What’s wrong with election polling and how to make it more accurate

The 2015 UK general election results left pollsters with egg on their face as every poll they did prior to election day proved to be wrong, very wrong. And not for the first time either. So why do polling companies continue to use such ineffective methods?

The problem they face is that asking people to articulate future voting intentions is flawed as many people just do not know how they will behave in the future or, alternatively, won’t tell you. Either way, those involved in collecting such information are using lazy and outdated methods to capture nonsensical data. A waste of time and money.

The exit poll results, released just before 10pm on the night of the general election, were based  on actual behaviour in the voting booth. This poll was the first, accurate indicator of what the result would be because it did not force people to articulate the future but simply recall the very recent past.

Even then, this exit poll, turned out to be short of the mark, failing as it did to predict the conservative party’s majority. It turns out that even in such circumstances, some people simply didn’t want to share such information.

It takes hard work to understand others. The business model of polling organisations is fixated on cheap, quantitative methods to collect information quickly, from large, robust samples of the population. The problem however is that such methodologies fly in the face of modern psychology and defy logic and common sense.

Large scale qualitative  data collection takes longer, is more expensive and data is much harder to analyse. But if polling organisations had used such methods, they would have shown that whilst there are die-hard segments who would never vote for any other party there are a great many people who take the decision in the polling booth. Understanding the emotional state of these people would have provided more valuable insights into their voting intentions than asking them who they would vote for. For instance, undecided people’s feelings about the incumbent party would have revealed a “better the devil you know” attitude  and a sense of loss aversion when faced with alternate options.

Customers are no different. Asking them to talk about future decision making is just as much a lottery. But getting to know them, understanding their feelings and concerns is a far superior way than cheap and lazy questions. But alas, organisations like politicians seem committed to using tools that are far from accurate and often totally misleading.

If insanity is repeating the same behaviour yet hoping to achieve a different result, perhaps its time to change the methodology.

Follow