservatives obtained an overall majority of seats in the Commons because the United Kingdom Independence Party (UKIP) obtained 15% of the national vote but this share was distributed over the whole of England and Wales. Very importantly, UKIP took more votes from Labour than from the Conservatives especially in the North and Midlands of England. Thus, UKIP won only one seat but provided the Conservatives with many seats that were Labour seats in previous elections.
UKIP is now saying it hopes to displace Labour as one of the two major parties.
In previous recent General Elections the national share of the vote had been a useful proxy for the share of the vote in almost all seats, and it was assumed it would be in this election. This assumption was wrong because the SNP had a landslide in Scotland, the LibDems collapsed, and UKIP obtained 15% of the votes and took more votes from Labour than from the Tories.
In summation, the polls WERE “correct” to within their measurement error but they were measuring the wrong thing to indicate the election result.
Richard
Richard
Your explanation of the reasons why the polls leading up to the election were wide of the final result seems far more likely than the other theories being put about, mainly by the pollsters who got it so wrong.
Are you aware that the British Polling Council (I hadn’t heard of it before either!), which I believe is headed by Professor John Curtice, the BBC exit poll man, is to hold an enquiry into what went wrong? Maybe you should contact them to see what they think.
Mr Green Genes
I was not aware of the British Polling Council, thankyou.
I posted a summation in this thread of a more detailed analysis I provided on WUWT here on the morning after the election.
I will try to find out about the “enquiry into what went wrong”. I will submit to them if I can and I will post the submission to here.
Richard
Mr Green Genes
I have made the following submission to the enquiry by email
Richard
Subject: A submission to the BPC enquiry into apparent failure of Election predictions
Dear Professor Curtice:
I write to make a submission to British Polling Council (BPC) enquiry into apparent failure of recent General Election predictions by opinion polls.
I submit that the error was in interpretation of the results.
In previous recent General Elections the national share of the vote had been a useful proxy for the share of the vote in almost all seats, and it was assumed it would be in the 2015 General Election. This assumption was wrong because the SNP had a landslide in Scotland, the LibDems collapsed, and UKIP obtained 15% of the votes and took more votes from Labour than from the Tories in England.
The polls WERE “correct” to within their measurement error. The polls showed that Labour and the Tories would have similar share of the national vote. In the event Labour won 31% and the Conservatives 37% of the national vote. These are very similar results to within +/-3% margin of error. However, on this occasion the small difference provided a misleading indication of the coming election result.
The indication was misleading because the polls were indicating national share of the vote and the UK’s First Past The Post electoral system provides seats to the largest share of the vote in each individual constituency.
The effect of this was most clear in Scotland. The Scottish Nationalist Party (SNP) obtained 6% of the national vote but won 56 seats because that 6% was all concentrated in the 59 Scottish seats and, therefore, in most of those individual Scottish seats the SNP obtained more than 50% of the vote.
Prior to the election, Labour had more seats in Scotland than the Conservatives so the SDP landslide increased the difference in number of Labour and Conservative seats.
The Liberal Democrats (LD) lost many seats but the lost LD share of the vote seems to have been lost to both Labour and Conservatives with little net effect on.
The Conservatives obtained an overall majority of seats in the Commons because the United Kingdom Independence Party (UKIP) obtained 15% of the national vote but this share was distributed over the whole of England and Wales. Very importantly, UKIP took more votes from Labour than from the Conservatives especially in the North and Midlands of England. Thus, UKIP won only one seat but provided the Conservatives with many seats that were Labour seats in previous elections.
UKIP is now saying it hopes to displace Labour as one of the two major parties.
In summation, the polls WERE “correct” to within their measurement error but they were measuring the wrong thing to indicate the election result.
I am providing a public record of this submission here
http://wattsupwiththat.com/2015/05/11/consensus-smensus-97-of-uk-voters-hiding-in-the-deep-ocean/#comment-1932953
Richard S Courtney
Ross McKitrick debunks the 97%.
http://business.financialpost.com/fp-comment/climate-change-consensus-among-the-misinformed-is-not-worth-much
And the next step is to pay the bill to the masters in Brussels, the money sink of Europe.
It is so frustrating that conventional terminology doesn’t allow one to accurately describe the likes of David Cameron. Applying the left-right spectrum to 21st century politicians is detached from current realities and accordingly is misleading and useless.
It is impossible to express what unifies these three dreadful men and their non-British equlvalents when the terms available place misleading and inaccurate emphasis on the differences between them. In the absence of terms which reflect the world as it is, purposeful debate about how things are and how they should change cannot happen.
An exploration of why Left and Right have been sucker terms for at least half a century:
https://chiefio.wordpress.com/2011/01/20/nationalist-socialists/
There have been several articles trying to explain the way a certain percentage of voters underestimated their conservatism to the pollsters. In a nutshell: the cultural hegemony of the left makes people ashamed to express their views, but in the polling booth, they do. To its credit, one of these articles is in the Guardian:
http://www.theguardian.com/commentisfree/2015/may/10/kissed-shy-tory-polls-labour-left-stop-demonising
Here’s another: http://www.kiwiblog.co.nz/2015/05/the_rise_of_the_shy_tory_why_pollsters_are_missing_voters_on_the_right.html
pdxrod
“Shy Tories” who don’t exist do not explain a polling error that did not occur.
Please see my above post here.
Richard