Conservative Home

« David Gauke MP: The men who came to dither | Main | Voice from the City: Peston's Darling get's it wrong »

Stephan Shakespeare: Frothy opinion polls

Stephen Shakespeare co-founded YouGov and is the owner of

I would like to be able to begin:

“As I predicted, Hillary wins in New Hampshire...”

Unfortunately I only said that Obama’s lead would be much, much smaller than the polls were indicating, and I made that prediction to friends (such as the editor of this site), and not in print. I enquired about the more attractive odds on Hillary to win the nomination, but I didn’t actually place the bet. So my claim is empty of glory.

However, I did on ConservativeHome advance my theory of frothy poll leads, with regard to the Brown/Cameron volatility, and that was substantiated in the New Hampshire primary.

My proposition is as follows:

A significant proportion of people tell pollsters that they will vote in an election then do not, in fact, end up voting. It’s not that they are deliberately lying about their voting intention, it’s just that they want to think of themselves as engaged in what they recognise as important matters, even though they’re not. It is very hard for pollsters to be confident about who is, and who is not, actually going to vote.

Those who claim to be voters but subsequently don’t vote are likely to know less and think less about political issues. When asked questions, their answers are more likely to reflect media buzz. If they don’t really have their own thoughts about an issue, but want to avoid saying ‘don’t know’ because they consider that to reflect badly on them, then they are likely to repeat whatever they happened to pick up around them.

My view therefore is that the media – indeed, the wider public buzz - has a stronger influence on polls than it has on actual behaviour. You will tend to see a ‘mean reversion’ (that is, a return to norms) as the froth settles down, and potentially a strong anti-reaction when reality hits, itself creating new froth.

Predictions are not easy, because as well as froth there is also genuine momentum: sometimes the buzz will convert into reality, norms do move. Increased volatility I take to be a sign that genuine engagement is declining; while sustained trends mean that engagement is increasing. By ‘engagement’, I don’t necessarily mean taking part in the political process, but connection (consciously or unconsciously) with the issues.

Related link: Stephan Shakespeare's archived ConservativeHome columns.


You must be logged in using Intense Debate, Wordpress, Twitter or Facebook to comment.