The next Welsh Political Barometer poll should be published early next week. It should have all our usual goodies – vote intention figures for Westminster, the National Assembly and the EU referendum – plus a few other things as well. Christmas is going to seem quite the anti-climax after that, isn’t it?
Just a word on methodology. After the much-publicised difficulties encountered by the various polling companies at the general election various enquiries were launched. Some polling agencies, such as ComRes, have already made some substantial revisions to their methods of weighting samples. YouGov, who do the fieldwork for our Barometer polls, have not yet completed their post-mortems (although I understand they will be doing so before very long). Our December Barometer poll will therefore follow the same sampling and weighting approach as did those in June and September. To remind you (and in the words of Adam McDonnell, a Research Executive at YouGov who has worked with us on all our recent polls):
“Currently, while we work out our new sampling frames and weighting, we are using the same sampling methods as pre-election and the same weight variables with the exception of Party Identification. Instead of Party Identification we are weighting by 2015 general election result.”
As I mentioned in a blog post prior to the June poll, given that the final, pre-election YouGov Welsh poll was actually very accurate – only being about one percentage point too low for the Conservatives and one point too high for Labour, with the other parties being estimated very accurately indeed – these changes should not make much difference to the results. They should probably tend to reduce Labour’s reported support very slightly compared with the methodology YouGov was using before the general election, and increase that of the Conservatives by a tiny amount, while leaving that of the other parties more or less unchanged.
Moreover, given that there have been no methods changes at all since June, if we see in our new poll any substantial changes in reported support levels for the parties then it will not be methodology that accounts for the shifts. Changes in party support levels might reflect normal sampling variation between individual polls, or they might indicate genuine changes in the public mood. I hope this clarifies how we should respond to next week’s poll.
(By the way, in case you are wondering – at time of writing this I have not seen any results from the poll. So I am not trying to offer tantalising hints of what the findings are. I can’t do that because I don’t know – and nor does anyone else yet either!)