In this paper we use longitudinal data to rest the strength of individual p
references and structural variables as explanations for married women's lab
or force participation. Data drawn from a subset of the Career Development
Study are used to compare gendered preferences measured toward the end of a
dolescence vs. work and family structural variables as predictors of the ac
tual number of hours married women work for pay. Family structures that pus
h women out of the labor force and pull them into family work prove to be t
he strongest predictor of married women's employment hours, with work struc
tures (e.g., aspects of "good" jobs) and the subjective definition of paid
work as a career also being substantively important for explaining hours in
the labor force, Our findings also indicate that attitudes formed before a
nd during early adolescence do have a weak but statistically significant ef
fect on married women's labor force participation, at least for baby boom w
omen.