Archive for the 'Pollsters/polling' Category


New LAB member polling has them rating Corbyn as their most favourable ever

Tuesday, January 21st, 2020

A reminder to LAB members about why so many GE2017 voters defected last month

Mike Smithson


Occasionally 52% of voters have impeccable judgment, here’s the proof

Sunday, December 29th, 2019

52% of voters are correct is not a sentence I expected to write but here we are. This finding isn’t a surprise, after all the star of the film, Bruce Willis, has himself said Die Hard is definitely not a Christmas film, after all film released in July 1988 can hardly be categorised as a Christmas film can it?



GE2019 – the general election that the pollsters mostly got right

Thursday, December 26th, 2019


I don’t want to move on from the general election before just looking back at the performance of the pollsters.

This election was important to them following  the disappointing final surveys that we saw in 2015 and 2017 when with one or two notable exceptions the polls were some way off.

Note  that national polling for Westminster elections generally takes place on a GB basis with Northern Ireland being excluded. So the comparisons should be made with what happened in the GB not the overall national UK vote totals.

As can be seen from the Wikipedia chart above the latest General Election saw the pollsters  getting the results pretty well in the right territory in their final polls. Top firms were Ipsos Mori and Opinium which had the right figures for the main parties.

Mike Smithson


The GE2019 polling race: The pollsters ranked by the CON lead % in their final surveys

Thursday, December 12th, 2019

Mike Smithson


The final polls – almost

Wednesday, December 11th, 2019

Unless there’s another polling disaster Johnson is heading for a workable majority


GE2019 polling and betting update

Tuesday, December 10th, 2019


Your regular reminder that hypothetical polls can be as accurate as an American war film

Sunday, December 1st, 2019

What the hypothetical polls said in October versus what the reality is now.

Hypothetical polls are a lot like Hawaiian pizzas, they should be avoided at all costs by right thinking people everywhere. This piece I wrote back in 2018 explains the systemic flaws with hypothetical polls, and nothing has changed.

As we can see with the two tweets the reality doesn’t bear any resemblance to the fantasy of the hypothetical. Boris Johnson did not die in a ditch and the Tory vote share did not plummet when the United Kingdom did not leave on Halloween.

If Boris Johnson does win his majority and it looks like we might have to extend the transition due to end in December 2020 there will be polling asking what the hypothetical voting intention will be if Boris Johnson extends the transition period.

My hunch is that such polling will be sub-optimal for Boris Johnson and the Conservatives, he would be wise to ignore such hypothetical polling, there’s a long history of it being very inaccurate.

It is also the reason why I’m a little cautious about polling about what will happen to Scottish Independence voting intentions if Boris Johnson wins a majority on December 12th. In the run up to the 2016 EU referendum there was polling showing the SNP and Scottish Independence would surge in the UK decided to Leave the EU, in 2017 the SNP lost around a third of their MPs, and current polling shows they aren’t course to achieve their 2015 high water mark on December 12th.



The latest Ipsos MORI government satisfaction ratings are worse for the incumbent than Major faced just before Blair’s GE1997 landslide

Saturday, November 16th, 2019

I was so taken by David Herdson’s observation in the previous thread header about how poor the current Ipsos MORI government satisfaction ratings that I thought I would dig into the pollster’s huge archive to see if there were historical precdents.

What’s great about the firm which has been polling UK politics since the 1970s is that it has been asking the same questions in the same format over the decades. You can therefore make comparisons.

Going through the the archive poor negative numbers were indicative of a government that was about to lose power.

The big difference between now and 1997, of course, is that the government ratings then were showing a broad picture that was similar to the voting intention ones. That is not happening now.

What it says about what will happen on December 12th I don’t know but it certainly adds to the overall uncertainty.

  • Just in case you think that the current ratings are an outlier they are the best ratings for the Government since Johnson became PM.

    Mike Smithson