In fact, we find only two cases in which the estimated effect of parental death is
statistically significant at the 0.05 level—sons are less likely to be married if their
fathers died early and daughters tend to eam more if their mothers lived longer. We
view this latter result with caution since it is not robust to seemingly minor specification
and estimation decisions. The former result, however, is consistent with the
findings reported in the previous section when we allowed the effect of father's
absence to be nonlinear. Thus, in this respect, we find a consistent effect of father's
death or absence on son's marital status. In addition, mother's death adversely affects
son's educational attainment and daughter's cognitive ability although these are significant
at only the 0.1 level.