Skip to main content
. Author manuscript; available in PMC: 2022 Oct 11.
Published in final edited form as: Soc Sci Res. 2021 Feb 13;96:102539. doi: 10.1016/j.ssresearch.2021.102539

Table 4.

Multinomial logistic regression of migration type for those who migrated before (ref. did not migrate) (N = 100,400).

Returned to county where grew up Migrated elsewhere
RRR SE RRR SE
Marital status (ref = separated)
Married .281*** .025 .590 *** .039
Never-married .415*** .044 .532 *** .041
Female .851* .054 .885 *** .028
Parent in county where grew up 2.490*** .228 .908 .053
Sibling in county where grew up 1.478*** .124 1.010 .056
Child in county where grew up 4.264*** .713 1.220 .208
Non-resident parent within 50 km .205*** .033 .566 *** .036
Non-resident sibling within 50 km .394*** .048 .583 *** .036
Non-resident child within 50 km .531*** .078 .493 *** .039
No non-resident kin in PSID data 1.167 .186 .661 *** .055
Parents living together .980 .077 1.074 .061
Resident parent or sibling .792 .108 .754 ** .074
Resident child < 18 .805* .068 .640 *** .032
Resident child 18+ .816 .137 .651 *** .051
Age .942*** .006 .969 *** .002
Race (ref = non-Latino white)
Non-Latino black .833* .076 .538 *** .038
Other .473*** .079 .508 *** .058
Homeowner .311*** .026 .333 *** .017
Total family income (logged) .906*** .025 .978 .023
Education in years .974 .014 1.089 *** .012
Employment (ref = employed)
Unemployed 1.829*** .181 1.442 *** .098
Not in labor force 1.900*** .160 1.795 *** .089
Size of largest city in county (ref=<25,000)
25,000–99,999 1.086 .099 1.025 .057
100,000 or more .784** .066 .838 ** .054
Region (ref = Northeast)
Midwest 1.153 .145 1.170 .103
South 1.558*** .175 1.406 *** .112
West 1.582*** .199 1.600 *** .143
Year (ref = 1983 to 1989)
1990 to 1999 .968 .083 1.122 * .061
2001 to 2013 1.030 .178 .928 .087
Study interval length 1.843*** .284 1.921 *** .153
Constant .452* .165 .115 *** .026
N (person-intervals) 1,411 5,001
*

p < .05,

**

p < .01,

***

p < .001.

Note: Pooled results of 10 imputation datasets; standard errors are clustered by family identification number.