YouGov admits it was reading Scotland wrong


by Dave Taylor

Almost everyone knows by now that the wee Scottish samples in GB polls have such a large margin of error that they are pretty useless in estimating voting intention.  However, if they are all pointing in the same direction, taken together, they are a reasonable indicator that that direction is right.

Since May this year all the Scottish samples in GB polls have shown the same phenomenon – instead of Westminster voting intention showing a big Labour lead over the SNP, the SNP either leads, or is level with Labour.  All, that is, except YouGov.

Today, YouGov admitted that their methodology for the SNP and Plaid support has been wrong.

Polling guru Anthony Wells of YouGov said today:  “YouGov are now weighting SNP and Plaid separately from the other ‘others’.  The actual targets that party ID is weighted to are all unchanged, but weighting the SNP in with the others was producing too few SNP identifiers in the Scottish sub-sample.”

In GB terms, that means little change.  Labour support in GB drops by around 1%.  However, in Scotland it means that YouGov are now showing the same pattern as the other pollsters.

Tuesday’s poll for the Sun gives the SNP a lead of seven points over Labour, in their wee sample of 142.

SNP 39% : Lab 32% : Con 17% : LD 5%

On its own, that Scottish sample means nothing.  It’s too small, and isn’t weighted to the Scottish demographic.  But it does mean that all the pollsters are showing the same pattern.  It is inconceivable that all of them, using different methodologies and samples, are all producing a bias in the same direction of increased SNP support.

We haven’t had any Scottish polls since the end of August, but the indications are that we can look forward to the next ones with a degree of expectation, and not just hope.

The YouGov poll also brings another spot of good news for the SNP.  YouGov occasionally ask attitudinal questions and, as Anthony Wells has noted, sub-samples on these do tell you that something is different.

In today’s poll, this question was asked: Thinking about the next two or three years, how worried are you that people like you will be victims of burglary, robbery or mugging?

People in England and Wales are evenly divided on this.  47% of those in England and Wales responded that they were worried, with 48% saying they were not worried.  However despite a seemingly constant barrage on the media about violent knife crime, Scots reported that they were markedly less worried by violent crime, with only 28% answering that they were worried compared to 71% who said they were not worried.  

Sounds like Kenny MacAskill seems to have got things right.  I’ll be keeping a close eye on YouGov’s polling results in future to see if they’re reading Scotland right.