Polling Results

Polling an individual constituency has become incredibly hard. The traditional method is to use telephone polling, where you continually phone people up from across the constituency until you have enough people who have answered and completed the full survey.

However, telephone response rates have fallen off a cliff in recent years. The long-running decline in response rates was sped up by PPI and other nuisance calls, making people less willing to stay on the phone for a few minutes to answer questions from a stranger.

In some areas, telephone response rates are now under 1% – meaning you would need to make over 50,000 calls just to get 500 responses.

In addition, GDPR legislation has meant that telephone lists that are geographically targeted are a lot harder to come by.

One solution is to use MRP modeling, where you survey a lot of people nationally, and then look at the demographic trends to create estimates for each local area. However, that doesn’t eliminate the need for constituency polls entirely, as it doesn’t allow you to account for times when hyper-local factors may be influencing people’s views.

Our solution is to go mixed-method, running half the fieldwork through the traditional telephone methodology, and the other half using online panels. This is what we have done for our recent constituency polls run for Greenpeace (tables at the bottom). 

Firstly, this allows us to run a large enough (500) sample in a way that is quick and cost-effective enough to work for most projects.

However, it also goes some way to smoothing over the problems faced by both methodologies. Online can pick up some of the younger in-work respondents that are hard to reach via phone. The telephone can reach some of the less politically active respondents that are harder to reach via online panels.

The data is then weighted by age crossed with gender, 2019 vote, and EU Referendum vote.

No methodology for constituency polling is ever going to be perfect, and therefore results should always be analyzed with those difficulties in mind. The 500 sample size is still smaller than a traditional national opinion poll, and therefore the margin of error is likely higher.

But given how hard the task has become, we believe this is the best methodology for producing reliable results.

Opinium ran three constituency polls for Greenpeace in Wokingham, Chingford and Woodford Green, and Wycombe. We also ran a poll across 41 “Blue wall” constituencies which were conducted entirely online. The full list of those constituencies can be found here.

The full tables are available…

Here < Blue wall

Here < Chingford and Woodford Green

Here < Wycombe

And here < Wokingham