Regular users of WhatScotlandThinks website during the referendum will remember that one of its more popular features was a ‘Poll of Polls’ of voting intentions in the referendum. It showed the average share of the vote for Yes and No recorded by the last six polls of referendum voting intentions.
Attention is now, of course, focused on the UK general election that is due to take place on May 7th. And thanks to a set of polls that have suggested the SNP could win a landslide victory (at least in terms of seats), the level of interest in the election in Scotland is high – and certainly much higher than at the last two or three general elections when there looked to be little prospect of anything more than the odd seat changing hands.
So with now just under 100 days to go to polling day, we thought we would revive the Poll of Polls, but now focusing on what the polls of voting intentions in the UK general election are saying. The first instalment is to be found here.
Polls of voting intentions for the general election in Scotland are being published rather less frequently than they were for the referendum. So to try and ensure that our figures are reasonably up to date, our Westminster poll of polls focuses on the just the last four polls rather than the last six. This does, however, mean that there is an increased likelihood that it will fluctuate simply as a result of random variation between one poll and the next. Small shifts (in either direction) should not be overinterpreted.
Still, in contrast to the position in much of the referendum campaign, systematic differences between polling companies have so far been less in evidence in their estimates of general election support. Thus our Westminster poll of polls may be less at risk of fluctuating simply because of a change in the set of companies on which it is based. The one exception to that rule is that so far Ipsos MORI’s estimates of SNP support have been noticeably higher than those of other companies.
It should also be noted that not all polls have been reporting a separate figure the Greens. Where that is the case we have assumed that the party’s support represents the same proportion of the combined vote for Greens and other parties as it did in other polls conducted at around the same time.
In any event, the polls of Westminster voting intentions have in fact been remarkably stable since the referendum. At the bottom of our graphic we show how our poll of polls has evolved since September 18th. The polls began to pick up the ‘surge’ in SNP support in October, and by the beginning of November that was fully reflected in our calculation. The picture has barely changed at all since then.
Note that our poll of polls is only based on polls that have been conducted entirely in Scotland. Britain-wide polls of voting intentions, of which there is now one almost every day, do of course include Scottish respondents in their samples. However, individually there are too few such respondents in these polls to provide a reliable measure of voting intentions. That problem can be overcome by combining several polls together, but it still has to be borne in mind that the samples have not necessarily been weighted to ensure they are representative of Scotland in particular. So while the Scotland only results of combined British polls are of interest (and we will keep our eye on them in the coming weeks), we are not including the results of such exercises in this poll of polls.
Much of the interest in polls of Westminster voting intentions lies of course in what they might imply for the outcome in seats. Such extrapolations are always a risky business under the single member plurality system, but to give an idea of the possible implications of what the polls are saying, we will show for the latest poll of polls the seat outcome to which it points on the assumption that the difference between each party’s poll rating and its share of the vote in 2010 is replicated in each and every constituency. The calculation assumes that errors will cancel out!
Blog originally published in The Conversation.