Table C4.

Sensitivity Analysis for Role of Neighborhood Violence in Explaining the Effect of Neighborhood Disadvantage on Teenage Pregnancy (males)

	β YU
rVU	.00	.01	.02	.03	.04	.05
.00	.0079*	.0079*	.0079*	.0079*	.0079*	.0079*
	21%	21%	21%	21%	21%	21%
.05	.0079*	.0074*	.0069*	.0064	.0059	.0054
	21%	20%	18%	17%	16%	14%
.10	.0079*	.0069*	.0059	.0049	.0039	.0029
	21%	18%	16%	13%	10%	8%
.15	.0079*	.0064	.0049	.0034	.0019	.0004
	21%	17%	13%	9%	5%	1%
.20	.0079*	.0059	.0039	.0019	--	--
	21%	16%	10%	5%
.25	.0079*	.0054	.0029	.0004	--	--
	21%	14%	8%	1%

Notes:

β_YU is hypothetical coefficient on standardized unobserved variable in linear probability model predicting teenage pregnancy and controlling for all other control variables.

r_VU is hypothetical partial correlation between standardized unobserved variable and violence scale.

In each cell, the top number is the expected coefficient on the violence scale variable if the unobserved variable where included in the linear probability model predicting teenage pregnancy, and the bottom number is the resulting proportion of the neighborhood disadvantage effect accounted for by neighborhood violence.

Note: 2 X SE of β on violence coefficient in linear probability model predicting teenage pregnancy = 0.0068

^*coefficient on violence scale would remain statistically significant