Fixing the problem with polls?

Sarah M. Inoue
2 min readMay 7, 2021
Photo by Arnaud Jaegars from

In 2016 and again, to some extent, in 2020, the polls before the presidential election were not as accurate as pollsters would like. Like most people, I was more interested in the election results than whether the pollsters got it right, but now that we are six months away from the election, I’ve started thinking about this problem.

Polls are done for two reasons: to predict the outcome of an election and to understand how people feel about an issue. I’m not sure I can fix predictions, but my work at TREND Community has been about creating better tools to understand how people feel and deal with an adverse situation by using social media listening. By listening to people who are talking amongst themselves about a problem, we are able to hear more about how they feel than a survey might tell us. Surveys can only find answers to the questions asked, whereas when we listen to people, we can hear what is important to them. For example, on my first assignment with TREND, we listened to a community with a rare disease. A survey had just been given to the community by a foundation set up to help find a cure. Our main finding was that people in the community were talking a lot about pain, but the survey didn’t ask a single question on pain.

I’m curious if the social media listening will help with polling data. Social media listening probably cannot replace polling and surveys, but it may be able to enhance their efforts. My plan is to attempt a few small projects on issues that are in the news to see what I find. If you have issues you would like me to work on, let me know. If you have ideas for how to use social media listening to improve the predictive power of polls, let’s discuss. Thanks for reading!