Insight & Research

November 7, 2020

Polling opinion: what can the US Election teach us?


The US voting system can seem convoluted, complex and strangely reductive to European eyes, with its focus on state by state response.  The long build-up of campaigning and slow-rolling release of results created a nail-biting story, with twists and turns and an unsure outcome days after polls closed.  In the four days between Election Day and Joe Biden being declared the winner, we turned our minds to what we can learn about consumer viewpoints and opinion measurement based on events across the pond over recent weeks. As marketeers and communicators, there is much to be gleaned about how to best harness and measure opinion.

Opinion polling has long been a central tenet of the story, and yet so often seems to get it wrong.  The ‘shock result’ of the last US election, where outcome differed from expectations based on polling, created shockwaves and yet the same outcome is a possibility again this time. Commentators have suggested various reasons for this: inconsistencies in methodology between polls (CNN’s telephone approach versus CBS’s online method), a proliferation of new poll sources with the advent of technology (almost anyone can set up a polling company with a few thousand dollars, it would seem), as well as sampling errors (Pew Research have put the figure at around 6% due to mismeasurement, non-coverage of certain cohorts and other sampling errors).

This is a reminder that robust sampling and best-practise quantitative rigor are key when seeking to predict future consumer behaviour, but in addition, it highlights the mismatch between intentionality and actual behaviour.  As marketeers we know that claimed future behaviour, i.e. asking people what they think they will do in the future, can often lead to incorrect steers.  In short, people often don’t do what they say they will do.

To avoid the same shocks from consumers that we may see from US voters, it is essential that we also ask the right questions when we seek to measure future behaviour relating to brands and services.  When gauging opinion via quantitative surveys, implicit testing (for example timed responses in a survey) helps us avoid overly rationalised responses (what people think we want to hear).  And integrated methodologies, combining both qualitative and quantitative approaches can play a vital role in accessing a better picture of what people may go on to do.  Indeed, using qualitative methodologies in support of surveys, enhances, illuminates and sometimes challenges the numbers, giving a better read on eventual outcome.  Best practise qualitative methodologies utilise projective and enabling techniques to get beyond surface level responses too, beyond the claim and closer to the truth of eventual behaviour.

We also learn from the US election process that the best insight sometimes comes from unusual sources.  In the run up to this election, campaigners used text messages extensively to reach out to voters, and in doing so, uncovered an informal yet illuminating data source. In the 2018 cycle, Democratic lobbyists sent some 350 million texts, capitalising on the power of peer to peer texting (and to circumvent spam folders). Whilst initially intended as a one-way communication, many recipients replied back (often negatively) and this became a rich data source, uncovering nuanced concerns amongst specific cohorts. A system of coding responses was quickly developed, and key findings used to nuance campaign messages in a way which conventional polling cannot.

In the US election, release of polling data as the story unfolds has been shown to affect campaign behaviour and ultimately the result. Whilst in the context of the voting process this can seem ‘unfair’, as it can alter people’s original voting intention, it does enable more effective and targeted campaign messaging.  In the same way, there is potential to engage our consumers in the findings throughout the insight journey in order to develop better communication strategies.  There can be a benefit to engaging the consumer in the inner workings of our strategy – we increasingly advise consumer-led methodologies, where the target is transparently engaged in the brief. Part of this can include letting consumers in on the evolving insight story and tasking them with finding solutions via co-creation, for instance as part of a Sprint process (where a task must be completed in a very short, fixed amount of time). This can nuance and refine potential communications campaigns – but the key is to know which insight to release to participants, and the need for intuitive analysis means these methodologies are invariably qualitative.

Finally, a simple but key learning is that we must always place data on intended behaviour within a broader context. We need to ask ourselves what other factors might be at play that need to be factored in when projecting future behaviours from stated intent.  Behavioural scientists have long argued that anxiety is the enemy of effective, rational decision-making. Due to the current pandemic, just as American voters exercised their democratic choices at a time of unique stress and uncertainty about the future, the same applies to consumers in their day to day choices of brands, products and services.  The ‘Covid effect’ means that incumbent and familiar choices may hold inherent appeal, despite what people say they may want or say they will buy into theoretically. This presented unique challenges to the Biden camp, and to us as marketeers.  As such, it is essential that change-aversion is treated as an additional dynamic at play, and to ensure our communications and messaging work even harder to minimise the distance between intent and action.


Written by Sarah Christie, Qualitative Research Director, Truth Consulting