When online pollsters told anyone who would listen in the year 2000 that the future of the research industry was online, few would listen and, despite their successes, this kind of attitude was still heard well into this millennium. Back then, the internet looked very different to how it does today; broadband was still some way off, internet penetration stood at just 25% and the notion of smart phones was consigned to Sci-Fi shows. However, despite the structural challenges to getting representative samples, internet surveys quickly proved their worth.
Poli’s innovative approach, conducting surveys through messaging apps used by over a billion people globally each month, faces far fewer structural challenges - more than seven in ten Brits use Facebook, including over 3 million people aged over 65. The barriers to using messenger apps are so low, and the scale with which they’re used so high, that the potential for conducting truly representative surveys using these tools is huge.
All survey methodologies, whether conducted face to face, by phone or online, must ensure that they can extrapolate from those who respond to surveys to those who do not. The polling misses of the last several years have shown just how difficult this has become with surveys still showing turnout approaching levels not seen in this country for two generations. This is at root of all the polling errors we have seen in recent times.
In 2015, the main problem was that too many people engaged in politics from Labour voting demographics were surveyed, inflating the Labour vote share. This led most pollsters to pre-set someone’s likelihood of voting in 2017 based on their demographics, however, the electorate changed enough between 2015 and 2017 that the Labour vote was understated in polls in June. Pollsters had looked to fix the symptoms of their errors and not the cause; their unrepresentative samples.
Poli’s approach will be different. Rather than manipulating our data to fit what we think the electorate will look like from one election to another, we will be making sure that we recruit enough people who do not vote in elections onto our panel. They will be recruited based on an interest in a variety of non-political topics and they will primarily get surveys on these, only getting political surveys when we need a representative sample of voters.
The way we conduct surveys, through short, interesting and visually engaging questions will also help to avoid the problems traditional companies face. Gone are the days of 15-20-minute surveys that most normal people find boring. Respondents will be able to answer surveys when it is convenient for them, whether this is on their commute, in front of the TV or on the toilet. We will also tap into people’s latent curiosity, showing them where they fit into the wider public based on their views on topics they’re interested in.
Traditional agencies ask a lot of respondents; signing up to their website and responding to an email all to answer 15-20-minute surveys on topics they do not care about. They must incentivise their respondents to take part, if they didn’t then nobody would join. Poli instead relies on an engaging product to incentivise respondents and will recruit the largest active panel in the UK. This will allow us to conduct surveys with huge samples and build more complex models than any other organisation. In the US, Pew analysed the state of polling and concluded “In general, samples with more elaborate sampling and weighting procedures produced more accurate results”. Poli’s approach will have the most elaborate procedures around.
Whilst we know that scepticism will persist until we can prove the benefits of our approach, we are convinced that we represent the future of the industry and we look forward to proving this.