Why Better Together should be worried by recent polls


By Dave Taylor

Last Sunday the first poll since the launch of the Scottish Government’s Independence White Paper was published.

Commissioned by the Mail on Sunday, the poll gave the No campaign a commanding lead – 56% to 27% with 17% undecided.

By Dave Taylor

Last Sunday the first poll since the launch of the Scottish Government’s Independence White Paper was published.

Commissioned by the Mail on Sunday, the poll gave the No campaign a commanding lead – 56% to 27% with 17% undecided.

The MoS announced the result in typical understated fashion: “The campaign to tear Scotland out of the United Kingdom has failed to gain any ground despite the launch of the Scottish National Party’s long-awaited prospectus for independence.”

According to the newspaper, polling expert John Curtice said: “There can be little doubt that the Yes side will be disappointed with this result because it does not show any evidence of any measurable swing of opinion.”

The poll was carried out by Progressive Scottish Opinion.  The same firm had carried out an earlier survey for the MoS in September that found 27% of Scots intended to vote Yes while 59% intended to vote No with 14% undecided.

Days after the Mail on Sunday ‘White Paper’ poll, we had another from TNS BNRB which found No on 45%, Yes on 25% and undecided on 32%.

The polls generated the predictable headlines and statements from the Better Together campaign hailing the results as proof that the Yes campaign was in the doldrums.

But no sooner had the noise subsided, than criticisms of Alistair Darling’s handling of the anti-independence campaign began to emerge.  They jarred with the supposed commanding lead enjoyed by Better Together.

However, closer inspection of these polls reveals all is not as is being portrayed – and the No campaign almost certainly know it.

Let’s take the TNS BNRB poll, which gives the No campaign a 20% lead.

In the TNS series of polls from August to the White Paper launch, the Yes vote has been virtually constant at 25/26%, while the No vote has lost 5% to Undecided.  Over the last month, the reports that Yes gained 1% from No are wrong.  In reality, Yes gained 0.4% each from No and Undecided, small movements but important.

Looking at likely voters only, since August Yes have lost 4%, and No 7%, to undecided voters.  Over the last month, the Yes vote was stable, while No lost 2% to Undecided.

Monthly changes can be within the margin of error, but if they are all in the same direction then the likelihood is that they are measuring an actual shift in opinion, as they result in a significant difference over time.  That is what we are seeing and appears to support what the Yes campaign are saying.

The Mail on Sunday poll is even more interesting, but offers no comfort to the No campaign.

Looking again at both MoS polls and we see the No camp losing support to the undecided.  Since the September poll, the No has dropped from 57% to 54% whilst the undecided has gone up by the same amount.

But what of the lead – even notwithstanding the slight erosion of the No vote, the MoS poll gives the No campaign a massive 29% lead.

Newsnet Scotland has acquired the details of the September poll which showed a remarkable deviation from other polls in party political support for parties at Holyrood.  If that poll was correct in measuring public opinion, then Johann Lamont is to be lauded for turning around Labour’s prospects.

While other pollsters found that more Scots would choose to vote SNP than Labour in recent polls, Scottish Progressive Opinion gave Labour not only an 8% lead over the SNP, but also a 6% lead in the North East, where the SNP currently hold every constituency seat.

That the MoS chose not to publish those findings is sufficient evidence that they realised how unrepresentative their sample was, and why they decided not to ask that question again.  However, if we accept that the PSO overweights Unionist voters, then the three percent swing from No to undecided is all the more striking.

Methodology is critical in understanding what polls are telling us.

PSO is not primarily a political pollster.  That they have no way of knowing the political ideas of their sample does not reflect on their ability to tell commercial companies if this or that advert is better.  It does mean though that their political polls may well be unrepresentative.

By contrast, TNS has a long experience of political polling in Scotland, but has a unique process of face-to-face interviews.  Respondents are quite happy to say they will vote SNP, Labour etc, but seem much less likely to commit to a view on independence in such a situation.  Hence their much larger proportion of undecideds.

Only the trends within each pollster’s series of polls is of importance. The PSO and TNS polls support the Yes campaign’s claim that they see a shift from No to undecided.  Whether that shift occurs at the necessary rate, and whether the undecideds then shift to Yes will determine the referendum outcome.

Finally it’s worth noting what the Scottish Mail on Sunday editor, who very kindly supplied Newsnet Scotland with the details of both polls – greatly appreciated – said when Newsnet Scotland pointed out that there appeared no account taken of party political support now, nor of how respondents voted in the 2011 Scottish election, in their post White Paper poll.

His reply revealed perhaps a naive understanding of political polling and an explanation of why some newspapers tend to exaggerate their headlines or strap lines when revealing their poll results.

“We are comfortable with the methodology used by Progressive as the sample is representative of the population.  Given that this is a referendum, a representative sample like this is the fairest way to carry out a poll like this.

“We chose not to ask about Scottish Parliament voting intentions on this occasion, although we did ask this in our September Progressive poll.”