Exciting news, blog readers – results of the next Welsh opinion poll will be published on the morning of Tuesday 22nd March.
There’s a couple of things that will be different about this set of results from those of previous Welsh Political Barometer polls. First, this won’t be one of the regular Barometer polls. Instead what will be published are findings from the first wave of the 2016 Welsh Election Study (WES2016) – an academic study funded by the UK’s Economic and Social Research Council. As with the Barometer polls, the survey fieldwork is still being conducted by YouGov, via the internet. However, we will have a much larger sample than normal (roughly 3,000 rather than the normal 1,000), and the fieldwork will have been done over a slightly longer period than usual. I’ll give you the precise details on all that when I release the information next week.
Second, this will be the first Welsh poll to incorporate the fully revised YouGov methodology. These revisions reflecting the problems experienced by the polls at last year’s general election, and the findings of the various enquiries conducted after the election. In the words of Anthony Wells, Research Director at YouGov (as well as the person behind the excellent UK Polling Report website):
“Changes to Welsh methodology largely mirror the changes we’ve made to the GB methods.
1) Newspaper readership is dropped as a sampling and weighting variable (its original purpose was largely a proxy for political sophistication, but these days not reading a newspaper is no longer a suitable proxy for not being interested);
2) Instead we add sampling and weighting by education (interlocked with age), and sampling and weighting by the self-reported level of attention respondents pay to politics (targets for this are based upon the average across the last few waves of the face-to-face element of the British Election Study).
3) The targets for people who did not vote at the last general election are also upped slightly to more realistic levels (it is higher than it was, but still deliberately lower than the actual level of non-voting, as we know from voter-validation that some people claim to have voted when they did not, and because there are presumably some people on the electoral register who are unable to vote OR take part in polls because they are incapacitated through ill-health, etc)
The overall aim is to increase the representation within samples of those with low political interest, by upping the proportion of people with low education, who did not vote, and who paid little attention to politics.”
In practice, I doubt that these changes will make an enormous difference to the findings we report – and, after all, our Welsh poll last year actually performed fairly well, so you probably wouldn’t want to change things a great deal. In general I suspect the revised methodology would tend to push reported Labour support down a little bit, and raise the Conservatives slightly, on vote intention measures compared with the previous YouGov methodology; that’s very much in line with what the polling errors last year indicated was necessary. But the precise impact of the changes would vary from sample to sample – it is difficult to be definitive about precisely how much difference the changes will tend to make.
I’ll be presenting the WES2016 findings at a public seminar here in Cardiff Bay on Tuesday morning. I’ll then be releasing those same findings at approximately 9.30am on the blog. I hope that they’re worth the wait.