Exhibit 3. Predicted Probabilities and Relative Risk of Insurance Status Among Gay Men Compared To Heterosexuals, Nonelderly Men Ages 18–64 In California, Average Of Years 2001, 2003, And 2005.

	Entire population				Employed population
	All men		Partnered/married men		All men		Partnered/married men
	Gays/heterosexuals (PP, %)	Gays compared to heterosexuals (RR)	Gays/heterosexuals (PP, %)	Gays compared to heterosexuals (RR)	Gays/heterosexuals (PP, %)	Gays compared to heterosexuals (RR)	Gays/heterosexuals (PP, %)	Gays compared to heterosexuals (RR)
Sample size		46,535		26,063		31,100		18,260
HEALTH INSURANCE STATUS
Uninsured	15/17	0.88	17/11	2.02**	13/14	0.82**	14/9	1.9
Public insurance	12/10	1.21	11/7	2.00**	4/6	0.66	3/5	0.67
Own ESI	56/55	1.03	58/61	0.9	73/69	1.09**	75/74	0.99
Dependent ESI	8/11	0.78	6/14	0.42**	8/8	0.93	5/11	0.44**
Privately purchased	8/7	1.07	8/6	1.23	3/3	0.82	4/2	1.96

Source: California Health Interview Surveys, 2001, 2003, and 2005.

Notes: We bootstrapped 95 percent confidence intervals (available in the full version of this exhibit in Appendix E, which can be accessed by clicking on the Appendix link in the box to the right of the article online), by sampling stratum. Relative risks (RR) were computed from multinomial logit models that adjusted for age, race/ethnicity, language of interview, education, income as percentage of federal poverty level, minor children in household, town/rural residence, employment status/work hours, working for a firm with more than 100 employees, industry (in employed group), citizenship, and survey year. Employed group excludes self-employed. Samples include bisexuals. PP is predicted probability. ESI is employer-sponsored insurance.

^**p < 0:05