What AI Told Me About Wearing a Mask
AI reveals a psychographic division in our nation
Over the past 8 months, wearing a face mask to protect yourself and others against the spread of the Covid-19 virus has become a political issue in America, partly due to a spotty and inconsistent regulation and guidance from political leaders and health experts. While this never should have become a polarized issue tied to anything other than health, for some, it undeniably has.
I wanted to understand more about this polarization, and the reasons behind the belief system and rationale for wearing, or not wearing, so I turned to a client of my company, Next Media Partners, that was perfect for the job.
Beehive AI is a unique platform that creates a mass open dialogue with consumers to help marketers understand customers and increase conversion. The technology interprets conversations at scale in order to uncover customer motivations and needs that can not be uncovered through traditional methods.
In my case, I used Beehive AI to collect and analyze data about wearing a face mask just prior to the U.S. national election. Using an online survey the platform engaged 18+ y/o Americans and selected the right questions to ask in order to encourage people to share their opinions and motivations openly and freely in their own words. The data collected continuously is enriched with demographics data and then automatically classified in order to construct psychographic profiles.
We ran the system for about a week spanning before and after the election. Some of you may have taken this simple online survey on Mask Wearing in America. Thank you, if you did!
The results were fascinating. Using artificial intelligence, Beehive AI analyzed the answers provided by more than 2,000 respondents and organized its findings into segments based upon their psychographic profile (values, opinions, attitudes, interests, and motivations). By parsing the open-ended free form answers from these participants, the AI created by Beehive semantically classified individuals with high granularity and then used clustering algorithms in order to uncover patterns and nuanced personas.
The survey achieved a good distribution of respondents across all 50 United States, with a little more than 50% female. Among the surprise findings were that 95% of these respondents are wearing masks. Media reports might lead one to believe that there would be a much higher percentage of those not wearing masks. However, I was fine with this distribution, because I was mostly interested in the individual’s motivations to wear or not to wear a mask. We branched our survey questioning at the point where the respondent informed us that they wore a mask or not. For those who wear masks we asked — “What do you think will convince people who are *not* wearing masks or face coverings to start wearing them?”
The Beehive AI solution was able to identify 50 unique motivations associated with the beliefs of these mask wearing persons. The top 5 were:
- Getting sick themselves (41%)
- Nothing (21%)
- Death of a family member (6%)
- Education (4%)
- A fine or mandate (4%)
Conversely, for those individuals who do not wear masks we asked — “What would you convince you *to wear* a mask or face cover?” The answers to this question did not exactly align with the belief system of the mask wearers, but there were similarities. Fifteen unique motivations were uncovered. The top 5 were:
- Nothing (34%)
- Getting sick (11%)
- If they are required (9%)
- Proof that they are effective (9%)
- An increase in deaths (7%)
Two items stood out for me in this comparison. For mask wearers, there was a fairly common belief that if the non-wearers were to get sick themselves it would change their attitude toward mask wearing. However, when a similar attitudinal question was asked of the non-wearers, they answered overwhelmingly that nothing would convince them to wear a mask — and getting sick was only a top issue for 11% of those respondents.
We also noticed a shift in empathy and caring over the course of the survey. As the platform continuously collects data , we were able to compare attitudes based upon when the questions were answered. Since we spanned across the U.S. Election Day we were able to see a dramatic shift in empathetic leaning before and after November 3rd. The sentiment “I care about others” peaked just before that date at more than 80%. However, immediately following the election that number dropped to less than 40% .
This alarming shift in attitudes uncovered through Beehive’s AI analysis has led us to launch additional analyses in a series we are calling “A Nation Divided”. It was apparent from this initial analysis that we uncovered something that many Americans may feel, but few are able to quantify and qualify. We will continue to publish the findings of these analyses and welcome additional ideas on the questions to ask, and commentary on the findings that we present.
Follow me here on Medium to get notified as we publish more updates.
