attitudes. Our independent variable of interest is the
share of an individual’s siblings who are sisters.3 Even
assuming that child gender is randomly assigned,
a major concern when analyzing the effect of sibling
gender is endogenous fertility choice. Parents may
choose stopping rules—whether to have additional
children—that depend on the gender of the children
they currently have so that an older sibling’s gender is
not randomly assigned. For example, parents who have
a preference for boys may continue to have children
when they have daughters, but not when they have
sons. Therefore, children with older brothers may tend
to be part of different kinds of families than children
who have older sisters.4 This means that the share of an
individual’s siblings who are sisters is endogenous since
parents’ choices about whether to have additional
children determine the overall share of female children.
On the other hand, conditional on parents choosing to
have an additional child, the gender of the next sibling is
assigned at random.5 We can use this random variable
to identify the effect of growing up with a greater
proportion of sisters.