As multiple recent elections on both sides of the Atlantic have proven, the pollsters can quite easily be wrong. 

To find out more about the crisis of confidence in polling, we caught up with Don Vaughn, Ph.D., Head of Product at data analytics firm Invisibly, which claims to have predicted the outcome of the US presidential election to the greatest degree of accuracy.

According to Vaughn, a fundamental overhaul to traditional polling methods is needed, with an emphasis on real-time research.

What is the difference between a poll and a survey?

Polls traditionally focus on the measurement of one thing, for example an election race. Surveys typically involve a broader series of questions on a topic and attempt to collect more data.

Why do you say that traditional polling has become obsolete?

To be clear, it isn’t that traditional polling is obsolete in the sense that it isn’t practiced anymore – it is obsolete in that it is outdated.

The problem is that these outdated data collection practices in polling are still the most popular and get the most attention. Many polls are done person-to-person via telephone which in itself cause biases in the data collected. We need to move to more anonymous polling that is suitable for the digital age which fits within our day to day life and is not disruptive. Invisibly Realtime Research solves this issue.

What prevents Facebook or Google from launching a similar project?

Google already does have a survey tool, but it’s built in a way that is meant to make money or drive views to content, not to collect completely unbiased data. Invisibly is different in that its goal is to protect user data and privacy, while also getting the most accurate results possible by not having a paywall, or content blocking function.

How far are we from a 100% AI polling solution?

I would imagine we aren’t that far off. There has definitely been innovation in this space. It’s one thing to ask someone who they are voting for, but there are examples of AI making election predictions based on sentiment on social media and not asking voters directly. AI is not perfect, but eventually inferences might  be able to be made accurately without directly interacting with voters.

What is the holy grail for survey professionals? 

The holy grail is being able to survey a large random sample of people and have them answer multiple questions. In addition, one needs to be able to reach specific groups of people (e.g. Nike might only want to survey people who have purchased Nike products in the past).

Invisibly’s Realtime Research product leads the polling and market research space in finding the largest random sampling of respondents, and we are building towards being able to facilitate communication between brands and individuals that fall into a desired cohort (e.g. people who have a verified purchase of a Nike product).

There are trends that don’t just apply to market research but research in general. There has been a decrease of in-person surveys due to pandemic. This means an increase in other types of surveys such as telephone surveying. However, we recently saw the benefits of Invisibly’s online style surveys proven over telephone surveys due to the chaos that occurred with the 2020 Presidential Election, for which many of the major polls still were done by telephone and ended up being very far off the actual election results. Some major research institutions such as Pew Research shut down in-person surveys due to the pandemic and were exploring telephone surveys. The question is whether or not virtual research trends will remain after the pandemic. 

Leave a Reply