For the first set of analyses that follow, we use the fourwave
PSP. The four waves were conducted in 1965,
1973, 1982, and 1997. In the first wave, most subjects
were in their senior year of high school; 98.5% of the
sample was between 17 and 19 years old. By 1997, most
of the respondents were about 50 years old. The datasethas detailed family information, including the gender
and age of a respondent’s siblings. These data thus
provide the share of a respondent’s siblings who are
sisters and the quasi-random indicator for the younger
sibling being a sister. Since our empirical strategy is
based on the random assignment of younger sibling
gender, our estimation sample consists of survey
respondents who had at least one younger sibling.
We consider the effect of having sisters on people’s
preferences on gender roles and their partisanship. The
phrasing of the gender-role question in 1973 was:
‘‘Recently, there has been a lot of talk about women’s
rights. Some people feel that women should have an
equal role with men in running business, industry, and
government. Others feel that women’s place is in the
home. Where would you place yourself on this scale or
haven’t you thought much about this?’’10
Response options were represented by a scale
ranging from 1 (‘‘Women and men should have an
equal role’’) to 7 (‘‘Women’s place is in the home’’).
Party identification was asked in the standard way
following the question wording used in the American
National Election Study (ANES). See online Appendix
1 for all question wordings and response options.
During the first two waves, data were also collected
on a sample of the subjects’ parents. We use these data
to conduct randomization checks to confirm that
having a younger sister (as compared to having a
younger brother) does not predict pretreatment baseline
characteristics for the PSP sample, as we would
predict if younger sibling gender is randomly assigned.
11 These variables indeed are very similar for
men (and women) with younger brothers compared to
those with younger sisters (see online Appendix 2).12
We chose variables that were entirely exogenous —
either characteristics of the parents’ own environments
when they were children or other characteristics
that could not be affected in any way by child
gender. In these checks, as with all the regressions
using these data, we cluster the standard errors at
the primary sampling unit (PSU) level.
To evaluate the robustness of the results from
the PSP, we also analyze a completely different
dataset: the National Longitudinal Survey of
Youth’s Children and Young Adults sample
(NLSY79-YA). The NLSY79-YA interviewed the
children of the women of the original NLSY79
survey starting in 1986. Questions about genderrole
attitudes and political partisanship were asked
in the 2006 and 2008 waves, respectively, when
respondents were between 21 and 38 years of age.
Most of the respondents were on the young side of
this range, with the median age for both female and
male respondents being 22 in 2006. Most importantly,
respondents were asked similar questions
about partisanship and the role of women as were
asked in the PSP. As a result, we can assess whether
the main results from the PSP replicate using an
entirely new sample surveyed in recent years. To
measure views on gender roles, respondents were
presented with the statement ‘‘A woman’s place is
in the home, not the office or shop’’ and were
asked if they strongly disagree, disagree, agree, or
strongly agree with it. In all analyses using the
NLSY, we cluster standard errors by mother since
sometimes more than one child in a household was
interviewed.
Unlike the PSP, the NLSY also contains data on
children’s experiences in their early household
environment. Starting at the age of 10, the same
children who were later asked about their partisanship
and political attitudes answered questions
about the types of chores they did. As a result,
we can use the NLSY to test for differences in the
household assignment of tasks according to respondent
gender, as well as differences according to
sibling gender.