White NH	Models 2: Stratified by Sex
		White NH
	All (n = 17,303)	Male (n = 8825)	Female (8478)
Ratio of family income to poverty	−1.384 **	−1.332 *	−1.449
	[−2.322; −0.446]	[−2.471; −0.193]	[−3.037; 0.140]
Ratio of family income to poverty (%) (Ref. low-PIR: 0–1.15)
Middle-PIR (1.16–2.80)	1.608 **	1.536 *	1.520
	[0.621; 2.596]	[0.330; 2.743]	[−0.097; 3.136]
High-PIR (2.81–5.00)	4.639 ***	4.424 **	4.474
	[1.892; 7.387]	[1.061; 7.788]	[−0.105; 9.053]
Socio-demographic variables
Female	0.150 *	NA	NA
	[0.00; 0.29]	-	-
Age	0.001	0.001	0.000
	[−0.002; 0.003]	[−0.003; 0.004]	[−0.003; 0.003]
Married	0.185 ***	0.224 **	0.150
	[0.081; 0.288]	[0.072; 0.376]	[−0.000; 0.300]
Education (Ref. less than high school)
High school graduate/GED	0.175 ***	0.180 **	0.173 **
	[0.084; 0.265]	[0.045; 0.315]	[0.046; 0.299]
More than high school	0.288 ***	0.381 ***	0.189
	[0.137; 0.439]	[0.162; 0.601]	[−0.022; 0.400]
Health system and behaviors
Covered by any kind of health insurance	0.298 ***	0.359 ***	0.236
	[0.146; 0.450]	[0.163; 0.555]	[−0.011; 0.483]
Smoking status (Ref. never smoked)
Former smoker	0.030	0.025	0.036
	[−0.047; 0.108]	[−0.112; 0.162]	[−0.060; 0.132]
Current smoker	−0.420 ***	−0.457 ***	−0.384 ***
	[−0.547; −0.292]	[−0.647; −0.266]	[−0.559; −0.208]
Drinking status (Ref. never drink)
Former drinker	0.339 ***	0.385 ***	0.325 ***
	[0.204; 0.474]	[0.162; 0.608]	[0.139; 0.511]
Current drinker	0.169 *	0.250 *	0.139
	[0.009; 0.329]	[0.003; 0.497]	[−0.083; 0.362]
Physical activities
No rigorous activities	0.190 ***	0.128	0.254 ***
	[0.089; 0.291]	[−0.038; 0.295]	[0.138; 0.370]
Self-reported health: Fair–poor (=1 if fair–poor)	0.297 ***	0.277 ***	0.317 ***
	[0.222; 0.373]	[0.167; 0.387]	[0.205; 0.430]
Live alone (=1, if alone)	−0.042	0.021	−0.082
	[−0.145; 0.061]	[−0.140; 0.182]	[−0.218; 0.054]
* p < 0.05, ** p < 0.01, *** p < 0.001. Note: For this analysis we used IV Poisson with generalized method of moments (GMM). It is frequently used in modern econometrics. The “GMM estimators make the sample moment conditions as true as possible given the data” [36].