Fig 1.
Conceptual framework of behavioral adaptation.
The framework focuses on three key phenomena of the interplay of autonomous and policy-induced adaptation: Confluence (blue), interactions (red) and time variance (green).
Fig 2.
Incidence, mobility changes and policy stringency in Germany.
The plot depicts data stemming from [98–100] for the 16 federal states of Germany between Oct 1, 2020 and Jan 31, 2021.
Fig 3.
Marginal effect of incidence on mobility during three phases of NPI deployment.
The plot visualizes the marginal effect of 7-day incidence on predicted relative changes in mobility (compared to 2019 baseline) during three successive phases of national NPI response: NPIs implemented at county-level (until Nov 1), a national-level “lockdown light” (until December 15) and a full-scale national lockdown (from December 16). The plot is based on Model D presented in S1 File and was generated using the R package ggeffects [101].
Fig 4.
Pandemic dynamics and public visibility in Germany.
The plot depicts various data [38, 100, 105–107] related to pandemic dynamics and public interest for Mar 2020 (left side) and Oct 2020 –Jan 2021. Note: Media interest refers to the share of articles in 68 national German online news media that mention “Coronavirus” or “COVID-19”. Google search interest is an index for the search interest in a topic over a specified period of time ranging from 0–100.
Fig 5.
Diminishing risk perception over time.
The plot depicts the marginal effect of time (measured as a linear variable) on perceived risk, beginning in August 2020 and assuming a constant level for 7-day incidence. The figure was generated using the coefficients of Model C (cf. S2 File) and the median incidence value in the sample (105.84). The light blue ribbon indicates a 95% prediction interval, generated with the R package merTools [113].
Fig 6.
Credibility of government information and assessment of containment measures.
Assessment of the credibility of information issued by the German government about COVID-19 (left) and assessment of the adequacy of containment measures (right). Data: Presse- und Informationsamt der Bundesregierung [108].
Fig 7.
Association of perceived credibility of government information with assessment of containment measures.
Left side: Each segment indicates a specific combination of response categories in the data set. Right side: Conditional effect of perceived credibility of information from the government on assessment of containment measures. The posterior mean estimate of the probability of responses in each opinion category is shown for each of the four categories of perceived credibility, with error bars indicating 95% credible intervals.
Fig 8.
Autonomous and policy-induced adaptation in behavioral-epidemiological models.
In behavioral-epidemiological models, changes in contact patterns and mobility result are modeled through varying mechanisms related to autonomous and policy-induced adaptation. However, how the two adaptation mechanisms come together (“confluence”) has not been addressed sufficiently.
Fig 9.
Autonomous and policy-induced adaptation, contact rates and infections.
The top panels show the number of daily new infections for an “early” and “late” intervention (dotted line, after 7 and 21 days, respectively). The bottom panels show relative changes in contact rates resulting from policy (blue dotted line), autonomous response (orange dotted line) and the combined effect (solid green line). Details on model specification can be found in S3 File.