Supplementary
Supplementary Figure 1, Linear regression of surveyed indicators over 40 surveys between May
5, 2020 and February 8, 2021. We estimate the coefficients using linear regression for each
indicator (stratified by factors in Supplementary Table 2) per 100 daily cases and only show those
with adjusted R2 larger than 0.3. The value on each edge indicates the coefficient of significant
linear regression of the two lined factors across the 40 surveys (Supplementary Table 1).
Supplementary Figure 2, Linear Stepwise regression of Google mobility measures and protective
behaviour proportions with information on coefficients in Supplementary Table 5. We examine
the correlation between each type of surveyed protective behaviours and six daily mobility
movement trends of Google by a stepwise regression analysis to add or remove predictors with
the criterion of p-value for F test. Lines denote those selected mobility indexes for each behaviour.
Supplementary Figure 3, Epidemiological model of COVID-19 transmission in Hong Kong. Upon
infection, susceptible individuals (S) progress to being exposed (E). A fraction of cases become
asymptomatic infectious (A) with lower infectiousness before recovering (R); the remaining cases
progress to presymptomatic (P), where they are moderately infectious but not yet symptomatic,
followed by symptomatic infectious (Y) and then either recover or die (R).
Supplementary Figure 4, Overview of survey and epidemic data. Weekly proportions of
protective behaviour and risk perception from weekly cross-sectional telephone surveys, daily
reported cases on average in a week by reporting date, and real-time reproduction number on
average in a week (Supplementary Table 2).
Supplementary Figure 5, Google mobility data for each of the location categories. Google
compares visitor daily numbers to specific categories of location to that during the baseline period
(the 5-week period from January 3 to February 6, 2020) before the pandemic outbreak. Six Google
mobility measures are collected to track how the numbers of visitors to places of (1) retail and
recreation, (2) grocery and pharmacy stores, (3) transit stations, (4) workplaces, (5) residential areas, and (6) parks have changed compared to baseline days4.