Table 3 Change in adherence estimates with time spent in residential locations as dependent variable (mobility data)

From: A worldwide assessment of changes in adherence to COVID-19 protective behaviours and hypothesized pandemic fatigue

 

Model 1

Model 2

Model 3

Model 4

 

Time spent in residential locations

Time spent in residential locations

Time spent in residential locations (residuals)

Time spent in residential locations

 

Estimate

CI

P

Estimate

CI

P

Estimate

CI

P

Estimate

CI

P

Level 1

            

Time

   

−0.903

−1.041 to −0.764

<0.001

−0.783

−0.918 to −0.648

<0.001

−0.744

−0.888 to −0.599

<0.001

Time2

   

0.191

0.136 to 0.245

<0.001

0.197

0.148 to 0.246

<0.001

0.156

0.099 to 0.212

<0.001

School closing

0.059

0.053 to 0.064

<0.001

0.047

0.042 to 0.052

<0.001

   

0.047

0.042 to 0.052

<0.001

Workplace closing

0.045

0.039 to 0.051

<0.001

0.056

0.052 to 0.061

<0.001

   

0.055

0.050 to 0.059

<0.001

Cancel public events

0.008

0.003 to 0.013

0.003

−0.007

−0.011 to −0.002

0.005

   

−0.005

−0.010 to −0.001

0.024

Restrictions on gatherings

0.005

−0.001 to 0.010

0.061

0.019

0.014 to 0.023

<0.001

   

0.019

0.014 to 0.024

<0.001

Close public transport

0.038

0.033 to 0.043

<0.001

0.034

0.030 to 0.039

<0.001

   

0.032

0.027 to 0.036

<0.001

Stay at home req.

0.036

0.029 to 0.042

<0.001

0.046

0.041 to 0.052

<0.001

   

0.043

0.037 to 0.048

<0.001

Restrictions on internal movements

0.028

0.024 to 0.032

<0.001

0.025

0.021 to 0.028

<0.001

   

0.023

0.020 to 0.027

<0.001

International travel controls

0.034

0.028 to 0.039

<0.001

0.021

0.016 to 0.026

<0.001

   

0.018

0.013 to 0.023

<0.001

COVID-19 searches

         

0.053

0.048 to 0.059

<0.001

Increase in deaths period

         

0.064

−0.038 to 0.166

0.217

Mean temperature

         

−0.124

−0.262 to 0.015

0.080

Northern Hemisphere × mean temperature

         

0.105

−0.042 to 0.252

0.160

Level 2

            

Northern Hemisphere

         

−3.888

−7.602 to −0.174

0.040

Observations

32,998

  

32,998

  

32,998

  

31,200

  

Number of countries (level 2)

124

  

124

  

124

  

120

  

R2 (level 1)

0.506

  

0.699

  

0.366

  

0.706

  

Var. (r1jk)

   

0.559

  

0.551

  

0.584

  

Var. (r2jk)

   

0.089

  

0.071

  

0.079

  
  1. Models 1, 2 and 4 have the change in time spent in residential locations as the dependent variable. Model 3 has the residuals from model 1 as the dependent variable. Coefficients were estimated in multi-level models with two levels: time (level 1, days) and countries (level 2). Time refers to time since first required measure. Coefficients of time and time2 are random effects at the country level, with unrestricted covariances across random effects and a first-order auto-regressive error term covariance structure. The intra-class correlation coefficient in the null model is 0.41 at country level. Residuals models were estimated with independent random effects for linear and quadratic time terms due to convergence issues. Results hold in OLS models with dummy variables representing countries (fixed effects) and clustered standard errors at the country level (Supplementary Table 1c).