Six months into the pandemic, we are still struggling to respond effectively to COVID-19.
In principle, it should be a relatively straight-forward process to gaining consensus on what to do. First, through the work of our vastly capable research institutions and scientific community, discover the best practices for mitigating the spread and danger of the disease. Then, create policies that communicate these best practices and provide incentives to maximize compliance. As we discover new information, we iterate from there. It is expected that information will change and that mistakes will be made. What is important is to have a system in place for discovering, communicating, and incentivizing best practices on a continual and consistent basis.
However, as of yet, we have no such system. Many of the questions that we had 6 months ago, we still have today: Should we all wear masks? Should we refrain from seeing our friends? Will vaccinations work? Is it possible to reach herd immunity? Despite 6+ months of data and experience, there is still little consensus on the answers to those questions. Yes, the scientific community has been largely consistent. But it takes more than just scientists. It requires a cohesive political apparatus to communicate that consensus, both at the national level and the local level, in a way that people can understand and act upon.
Instead, the messaging on how we should best protect ourselves has been, to say the least, inconsistent. Accordingly, it is natural to wonder:
How do differences in demographics and behaviors correlate to complying with recommended COVID-19 safety measures?
To help answer this question, UC Berkeley recently used Embee’s DIY research portal, ResearchDesk, to field a survey to 1,301 panelists. This survey asked a number of questions about the observance of COVID-19 safety measures, such as the propensity to wear masks on public transportation or when shopping, the frequency of washing hands and other safety measures. Embee compiled the responses into a scoring system and bucketed the respondents into equal buckets of “high observers” and “low observers.”
Using this analysis, we can quickly see several interesting correlations between compliance with safety precautions and various demographics, news sources, political opinions, and other information. The chart below summarizes several of these findings. (For example, in the chart below: male respondents were 3.7% more likely to be in the “high observance” group, with females 2.5% less likely.)
Looking closer at the data
- 18-24s and 55+ buckets are much more likely to report observant behaviors. The older group, who are the most vulnerable in general, are obeying COVID measures as expected. But the fact that the youngest group over-indexed on observing is interesting. Maybe this group has fewer pressures on them to live life normally.
- There is a strong correlation between the stability of a home and observance of COVID measures, which may be because the more stable a home, the easier it is to shelter in place. If authorities are to increase social distancing compliance, these results suggest that making sheltering at home better is critical.
- When UC Berkeley asked the source of people’s COVID news (with multiple answers allowed), there was a correlation between COVID compliance and the breadth of the news source: National news sources correlate to COVID compliance, while friends and family or social media correlate to the least level of COVID compliance.
- It should not be surprising that there is a high correlation between liberal political attitudes and complying with COVID restrictions. This has been widely documented. Our data confirms that the strongest correlation we looked at regarding compliance with COVID restrictions was the person’s position on the liberal-conservative scale.
In addition to evaluating the correlations above, we also looked at correlations between levels of observance and mobile behaviors.
For example, a person’s location data—where they go and what they do—is a window into the behavior and observance of COVID-19 restrictions. Embee can extract that kind of data to see where a respondent goes every week and how frequently they’re out. Since the start of the pandemic, we have noticed that people who are out and about the most are also the least observant of COVID-19 restrictions.
There are good reasons that people disagree
It can be tempting to dismiss disagreements on how best to respond to COVID-19 as being principally rooted in misinformation. However, our view is that is too limited a perspective. One of the fundamental assumptions that underpins our research partnership between UC Berkeley and Embee on COVID-19 is that, when people disagree, they do so for diverse but important reasons. These reasons may be related to information sources, yes, but they may also be related to physical and mental health, economic and social needs, beliefs, or other factors.
Using a survey, mobile data, and the best principles of AI and behavioral science, we aim to learn the many reasons and present their diversity and validity so that America’s leaders can then provide the kind of assistance, information, and education people need—which will be very important for the control of COVID-19.
We’ll keep an eye on things
There is a significant amount we can learn about adherence to COVID-19 restrictions from the data that streams our way every day. UC Berkeley and Embee Mobile will continue to track it over the coming weeks and months through the unique view we have into our respondents’ mobility and behavior.
(Note: after the original version was published in our newsletter, some edits were made for clarity)