Archive for the 'Pollsters/polling' Category


Here’s a funny thing. Polls that carry out fieldwork online are 3-9 points more favourable to the “big 2″ than phone surveys

Sunday, November 23rd, 2014

The big methodology difference is in sampling. The phone firms used randomised dialling and theoretically anybody with a landline, and now mobile, can be included.

Online polls are generally carried out amongst members of a polling panel who perhaps are slightly different from the electorate as a whole in they’ve signed up in the first place and are doing the survey for money.

It was only when I was looking through recent findings from the different firms that I found the split featured in the chart above.

Mike Smithson

2004-2014: The view from OUTSIDE the Westminster bubble


New Survation/UNITE poll has CON holding onto Stockton S where it’s defending a majority of just 310

Wednesday, November 19th, 2014

Finally for tonight

UPDATE: CON back in lead with YouGov


What LAB voters like about their party: Some surprising Opinium findings that conflict with the media narrative

Tuesday, November 18th, 2014


Survation finds dramatic LAB collapse in Scotland but not on the scale of Ipsos-MORI in October

Tuesday, November 18th, 2014

What we need now are Scottish constituency polls

The second part of the Daily Record Survation poll of Scottish voters was published overnight and finds a big increase in SNP support since the IndyRef with an even bigger drop in the Labour share.

It doesn’t really need to be said that the prospect of losing a significant part of its current base of 41 seats North of the border is going to make Labour’s challenge at the General Election that much harder.

    On some computations these latest numbers would leave Scottish LAB with just five seats out of the 41 it currently has.

That might sound an awful lot but it is far better than the same calculations that were being made three weeks age when the Ipsos-MORI quarterly Scottish poll had the SNP on 52% with LAB on 23%. Then the talk was of just one Scottish LAB MP remaining.

So in one sense today’s Survation poll could have been worse for LAB. It wasn’t as bad as Ipsos-MORI.

The best way of working out the impact in terms of seats is through single constituency polling which Lord Ashcroft had said he had in the pipeline. He asks a two stage voting question with the second requesting those sampled to focus on the specifics of their own seats. And as we have seen this can make a big difference.

One thing of course to bear in mind is that the loss by Labour of Scottish seats to the SNP has no impact on the Tory ambition of winning an overall majority. That requires the blues to win seats not LAB to lose them to another party.

As I wrote yesterday full Scottish polls have been something of a rarity since the IndyRef. This is just the third so we have not got too much data to base things on. But all three surveys have pointed to catastrophic LAB losses the only differences between on the scale of what might happen.

And to reiterate Lord Ashcroft’s regular caveat – polls are a snapshot not a prediction.

Mike Smithson

2004-2014: The view from OUTSIDE the Westminster bubble


For the first time in a month the Ashcroft National phone poll has LAB in the lead

Monday, November 17th, 2014

Ashcroft becomes the 4th pollster in a row to have LAB ahead

All the movements are very small and well within the margin of error but it will come as a relief in Miliband towers that the national VI polls seem to be moving back to LAB.

The shares vary considerably across the firms no more so than today. Just look at the chart to see the very real differences between Ashcroft and Populus – the latter having a CON+LAB aggregate that is is five points up on the 2010 general election. This is all very different from some other recent surveys that have had CON+LAB at below 60%.

    UKIP at bottom of party positive/negative ratings
    Cameron at top of leader ratings


Some relief for LAB/Ed with ComRes online as lead moves up to 4%

Saturday, November 15th, 2014

LAB 34%= CON 30%-1 UKIP 19%= LD 8%+1 GRN 3%-1

So only very slight movement well within the margin of error in voting intentions.

The findings from the poll that I’ve highlighted in the chart are on perceptions of Dave which I think get to the heart of the challenges for both leaders. People can’t imagine Ed at Number 10 while Dave is not seen as someone who stand up for working people. Also the perception is that a CON government will cut a lot more than a LAB one.

All of this is what is likely to play in the election campaign.

Mike Smithson

2004-2014: The view from OUTSIDE the Westminster bubble


The big and only real question is how the changes in the national mood are playing out in the marginals

Friday, November 14th, 2014

Have seats that were in LAB’s grasp now fallen away

There’s no doubt that this has been a dramatic polling week with apparently a move from LAB that is changing the long established view that the red team was heading for victory.

But these are national polls of 1,000 sample sizes for phone surveys and up to 2,000 for online ones.

    What we need to see before jumping to conclusions is whether the changes are also being seen in 75 or so seats that could change hand – the key battlegrounds.

General elections as I repeatedly observe are not decided by national party aggregate vote shares but by the outcomes in 650 separate constituency elections fought under first past the post.

Voters are not being asked to choose a Prime Minister or a party but an individual who will represent the area at Westminster. In some fights the personalities, popularity and overall voter appeal of the contenders will matter more than party branding.

All this is why in such a confusing national picture the single seat polling by Lord Ashcroft and others is becoming the best guide to GE15.

We see from the Ashcroft two stage voting intention questioning how things can shift sharply when those sampled are asked to think specifically about their own constituency. That fact alone should cause us to be more sceptical about the national surveys.

Lord Ashcroft is organising some Scottish seat polls as well as moving up the LAB target list to those that are less marginal. At some point the constituency polling will highlight a group of seats beyond LAB’s reach. We are not there yet.

To underline the scale of what Lord A is doing his last round of marginals involved telephone interviews with more people than Ipsos-MORI call in a year. And we are getting new batches every few weeks.

Mike Smithson

2004-2014: The view from OUTSIDE the Westminster bubble


Beware the over-prompting of Mark Reckless and UKIP in Rochester by-election polls

Thursday, November 13th, 2014

The intro to voting questions in the Ashcroft poll

As you may know, the Member of Parliament for Rochester & Strood, Mark Reckless, has announced that he is leaving the Conservative Party and joining the UK Independence Party (UKIP). He has therefore decided to resign as an MP and to fight the resulting by-election in Rochester & Strood as the UKIP candidate. Many local people are saying that they probably won’t vote at all in the by-election, while others say they definitely would vote. Please say how likely you are to vote in the by-election, when it happens?

if under-prompting depresses the shares then over-prompting could have the opposite effect

Much has been written in recent months about prompting in voting intention polls. UKIP has argued strongly that it should be treated on the same basis as the traditional three main parties and that those polls that don’t do this are understating its position.

    But could the opposite be happening in Rochester by-election polls? Could the form being used by pollsters actually be over-emphasising UKIP and its candidate?

Above I have reproduced the precise wording that Lord Ashcroft used in the latest Rochester poll. This is the very first question that participants are asked and it very much sets the scene for the questions that follow. Voters are reminded that Reckless is the incumbent, UKIP is name checked twice, and no other candidates are mentioned. The impression is this is all about Mark Reckless and his decision to switch parties.

My reading is that opening in this way could possibly could cause participants to overstate their voting certainty and then to respond to the voting question that follows on their view of the Reckless decision.

A more neutral, and possibly a better way to start, would be simply to say “As you may know there is a by-election in Rochester and Strood on November 20th. Please can you say how likely you are to vote”.

An interesting feature of the poll is that a second voting intention question, asking about the general election, was put and this produced very different numbers from the by-election one. The Tories were 1% ahead suggesting that if Mark Reckless won next Thursday he could lose next May. In that question there was no reference to Mr. Reckless/UKIP or any of the other parties.

It should be noted that similar question formats were used for Clacton and the polls were broadly in the right area. But that outcome was so much more clear cut right from the start and Carswell had a very different relationship with his constituents.

  • Just to add. Lord Ashcroft invited me to his office yesterday where we met for the first time and had great discussion on polling. I mentioned the above thoughts on Rochester. He’s a really nice guy and it was real pleasure seeing him
  • Mike Smithson

    2004-2014: The view from OUTSIDE the Westminster bubble