OPINION7 May 2010

A Case of Social Desirability Bias in Polling

Cleggmania and the election result perhaps reveals some of the pitfalls in polling

So, like 1992, the opinion polls proved wide of the mark. But I wouldn’t throw the baby out with the bathwater. It just goes to show that what really counts isn’t the survey result but actual behaviour, and that polls seem only to really hit the mark when there is a degree of electoral stability.


See: http://news.bbc.co.uk/1/hi/uk_politics/election_2010/8667801.stm


Perhaps what has been most confusing, though, is the degree to which Cleggmania (at least that’s the media headline) fell apart.  My own view is that there is a certain implicit social bias in response to researchers. 


When the whole media space builds these things up, people feel they have to express agreement with the social buzz and say what feels like the socially correct thing to say – to follow the ‘cultural leaders’ who try to define the moral high-ground. 


But this is a very fickle arrangement.  What happens ‘in the moment’, in that personal time when an individual decision is made, reflects true desires not those bound by artificial convention or the media definition of what you should do and say.   


The question is, how to take account of this type of error in the future.


1 Comment

11 years ago

Not sure the polling errors are quite at 1992 levels, Steve. Predictions of Tory and Labour vote shares were largely on the money – save for a few, like Angus Reid, which had Labour in third place. The real issue, as you point out, is the overstatement of Lib Dem support. Its something our features editor, Robert Bain, is investigating at the moment. Post to follow on the Electionwatch blog very soon...

Like Report