Andy Sawford: Now the pollsters can’t win either way

Written by Andy Sawford on 31 May 2017 in Opinion

Many in the industry no longer believe that election polls will be accurate within the margin of error.

With a week to go until the general election, the stakes are high for everyone involved professionally in forecasting political outcomes, including media pundits, pollsters, bookmakers and public affairs experts.

Opinion pollsters reacted to criticism that they got it very wrong at the 2015 general election, with an industry wide inquiry to defend their reputation. They rightly remind us that polls are a snapshot of opinion at a point in time, rather than a prediction. Similarly, it is not their fault that the media treat the ‘margin of error’ as a minor detail, rather than a critical consideration.   But there are deeper questions for the polling companies about their methodologies, and it is clear that they are addressing this in different ways.

The most recent poll from ComRes, for example, points to a landslide Conservative majority, while YouGov’s latest figures suggest Tory losses and a hung Parliament. They cannot both be right.

The PR industry has also been involved in some introspection itself, through a review by the PRCA.  As part this, we are now publishing the results of a panel of over 100 PR professionals, mainly working in public affairs, about the general election.  

Our panel were asked to both make their predictions and tell us the basis for them. While many told us that they do consider opinion polls to be important, only half believe that the polls will be accurate within the margin of error.  This is no surprise given that there is near unanimity in predicting a large Conservative majority.

In our work over recent months reviewing how predictions are made, there has been much soul searching about two issues: firstly, whether people’s predictions are bias towards the outcome they personally want; and secondly whether there is a ‘Westminster bubble’ bias in who people listen to and how predictions are influenced.  Our survey shows that the industry is addressing this in two ways.

First, people are now more likely to take account of personal insights, such as doorstep conversations, than the views of their colleagues or media commentary. Listening to the man or woman on the Clapham, Cleethorpes, Clitheroe or Clyde omnibus is seen as important by over 80% of our panel, with 40% saying it is very important. Our panel rated these insights as of almost equal importance to opinion polls when making predictions.

Second, while our respondents are almost equally divided about the outcome they personally want, in terms of either a Conservative or Labour government, they are clearly able to put this aside when making their predictions about the likely outcome of the election. Many will have learnt from Brexit and Donald Trump’s win to see beyond their own preferences and prejudices, when forecasting political events.

We will be looking back on the survey after the election to see how the industry predictions faired, and using this as a benchmark for the future.  In the meantime, spare a thought for the pollsters, who now can’t win either way. Some of them are going to get egg on their face next week, the question is which ones? 


Andy Sawford is managing partner of Connect.  He is currently leading the PRCA Review of Political Predictions. 

Share this page