MORI’s June political monitor has been published and is online here. The topline voting intention figures are CON 45%, LAB 28%, LDEM 16%. It was carried out between the 13th and 15th June.
I haven’t put a change since the last MORI poll for two reasons – firstly MORI have released some hitherto unpublished data from during that gap, secondly there are two major changes in their methods since the last poll. It is, however, the largest Conservative lead MORI have reported for many years, even taking into account methodological changes which this month have increased the Tory lead by 2 points.
The main changes in methodology are, firstly, a switch from face-to-face polling to phone polling, and secondly, the introduction of weighting by whether people work in the public or private sector. The full details of the review of methodology since the London elections is summarised here. It looked at four areas – sample representativeness, predicting turnout, differences between face-to-face and phone polling and how the polls are presented.
1) Ipsos-MORI’s face-to-face and phone samples contain a higher proportion of public sector workers than in the population as a whole. In the future they are going to begin weighting by public or private sector employment, using figures drawn from the Economic & Labour Market Review. Since they haven’t always asked this in the past it’s impossible to say for sure what the party partisan effect will be, but it is likely to decrease the Labour share of the vote and, in this latest poll, its effect was to increase the Conservative share by 1 point, and decrease Labour by 1 point.
MORI rejected weighting by past vote for the same reasons as they have in the past – that in their opinion the level of false recall of voting intention is liable to change in line with their political opinions, and that target weights for past vote must be based to some degree on subjective judgement, not hard fact.
2) Turnout. Ipsos-MORI are going to continue with their present method of filtering by turnout, taking only those 10/10 certain to vote for their topline figures. They are going to be experimenting in the future though with the sort of attitundinal questions ICM have starting factoring in, and might add trends based on other filters in the future.
3) Telephone. Previously Ipsos-MORI got much the same figures from phone polling as face-to-face polling. However, more recently they have found their telephone data to be closely inline with other phone pollsters (as one would expect, they are collected in much the same way) but their face-to-face data has shown a higher Labour share and a lower share for the Lib Dems and “others”. In their analysis Ipsos-MORI say that when they look at the data and include all the previously unpublished voting intention data from their phone polls (which is now on their website here – I will update my tables over the weekend) this had become a small, but consistent difference. Because of this they have decided to switch all the voting intention polls over to telephone sampling. The reason for the difference is, incidentally, a mystery – MORI have checked it isn’t to do with homes without a landline, eliminated the chance of it being the showcard used in homes as compared to vocally prompted questions, or question order.
One regular question that will still be asked face-to-face is, thankfully (since regular readers will know I am a huge fan of it), their “most important issues facing the country” question, which unlike other questions of this time is entirely unprompted.
4) Presentation – Ipsos-MORI are releasing their data in a way that makes clear how figures should be interpreted. This month’s figures are in a comprehensive pdf, which includes things like warning people that the unfiltered voting intention tends to overestimate Labour support compared to election results in these days of lower turnouts.