Fatigue during the COVID-19 pandemic: Evidence of social distancing adherence from a panel study of young adults in Switzerland

Abstract

In this paper we analyze panel data (N = 400) to investigate the change in attitudes towards the Covid-19 measures and the change in compliance behavior between the first and second lockdowns in a sample of young adults from the University of Bern, Switzerland. We find considerable fatigue. While respondents expressed high acceptance of and compliance with the Covid-19 measures during the first lockdown, both acceptance and compliance behavior decreased substantially during the second lockdown. Moreover, we show via a structural equation model that respondents' compliance behavior is largely driven by the perception of how others behave and by the acceptance of the Covid-19 measures. All other effects scrutinized e.g., individual and social risk perception, trust in politics, and pro-social orientations affect compliance behavior via the acceptance of Covid-19 measures. We also conduct two tests of causality of the estimated relation between attitudes towards the measures and social distancing behavior. The first test incorporates the effect of compliance behavior reported during the first lockdown on attitudes during the second lockdown. The second test involves estimating a first difference panel regression model of attitudes on compliance behavior. The results of both tests suggest that the effect of Covid-19 attitudes on social distancing behavior can be interpreted causally.

Fig 1. Results of a structural equation model explaining compliance with Covid-19 measures during the first lockdown in Switzerland.

Note: N = 493. All reported coefficients are unstandardized and statistically significant at least at the 5% level [1].

Fig 2. Acceptance of Covid-19 measures during the first and second lockdowns.

Note: N = 364. Results for respondents who participated in both waves. Each measure was surveyed via five-point Likert scales ranging from “do not support at all” (1) to “support very much” (5). The figure displays the proportion of respondents supporting a measure weakly or strongly (categories 4 and 5).

Fig 3. Adherence to social distancing measures during first and second lockdowns.

Note: N = 364. Results for respondents who participated in both waves. Adherence was surveyed via five-point Likert scales, and the number of people met in the last 7 days. The figure displays the proportion of respondents, who stayed at home “mostly” or “strictly” (categories 4 and 5), the proportion of respondents, who “rarely” or “never” made exceptions (categories 4 and 5), and the average number of people met. The differences between the first and second wave are all statistically significant.

Fig 4. A structural equation model of compliance with the coronavirus distancing measures.

Note: N = 364. All reported coefficients are unstandardized and statistically significant at least at the 5%-level.

Fig 5. A test of the assumed causal structure via structural equation modeling.

Note: N = 364. All reported coefficients are standardized and statistically significant at least at the 5%-level. Included control variables are individual risk, social risk, household risk, descriptive norm, trust in politics, prosocial, gender, and social desirability.