Table 5.
Heterogeneity and Robustness Checks
Stratified sample | coefficient on accepted |
N | Mean of rejected applicants |
Effect of acceptance in years |
p-value for test of equality across groups |
---|---|---|---|---|---|
Panel A. Subgroup analysis | |||||
Income above median | 0.0171* [0.009] | 3,944 | 72.78 | 1.24 | |
Income below median | 0.0200* [0.012] | 3,915 | 72.02 | 1.44 | 0.84 |
Fraction urban above median | 0.0225* [0.012] | 4,282 | 72.04 | 1.62 | |
Fraction urban below median | 0.0162** [0.007] | 3,577 | 72.87 | 1.18 | 0.64 |
Fraction foreign born above median | 0.0176** [0.008] | 3,729 | 72.98 | 1.28 | |
Fraction foreign born below median | 0.0185 [0.012] | 4,130 | 72.02 | 1.33 | 0.95 |
Panel B. Additional robustness checks | |||||
Full sample | 0.0182** [0.007] | 7,860 | 72.44 | 1.32 | |
Age <= 14 | 0.0201*** [0.006] | 7,408 | 72.32 | 1.45 | 0.25 |
Age <= 10 | 0.0173** [0.008] | 5,202 | 72.32 | 1.25 | 0.86 |
Born 1900–1920 | 0.0219** [0.008] | 6,798 | 72.32 | 1.53 | 0.12 |
Born 1900–1910 | 0.0211** [0.010] | 2,524 | 72.32 | 1.53 | 0.69 |
Born 1911–1920 | 0.0189 [0.012] | 4,274 | 72.32 | 1.37 | 0.92 |
Family size between 3 and 7 | 0.0244** [0.009] | 5,178 | 72.32 | 1.76 | 0.22 |
Notes: Table shows effect of “accepted” on log (age at death) for selected subsamples. Standard errors (in brackets) are clustered at the county level. Effects computed relative to the average for rejected boys. All coefficients are estimated from separate regressions, with standard errors clustered at the county level and include all controls: individual characteristics (age at application, age of oldest and youngest in family, dummies for number of siblings, number of letters in name, a dummy for whether date of birth is incomplete, year of application, and dummies for the marital status of the mother); state characteristics at the time of application include manufacturing wages, education/labor laws (age must enter school, age can obtain a work permit, and whether a continuation school law is in place), state expenditures in logs (education, charity, and total expenditures on social programs), and state laws concerning MP transfers (work required, reapplication required, the maximum legislated amount for the first child, and the legislated amount for each additional child); county dummies and cohort dummies.
significant at the 1 percent level.
significant at the 5 percent level.
significant at the 10 percent level.