Table 2.

Generalized linear model results for crime rates by year and crime type (1:100,000)

	Effects
Crime type	Year Waldχ2b pES	Lock Waldχ2b pES	Inter Waldχ2b pES
Murder and manslaughtera	0.22 0.00 {1.00}	0.02 0.19 {1.21}	0.22 − 0.54 {0.58}
	.642	.889	.642
	.046	.014	.046
Assault	13.51  − 1.15	19.52  − 1.51	1.79 − 1.31
	< .001	< .001	.182
	.360	.433	.131
Sexual assault	3.70 − 0.14	1.54 − 0.03	1.15 − 0.36
	.054	.214	.283
	.189	.122	.105
Drug-related crimes	10.03 − 1.84	17.37 − 2.72	1.13 − 1.86
	.002	< .001	.289
	.311	.409	.104
Robbery	0.33 − 0.46	0.11 − 0.06	1.87 0.16
	.565	.736	.171
	.056	.033	.134
Property-related crimes	1.24 − 0.23	2.19 − 0.94	3.91 1.07
	.266	.139	.048
	.109	.145	.194

Year 2019 — 0, Year 2020 — 1; Lockdown — 1 and 0 — otherwise. This data series was assumed to have the binomial distribution and was linked to the Logit function rather than the normal distribution and the identity link which were set for all the other series. The odds of the Logit coefficients are given in curly brackets. For detailed results, see Table 5

Int. interaction effect, Lock. lockdown periods

^aIn each cell for effect we show: Wald’s χ² with one degree of freedom, the p-value of that effect (p), and effect size (ES), b for the unstandardized regression coefficient