Table 2.

Random effect models for repeated measures for the relative reductions in the average degree of the co-location network with respect to the baseline.

independent variables [estimates (s.e.)]	(A)	(B)	(C)
dependent variable: relative reduction of 〈k〉 in pre-lockdown
population density	3.33*** (0.89)	2.85*** (0.85)	3.29** (0.89)
old age index	2.43* (1.04)	—	0.74 (1.29)
high education	3.17*** (0.99)	—	3.36** (1.14)
female	−1.93 (1.11)	—	−3.21* (1.26)
commuters	—	−0.49 (0.89)	0.24 (1.08)
residential buildings	—	2.26** (0.87)	2.38* (1.08)
intercept	16.74*** (3.01)	16.74*** (3.08)	16.74*** (3.08)
no. of days	16	16	16
AIC	5429.47	5439.31	5428.65
dependent variable: relative reduction of 〈k〉 in lockdown
population density	0.001 (0.45)	0.75 (0.47)	0.08 (0.45)
old age index	−0.10 (0.52)	—	−0.80 (0.65)
high education	9.79*** (0.5)	—	10.88*** (0.57)
female	−4.32*** (0.56)	—	−4.66*** (0.63)
commuters	—	−2.88*** (0.49)	1.98*** (0.54)
residential buildings	—	0.15 (0.48)	1.01 (0.54)
intercept	62.21*** (0.87)	62.21*** (0.87)	62.21*** (0.87)
no. of days	38	38	38
AIC	12094.05	12399.40	12082.48
dependent variable: relative reduction of 〈k〉 in phase 2
population density	0.39 (0.83)	0.48 (0.84)	0.27 (0.82)
old age index	0.63 (0.96)	—	−1.37 (1.19)
high education	10.31*** (0.92)	—	9.8*** (1.05)
female	−5.32*** (1.03)	—	−6.98*** (1.16)
commuters	—	−4.68*** (0.88)	−1.09 (0.99)
residential buildings	—	1.34 (0.86)	2.8** (0.99)
intercept	39.34*** (1.68)	39.34*** (1.68)	39.34*** (1.68)
no. of days	14	14	14
AIC	4584.88	4665.17	4579.13

Significance levels: *p < 0.05; **p < 0.01; ***p < 0.001.