Can We Still Trust Opinion Polls? A Young Researcher’s Response

Can We Still Trust Opinion Polls? A Young Researcher’s Response

4 August 2017

As we all know, the reliability and reputation of pollsters is at a low ebb. Last week, Simon Atkinson of IPSOS delivered a reasoned defence of pollsters at the MRS’s Guest Speaker Evening.

The context for polls is important here. Unlike most market researchers, pollsters find their results shared with national and global audiences. With this comes added public accountability and increased pressure on the validity of their results – pollsters are effectively accountable to, and judged by, the public.

A Challenging Climate

Whilst pollsters now find their work increasingly disseminated amongst the public, they are also discovering the market they are operating within offers far greater challenges. In today’s environment pollsters are finding an increasingly agitated populous, leading to more disruptive and un-predictable elections.

The Pressing Challenges

As the market has become increasingly agitated, sampling has become more fluid and less predictable. In the previous era of “party identification”, where belonging to a party was part of the social fabric it was easier and more reliable to sample effectively. Today following the rise of “partisan dealignment”, party identification has reduced in salience. With this, pressure has built on pollsters to try and accurately sample in an era where partisan roots are more saturated.

Whilst sampling is troubling for pollsters, varying turnout is their achilles heel. Turnout has varied significantly in recent elections globally, both in general and within different segments, age being the most important. For pollsters, this adds to the climate of unreliability as when turnout varies so drastically from one election to the next, it makes it hard to set an accurate turnout filter. Where this filter is not accurate, discrepancies occur, increasing the risk errors emerging.

What can pollsters do?

Whilst these challenges are hard for pollsters to remedy, it is important to recognise that they are not completely innocent. As Simon himself accepted, there are areas in which pollsters and polling companies need to improve:

  • Pollsters need to, and should act, with increased bravery. Fear of publishing mass data sets into the public sphere instead of just a single table should be avoided. There is also an argument to be had about breaking from convention in the sense of visualising polls in new formats to account for the volume of data now at our disposal
  • Pollsters need to recognise that when their results are disseminated there needs to be greater importance placed on educating the public who consume them. The public at large need greater explanation of the data on offer and need to understand that an opinion poll should not be taken as fact alone
  • As Simon highlighted, pollsters could also benefit from integrating new models into their methodology, while also taking account of new variables when sampling, education being significant

A Golden Opportunity?

Ultimately, for all the criticism and negative press, opinion polls are here to stay. Opinion polling has become an important characteristic of the electoral process with each poll becoming eagerly anticipated by media and politician alike.

As an industry, opinion polls make up a significant part of our identity to the wider public, and while pollsters do make mistakes there is no doubt that we should continue to support this method. Resultantly, if as an industry we can work towards a point whereby our polls are highly accurate, it would be a huge benefit for us all and potentially lead to wider benefits for data collection throughout market research and the industry’s identity as a whole.


Please contact Charlie Rollason on 0207 259 1754 or for further information.

Comments are closed.