Facebook surveys provide rapid insights into people’s behaviors during the COVID-19 pandemic

How did people change their behavior during the COVID-19 pandemic? In an article recently published in EPJ Data Science, Daniela Perrotta and co-authors illustrate how timely data collection via Facebook, along with appropriate survey designs and statistical methods, can help answer this question.

The COVID-19 pandemic hit our lives in unprecedented ways.

a lack of data makes it difficult to assess people’s responses during outbreaks

In an effort to curb the spread of SARS-CoV-2, governments have implemented a variety of so-called non-pharmaceutical interventions (NPIs), such as stay-at-home measures, physical distancing regulations, constraints on non-essential movements, and recommendations for good hand hygiene and use of face masks.

While achieving universal adoption of public health recommendations and compliance with the NPIs is critical to minimizing the burden of COVID-19, this is only possible to the extent that people are willing to change their behavior and comply with the recommended or mandated NPIs.

Assessing whether people actually change their behavior and understanding the behavioral choices, barriers and drivers within a population, is therefore key to increasing situational awareness and informing interventions.

Yet, a lack of data makes it difficult to assess people’s responses during outbreaks. This may hinder effective communication strategies and evidence-based decision-making.

Nonetheless, over the past year we have witnessed a growing effort in collecting primary data and leveraging alternative data sources to derive insights to fight the global outbreak.

Rapid data collection via Facebook

we provide insights from a total of 71,612 completed questionnaires

In this work, we present our efforts to collect key data on people’s behavioral responses to the pandemic.

This is based on the COVID-19 Health Behaviour Survey (CHBS), a cross-national online survey that we conducted via Facebook at a crucial time when the first wave of the pandemic was sweeping Europe and North America.

We leveraged Facebook’s advertising platform to disseminate ad hoc advertisements among Facebook users and draw their attention to our survey.

This approach allowed us to quickly and continuously reach large numbers of survey respondents in several countries at the same time.

Participants recruited from Facebook are not necessarily representative of the general population and attention to who is represented in the data is critical to ensure that the conclusions drawn are valid.

In our analysis, we ensure this by appropriately reweighting the respondent samples in central observable characteristics such as age and gender.

Based on this, we provide insights from a total of 71,612 completed questionnaires collected between March 13 and April 19, 2020 in Belgium, France, Germany, Italy, the Netherlands, Spain, the United Kingdom, and the United States.

Compared to other studies, our survey stands out thanks to its timeliness, cross-national and comparative nature, and population coverage, which allowed us to provide important insights related to COVID-19.

This includes insights into people’s perceptions of the threat, adoption of preventive behaviors, and trust in governments and healthcare systems in their preparedness to deal with the pandemic.

Behavioral insights related to COVID-19 for communication campaigns

we have seen a rapid adoption of protective face masks when masks were not mandatory yet

Our findings show, for example, that women are more likely to perceive the pandemic as a very serious threat and to adopt and comply with public health recommendations (Figure 1).

Since the case fatality rate for COVID-19 is substantially higher for men, this evidence is relevant for policy makers in designing communication campaigns, which may need to be gender-based in order to tackle this difference in attitudes and behaviors.

Figure 1. Adoption rate of behaviors by country and gender. The adoption rate is defined as the weighted proportion of respondents who adopted a given behavior compared to the pre-pandemic period.
© The Authors

Likewise, the age-specific patterns we observe in the perception of threats and adoption of preventive behaviors can inform future communication campaigns.

For example, we observe that younger people perceive high threat to their family, whereas older adults perceive high threat to themselves (Figure 2).

This is consistent with the evidence that older adults are at higher risk of complications following a COVID-19 infection.

Figure 2. Relationship between the threat of COVID-19 perceived by women (y-axis) and men (x-axis), where colors indicate the different levels of society and sizes indicate the age of respondents. Perceived threat data points are means of Likert-scale values normalized in the range 0-1.
© The Authors

Furthermore, the fact that we collected data from multiple countries simultaneously enabled us to capture differences across countries that were at different stages of the pandemic and with different control strategies in place.

For example, we have seen a rapid adoption of protective face masks in the early days when masks were not mandatory yet, but this varies substantially from country to country as a result of different NPIs.

In conclusion, our study contributes to narrow the gap in behavioral data and provide important insights to enhance our understanding of the COVID-19 pandemic.

As the pandemic progresses with new waves of infections, such insights are crucial for policy makers to increase situational awareness and guide the decision-making process.

View the latest posts on the On Physical Sciences homepage