By Dave Taylor
The Panelbase poll for Wings Over Scotland confirms the reduction in the No lead that our poll showed 3 weeks ago. Indeed, it has the Yes vote a shade higher than our poll found.
Whilst it’s always nice to have your own poll results confirmed, the actual level of support for Yes and No isn’t guaranteed to reflect what these two Panelbase surveys show.
Those who follow such polls will know it’s the trend within polls by the same pollster that allows us to see how opinion is changing, and it’s this that both campaigns will be monitoring.
We’ve had three Panelbase polls over the last 6 weeks – conveniently spaced 3 weeks apart.
The respective trends are:
Yes : 37.0% -> 39.9% -> 40.8%
No : 46.7% -> 45.2% -> 45.6%
DK : 16.2% -> 14.9% -> 13.6%
Or, excluding undecideds
Yes : 44.2% -> 46.9% -> 47.2%
No : 55.7% -> 53.1% -> 52.8%
While we don’t normally report percentages to a decimal point, rounding of numbers often conceals or exaggerates changes. Interestingly, prior to the rounding up of the last two polls, the Yes vote is marginally higher than 3 weeks back, while the No vote is smaller.
As always two caveats –
1. There are wide disparities between the voting intention figures produced by different pollsters, because they all use different methods in their genuine attempts to honestly measure public opinion in a situation that no one has previous experience in. On 19th September, we’ll find out which of them (if any) had the correct methodology.
2. Pollsters balance their entire sample to try to represent the population, not the sub groups. Not only is the margin of error much higher for the smaller groups, but if a particular poll has a greater number of Yes people in one group that doesn’t affect the overall poll figures if it’s balanced by a greater number of No people in another group.
Despite that, looking at the sub groups is always interesting. If a particular pattern is found in particular groups in a number of different polls, then it’s safe to say that pattern is solidly based. Hence, we know that men are generally more pro-independence than women, and over 55s are generally more pro-Union.
After our own poll, we reported that the Yes vote had risen in every demographic group since the February poll, except over 55s and 2011 non-voters. The Wings poll confirms that. It finds the same pattern except among C2Des where we found a tiny rise in the Yes vote, while they found a tiny rise among Nos. No change seems the most likely answer.
Other changes in demographic groups also largely cancel each other out in a comparison with our poll. For example, our poll showed an unusually high shift among women from No to Yes, and among over 55s from Yes to No. The Wings poll does show these shifts, but to a lesser extent. So, it’s safe to say that Yes is making some progress in an area they have struggled – attracting the female vote.
We should expect such churn among sub groups. It can be a genuine effect as people move in and out of the undecided group or, more probably, it’s random sampling error. The trend, as always, is what matters.
Naturally, some Yes voters will exaggerate the importance of the poll, while the No campaign will try to discredit it.
But we, all of us, have to be careful in how we interpret reactions from our opponents in this increasingly difficult to call referendum campaign. We don’t have to follow the Wings Over Scotland hypothesis that exceptionally rapid blinking by the head of Better Together is an indication of lying.
Blinking can also be an indicator of panic from someone who is being asked about something they wholly fail to understand.
Having said that, maybe the former Chancellor would prefer to be thought of as a liar rather than someone who doesn’t comprehend basic numbers and statistics.