Yet, Corak (2000) provides evidence that divorce is a much more endogenous
source of family dissolution than parental death. If divorce is much more highly
correlated with adverse unmeasured family background characteristics than is death,
we will measure a much more negative impact of divorce than of death even if the
true impact is identical. Comparing the outcomes of respondents from single-parent
families generated by divorce, those generated by parental death and intact families
actually creates a comparison (divorce/intact) that is more likely subject to omitted
variable bias than the standard single-parent/intact comparison. By excluding the
relatively exogenous single-parent families generated by parental death, previous
authors may have worsened the bias.