Racial/ethnic and educational inequities in restrictive abortion policy variation and adverse birth outcomes in the United States

Table 2 Linear Probability Models Examining Moderating Effects of Race/Ethnicity and Education Level on Relationship between Restrictiveness Index and Adverse Birth Outcomes: Period Linked Live Birth-Infant Death Files, 2005–2015 (N = 2,250,000)

		[1]	[2]
		Preterm Birth (n = 2,058,512)	Low Birthweight (n = 2,061,512)
*Section A:* *Race/ethnicity (categorical) x Restrictiveness Index (RI)*	Interaction Term
	RI*AIAN (Non-Hispanic)	−0.00243 [− 0.00920, 0.00433]	0.0000550 [− 0.00460, 0.00471]
	RI*AAPI (Non-Hispanic)	− 0.00173 [− 0.00399, 0.000526]	0.000101 [− 0.00137, 0.00157]
	RI*Black (Non-Hispanic)	0.00247^* [0.000145, 0.00480]	0.00225 [− 0.000804, 0.00531]
	RI*Hispanic/Latinx	−0.00140 [− 0.00406, 0.00126]	−0.00171 [− 0.00371, 0.000281]
	RI*White (Non-Hispanic)	Ref.	Ref.
	Wald Test of Interaction Term	Χ² = 1.69, p = 0.1681	Χ² = 2.01, p = 0.1068
*Section B:* *Race/ethnicity (dichotomous) x Restrictiveness Index (RI)*	Interaction Term
	RI*Black (Non-Hispanic)	0.00334^* [0.000809, 0.00587]	0.00292 [−0.000134, 0.00597]
	RI*Non-Black	Ref.	Ref.
	Wald Test of Interaction Term	Χ² = 7.02, p = 0.0107	Χ² = 3.69, p = 0.0606
*Section C:* *Education x Restrictiveness Index (RI)*	Interaction Term
	RI*Less than HS	0.00190 [−0.000200, 0.00399]	0.00417^*** [0.00178, 0.00656]
	RI*HS grad	0.00241^* [0.000446, 0.00436]	0.00360^*** [0.00164, 0.00557]
	RI*Some college	0.00186^* [0.000327, 0.00340]	0.00153^** [0.000407, 0.00266]
	RI*College grad	Ref.	Ref.
	Wald Test of Interaction Term	Χ² = 2.42, p = 0.0766	Χ² = 5.17, p = 0.0034

Note: Results are marginal effects and 95% confidence intervals (CIs) from multivariate linear probability models estimating moderating effects of race/ethnicity and education level on the relationship between the standardized lagged restrictiveness index and the probability of preterm birth and low birthweight among all 50 states and Washington, D.C. Final sample size included people not missing any data on moderators, restrictiveness index, outcomes, and covariates. All models adjust for individual-level sociodemographic characteristics, state-level sociodemographic, economic, and political characteristics, and state and year fixed effects. Standard errors clustered at the state level. Wald test of interaction term tests if the interaction term as a whole is statistically significant. p-values significant at ^* p < .05, ^** p < .01, ^*** p < .001

ISSN: 1472-6963