After an unusually long delay, the datasets from this weekend's Panelbase poll have finally emerged. Of course the first thing I always look for with independence polls is whether the unrounded numbers are slightly better or worse for Yes than the headline numbers, and in this case the answer is 'better'. In fact, the headline numbers effectively exaggerated the No lead by a full 1%.
Should Scotland be an independent country?
That suggests there has been a decent 2.8% swing to Yes since referendum day last September. The breakdown by recalled referendum vote reveals the broad nature of that swing, and it's very much in line with the pattern shown by other polling firms in recent months - slightly more No voters (6%) than Yes voters (4%) have made a direct switch to the other side, and more No voters (6%) than Yes voters (3%) have drifted off to the Don't Knows.
As I suspected, weighting by recalled referendum vote has boosted No quite significantly in this poll. The 480 respondents who recalled voting No have been upweighted to count as 506, and the 451 who recalled voting Yes have been downweighted to count as only 416. It does look likely, therefore, that if Panelbase were still using their pre-referendum weighting scheme, Yes would have a very slight lead. The superficial appearance from the headline numbers that nothing much has changed since the pre-referendum Panelbase polling (which often had Yes at around 47%) is highly misleading.
It's interesting to note that Panelbase are still suffering from the slightly mysterious problem that has plagued most online pollsters - they've got far too many English-born people in their unweighted sample, and are having to make a hefty adjustment to correct for that. Among Scottish-born respondents, Yes have a narrow 51% to 49% lead, which heightens the sensitive nature of the finding that, by a 52% to 48% margin, Scottish-born respondents think the franchise for any future independence referendum should be restricted to Scottish-born people - in line with the "Cameron Doctrine" of excluding EU citizens who are resident in this country from the EU referendum.