h1

It is a mistake to assume that all polling bias is against the Tories

January 18th, 2016

Remember what happened at GE2010 and the last London Mayoral election

This week’s polling news is going to be dominated by the publication tomorrow of the inquiry into what went wrong the GE2015 polls when all the firms undershot the Tory share by big margins. Unfortunately I’m at a memorial service tomorrow and won’t be able to attend the big event.

In the build up we’ve started seeing some interesting explanations including one which suggests that CON voters are much less likely to take part in polling surveys than non-Tories.

There’s a tendency to think that all polling bias is in one direction against the Tories but this is simply not the case.

The question will be post Tuesday will be what should pollsters do about it in order to ensure more accurate polling at the next election. My concern is that there could be an over-reaction.

    Let’s not forget that at GE2010 all the final polls understated Labour’s eventual share and we had a similar experience at the last London mayoral election as seen in the chart above.

In May 2012 all pollsters got the CON/Boris lead over LAB wrong some by quite huge margins. If there is a systemic pro-LAB effect then why didn’t it show up 4 years ago in the capital? Maybe there’s something specifically about Boris or London that impacts on the general theory?

One factor is that turnout in the mayoralty election was considerably lower than in the general election. Maybe the former attracts just the engaged while with general elections embrace many who are less engaged in the process?

It might be that there is no “one size fits all” solution to UK political polling.

Mike Smithson