Do measures matter? Comparing surface-density-derived and census-tract-derived measures of racial residential segregation

Table 3 Association of MSA segregation and very preterm birth among Black and White mothers, 231

MSAs
	Main effect of segregation			Main effect of race			Segregation × Race Interaction			Model Fit (Deviance)
	RR	95% CI		RR	95% CI		RR	95% CI
BASELINE				3.11	3.07	3.15				2104
Continuous
Dissimilarity
Tract	0.91	0.83	1.00	2.96	2.92	3.01	1.68	1.52	1.87	1990
500 m	0.95	0.85	1.07	2.96	2.92	3.01	2.12	1.84	2.43	1948
4000 m	0.95	0.87	1.05	2.99	2.94	3.03	1.69	1.50	1.89	2001
Isolation
Tract	0.91	0.86	0.96	2.91	2.86	2.96	1.51	1.40	1.62	1967
500 m	0.96	0.91	1.01	2.91	2.87	2.96	1.52	1.41	1.64	1963
4000 m	0.93	0.88	0.98	2.93	2.88	2.98	1.44	1.34	1.54	1990
Binary ^†
Dissimilarity
Tract	0.97	0.95	0.99	2.91	2.85	2.97	1.11	1.08	1.14	2038
500 m	0.98	0.95	1.01	2.80	2.69	2.91	1.12	1.08	1.17	2067
4000 m	0.97	0.95	0.99	2.99	2.94	3.05	1.08	1.06	1.11	2062
Isolation
Tract	0.96	0.93	0.98	3.00	2.95	3.04	1.17	1.13	1.20	1978
500 m	0.96	0.94	0.98	2.97	2.92	3.01	1.13	1.10	1.16	2011
4000 m	0.91	0.88	0.94	3.03	2.99	3.07	1.18	1.14	1.22	2022

NOTE: Race is coded Black = 1, White = 0, and all continuous segregation indices are mean-centered so that the main effect of segregation is the effect for white women only, and the main effect of race is the excess relative risk for black as compared with white women at average segregation levels (e.g. mean-centered segregation equals zero for in an MSA with average level of segregation). The interaction term is then the excess relative risk for black women in the most compared with the least segregated MSAs. All models are adjusted for census region and MSA population size.
† Dissimilarity index dichotomized into low (< 0.6) = 0, and high (≥0.6) = 1. Isolation index dichotomized at 0.7
Data source: National Center for Health Statistics; all singleton live births to non-Hispanic white and non-Hispanic black mothers in 231 US MSA's, 2000-2002

ISSN: 1476-072X