Skip to main content
. Author manuscript; available in PMC: 2014 Dec 31.
Published in final edited form as: Econ Inq. 2012;50(1):82–93. doi: 10.1111/j.1465-7295.2010.00345.x

TABLE 4.

Effect of Education on Endogamy, Robustness Checks

Endogamy County Group
1		 Country
2		 State of Residence: Movers
3		 Birth State: Movers
4		 County Group: Non-movers
5		 County Group: Had First Child After Age 25
6
Years of Schooling −0.005**
(0.002)		 0.002
(0.002)		 0.000
(0.002)		 0.000
(0.003)		 0.008
(0.010)		 0.002
(0.004)
Age 3.997**
(0.394)		 5.505**
(0.300)		 7.377**
(1.264)		 4.429**
(1.364)		 2.885*
(1.282)		 3.622**
(0.481)
Non-English Native Language −9.265**
(1.808)		 −12.789**
(1.132)		 −42.544**
(11.928)		 −24.798*
(11.723)		 −1.426
(3.738)		 −5.991**
(1.268)
Percentage Sharing Ethnicity 0.002*
(0.001)		 0.009**
(0.002)		 0.009**
(0.001)		 0.010**
(0.002)		 0.011**
(0.004)		 0.007**
(0.001)
Percentage Sharing Ethnicity Squared −0.043**
(0.014)		 −0.139**
(0.020)		 −0.133**
(0.013)		 −0.129**
(0.019)		 −0.182**
(0.055)		 −0.109**
(0.020)
Years of Schooling × (Mean Ethnic Schooling − Mean Schooling) 0.007**
(0.000)		 0.006**
(0.000)		 0.006**
(0.001)		 0.006**
(0.001)		 0.004*
(0.002)		 0.004**
(0.001)
Mean Ethnic Schooling − Mean Schooling 0.062**
(0.010)		 0.081**
(0.011)		 0.088**
(0.011)		 0.106**
(0.012)		 0.048
(0.051)		 0.035
(0.024)
Constant −0.456**
(0.053)		 −0.287**
(0.036)		 −0.250**
(0.044)		 −0.232**
(0.067)		 −0.313+
(0.177)		 −0.160*
(0.079)
Ethnicity fixed effects Yes No No No No No
County group fixed effects Yes No No No No No
Observations 39,489 39,489 7,590 7,590 341 5,275
R-squared 0.23 0.15 0.11 0.08 0.25 0.13

Notes: Column 1 adds ethnicity and county group fixed effects to the last specification shown in Table 3. Column 2 computes the percentage sharing ethnicity and average schooling variables at the country-wide level as opposed to the county group level. Results shown in columns 3 and 4 are for a sample of people who reside in a different state than their state of birth. The percentage sharing ethnicity and average education variables are computed at the state of residence and state of birth levels, respectively. The state of residence and state of birth regressions are run using the Form 2 State sample of the 1970 U.S. Census instead of the Metro sample in order to increase sample size. In column 5, the main specification is run on a sample of males who did not change counties within the 5 years prior to the survey but whose spouses did. In column 6, the baseline specification is run on a sample of males who had their first child (eldest child living in the household) after the age of 25. To increase the probability that the male’s first child is the eldest child living in the household, the sample consists of only males whose eldest child is younger than age 13. Standard errors are clustered on ethnicity-county group cells in columns 1, 5, and 6; on ethnicity in column 2; ethnicity-state cells in column 3; and ethnicity-state of birth cells in column 4.

**

p < .01,

*

p < .05,

+

p < .10.