Table 3.
Linear probability model results for the effect of disability on educational, labor market, and marital outcomes.
| Currently enrolled | Appropriate grade for age | Educational attainment | Currently employed | Employed in a salaried job | Married before age 15 | Married by age 30 | Age at first marriage | |
|---|---|---|---|---|---|---|---|---|
| Has disability | −0.165*** | −0.069*** | −0.433*** | −0.214*** | −0.009 | −0.011** | −0.300*** | 0.443*** |
| (0.006) | (0.007) | (0.067) | (0.010) | (0.011) | (0.003) | (0.007) | (0.062) | |
| Female | −0.017*** | 0.012*** | −0.653*** | −0.188*** | −0.104*** | 0.117*** | −0.172*** | −2.785*** |
| (0.001) | (0.001) | (0.019) | (0.003) | (0.004) | (0.002) | (0.003) | (0.024) | |
| Has a disability × Female | −0.006 | −0.025 | −0.778*** | 0.033 | 0.035 | 0.021 | −0.060*** | −0.169 |
| (0.008) | (0.010) | (0.156) | (0.021) | (0.029) | (0.011) | (0.012) | (0.152) | |
| Constant | 0.684*** | 1.001*** | 2.679*** | 0.083*** | 0.315*** | 0.636*** | 0.786*** | 13.156*** |
| (0.003) | (0.003) | (0.213) | (0.021) | (0.039) | (0.023) | (0.017) | (0.243) | |
| Number of individuals | 1,234,095 | 1,051,251 | 607,486 | 644,173 | 528,889 | 1,022,169 | 760,889 | 1,022,169 |
| Number of sibling-pairs | 566,855 | 519,749 | 474,317 | 495,273 | 437,311 | 798,056 | 667,848 | 798,056 |
| Within R-squared | 0.12 | 0.28 | 0.19 | 0.30 | 0.05 | 0.14 | 0.17 | 0.19 |
Notes: This table shows the coefficient on whether an individual has a disability, their gender, and the interaction between the two, obtained from estimating equation (1) in the sample of individuals shown in Table 2. All regressions include sibling-pair fixed effects as well as birth year fixed effects. They also include the “generation” the individual belongs to (see text). In addition, models with ‘currently employed’ and ‘employed in a salaried job’ include educational attainment as a covariate. In the models for labor market outcomes, the coefficient does not change substantively when the sample is limited to individuals above the age of 18. The standard errors are clustered at the household level. As discussed in the text, the sample on which the regression is estimated differs based on the outcome. *p < 0.0125, **p < 0.00625, ***p < 0.000125. The cutoffs are Bonferroni-corrected for multiple (specifically, eight) hypotheses and correspond to 10%, 5% and 1% significance level, respectively.