Table B.2.

Sensitivity analysis comparing logistic regression using imputed data versus complete case analysis of White and Black men shot and killed by police during 2015–19

	Complete Case (N=3,043)						Imputed (N=3,090)
	S_OR	LB 95% CI	UB 95% CI	S_OR	LB 95% CI	UB 95% CI	S_OR	LB 95% CI	UB 95% CI	S_OR	LB 95% CI	UB 95% CI

Black^*Armed	^*0.613	0.462	0.813	0.772	0.558	1.069	^*0.597	0.451	0.790	0.754	0.546	1.040
Age				0.953	0.946	0.961				0.953	0.946	0.961
Mental Illness Signs				0.391	0.312	0.490				0.399	0.319	0.499
US Region				1.193	1.071	1.328				1.190	1.069	1.324
Black Population				147.7	69.3	315.1				164.2	77.5	347.9
Median Gross Rent				1.001	1.000	1.001				1.001	1.000	1.001
Disabled & age <65				4.E-05	4.E-07	0.005				4.E-05	4.E-07	0.004
Violent Crime Rate				1.156	1.109	1.205				1.152	1.105	1.200
Intercept	0.833^a			0.784^b			0.860^c			0.847^d
BIC	3918.0			3029.6			3995.2			3081.0

Abbreviations: WaPo=The Washington Post; S_OR=odds ratio interaction estimate; LB=lower bound; UB=upper bound; BIC=Bayesian information criterion

Interaction reference group=White and Armed

95% CI=0.635–1.094

95% CI=0.304–2.020

95% CI=0.656–1.126

95% CI=0.332–2.166

Bolded=P<0.01;

P<0.001

Note: All covariates in the adjusted models had p-values less than 0.01 but no asterisk is shown on estimates as they were not the main associations of interest.