. Author manuscript; available in PMC: 2014 May 27.

Published in final edited form as: Demography. 2012 May;49(2):629–650. doi: 10.1007/s13524-012-0093-0

Table 3.

Decomposition of the difference in the log odds of educational homogamy in prevailing and new first marriages

	Odds Ratio	95% CI for Odds Ratio^a	Percentage Contribution (%)^b
Panel A. Wives Aged 18 to 41
(1) Total Difference in Log Odds
Observed Prevailing Marriages – 1.007 – 0.868 = 0.139 Observed New First Marriages	1.15	[1.04, 1.26]	—
(2) Portion Due to Age Patterns of Homogamy and First Marriage
First Marriages (S1) – 1.058 – 0.868 = 0.190 Observed New First Marriages	1.21	[1.08, 1.34]	137.1
(3) Portion Due to Marital Dissolutions, Remarriages, Educational Upgrades
Observed Prevailing Marriages – 1.007 – 1.058 = −0.052 First Marriages (S1)	0.95	[0.85, 1.05]	−37.1
Panel B. Wives Aged 18 to 29
(1) Total Difference in Log Odds
Observed Prevailing Marriages – 1.006 – 0.885 = 0.121 Observed New First Marriages	1.13	[1.00, 1.26]	—
(2) Portion Due to Age Patterns of Homogamy and First Marriage
First Marriages (S1) – 1.014 – 0.885 = 0.129 Observed New First Marriages	1.14	[0.97, 1.30]	106.6
(3) Portion Due to Marital Dissolutions, Remarriages, Educational Upgrades
Observed Prevailing Marriages – 1.006 – 1.014 = −0.008 First Marriages (S1)	0.99	[0.89, 1.09]	−6.6

Note: Data are weighted using 1979 sampling weights.

Source: National Longitudinal Survey of Youth (NLSY79), 1979–2002.

95% confidence intervals (CIs) are estimated using 1,000 bootstrapped samples.

Estimated using the difference in log odds.