208 / 126
208
E
UR
A
MERICA
is true for both Filipino cohabiting and marital unions and is
stronger in the latter form of relationships. Whether this pattern is
due to geographic proximity (e.g., more Filipinos live in major
immigrant destinations) or structural barriers would be an
important topic for future research to address.
A number of study limitations should be noted. First of all, it
would be more ideal if discrete-time event history analyses were
used. However, data requirements are also more intense for event
history analyses than regular logistic regression. That is, any
variable that could be time-varying should have complete
yearly/monthly records. For the current study, family structure
should be time-varying, but it is not possible to know whether a
respondent went through parental divorce or remarriage or
re-partnering between Waves 1 and 4 (a time span of 12-14 years).
In turn, a strong assumption of
no change
in family structure would
need to be posed if event history analyses are to be used. This is not
appropriate and hence logistic regression models were used instead.
Secondly, finer ethnic categories could not be used due to the
limited number of Asian American youths in the Add Health data.
It is not ideal to group youths of Asian Indian and other
South/Southeast Asian origins together, since the former often have
parents with better socioeconomic profiles than the latter. This is
largely due to the fact that the majority of post-1965 immigration
from India is composed of students seeking advanced scientific
studies in the U.S. (Sheth, 1995), while most of the South and
Southeast Asian Americans entered the U.S. through the refugee
resettlement program (Rumbaut, 1995). Lastly, substantial sample
attrition took place between Waves 1 and 4 of Add Health that
could potentially lead to some bias in estimations, which is a
common data issue among longitudinal surveys.
In conclusion, this paper adds to the current literature on
ethnic variations in union formation among Asian Americans in two
ways. First, the data structure of Add Health offers a unique
opportunity to investigate whether a first marriage is preceded by