Cafe Scientifique

Cafe Scientifique

From the archive

This event last happened on 5 February 2018

Next event:

Opinion Polls-Why do they get it wrong?

Monday 5 February

Details coming soon...

Speaker info

Charles Pattie is Professor of Politics at the University of Sheffield. He has spent much of the last 30 years researching and writing about elections, voting behaviour and party campaigning in the UK and elsewhere. As a long-term user of, and commentator on, survey data and opinion polls, he both relies on the polls as sources of data and frets about their quality.

Citation if available:  https://www.sheffield.ac.uk/politics/people/academic/charles-pattie/profile .

Abstract

We live in an increasingly data-rich world, and are we bombarded, on an almost daily basis, with results from an ever-increasing number of opinion polls and social surveys. At their best, these offer valuable insights into trends in public opinion, into support for our government and opposition parties and for various public policies, into our views of the world around us, and how we behave and think. But how reliable are they? In recent years, opinion polls have come unstuck in very public and embarrassing ways. Election polling, for instance, has not had a good track record in recent contests, predicting the wrong outcome on several occasions – and most embarrassingly in 2015, when almost all major polling houses failed to spot the final result and, worse, mostly ending up being wrong in the same ways. This has lead to a number of inquiries into what has gone wrong. And it has also lead to suggestions that polling should be regulated more tightly (and to some suggestions that polls should be banned during lection and referendum periods, for fear that they might distort the real vote).

So what has gone wrong? Should we be worried? Are polls little better than expensive guesswork? Or is three still something to be gained from polling? If so, what – if anything – can we ‘fix’ to make polls more reliable? Do we focus too much on ‘the polls’? Or should we, if anything, focus on them rather more than we do?

Explore

We use cookies to help us provide you with a better service, but do not track anything that can be used to personally identify you.

If you prefer us not to set these cookies, please visit our Cookie Settings page or continue browsing our site to accept them.