We have seen many elections and referendums get it wrong in the last 18 months, whether it was the Brexit vote or this week’s U.S. Presidential election. We are all wondering why.
There is a good reason why these polls are inaccurate and an even better reason to be optimistic about the future of polling.
A poll today often asks us questions that we answer via our short-term memory. If we are lucky, we can remember an average of seven items per topic and that’s it. We’re tapped out. That’s not a problem when you are asked simple questions.
The bigger issue relates to how we effectively utilize big data to understand what is really happening inside our heads.
If a topic is safe and positive, we answer accurately. However, if we are worried about how we will be judged or if we are filled with negative emotions, our judgement changes and we start to act differently. Here are a few actions we take as a result.
We give the answer we think others want to hear, since we do not want to be judged negatively. And that leads us into a new set of metrics for polling that will revolutionize how we measure an election or public opinion of any type. In a highly emotional and partisan situation, we’ll consider the following:
We will evaluate subconscious behavior. For example, what are we searching for vs. what we are retweeting vs. which sites we visit. We will start to see that the same group of people who are saying they will vote for candidate X are actually increasing their searches for topics that are more positive for candidate Y or they say they like candidate X, but never show their support online. We will see the differences in what we say verbally vs. what we do when no one is watching.
We will measure passion. As a result, “events” will be redefined as “moments where the voter can express their support”. It could be a physical event like a rally or a photo you can like on Instagram or a video you can watch for five minutes or a speech that you reviewed and shared. By looking closely at dozens and ultimately hundreds of “events”, we can see if passion is building or waning and then zero in on who, why, where and when. We’ll be measuring every place an impact can be made or a sign of respect or disrespect can be shown.
We will measure silence. Imagine having a normative data set of all voters for your party and you start to see that 500,000 of these potential voters have stopped participating in any form of support, yet they have shown support in prior elections or for other relevant political causes. These may be silent voters who judge the price of being public as too high, but they will still vote. We could also start to see if we are losing voters to another candidate, as well.
We will measure apathy. If we look over the course of a campaign, where is interest non-existent or where has it decreased significantly? What is happening that is causing an entire area of people to shift their views? Why are less people in Detroit interested in candidate X over the last six months? We’ll be able to see what is happening at the neighborhood level, which will impact our choice of content and location so we can regain interest in the candidate.
We will measure the importance of a key topic by town by influencer. In today’s world, you can see which topic matters by town by candidate and which people have the most influence online. So, if you are visiting a key city in Ohio, you know you should talk about manufacturing first and health second, you know who to invite to your event and you know what other towns in Ohio care about this exact topic. You will know that 97 cities/towns matter for your candidate in Ohio and 130 in Texas and so on. You won’t guess at all. You will also know which towns and cities do not rate these topics as a priority, so you can streamline where you visit. We can see this, of course, by analyzing what content people consume at the local level.
The list goes on. The result will be a renaissance moment for polling. We will have completely new pools of people to choose from to define where an election is heading. We will understand better how to look at historical data and subconscious behavior and other data points to know what questions we need to ask, when to ask them and where to ask them.
The art of polling will increase in value as researchers, pollsters and those they serve learn how to look at all “data inflection points” and master the art of understanding the subconscious behavior of all of us, particularly in these highly partisan times. Those who figure out how to leverage this influx of new data points will continue to find that “edge” in each election until this becomes standard business.