Data Source
Analyses were based on wave 1 of the
2000–2001 Los Angeles Family and Neighborhood
Survey (LAFANS-1). LAFANS-1 wasa survey of adults, children, and neighborhoods
in a stratified probability sample of
census tracts in Los Angeles County. The
1652 census tracts in Los Angeles County
were divided into very poor, poor, and nonpoor
strata based on the percentage of the
population living in poverty in each census
tract. A total of 65 tracts were sampled: 20
each from the very poor and poor strata and
25 from the nonpoor stratum. Within each
sampled tract, 40–50 dwelling units were selected
at random, with an oversampling of
households with children. Within each household,
LAFANS-1 randomly selected 1 adult
(aged 18 years and older) for interview. Interviews
were conducted in English and Spanish.
Two thousand six hundred twenty-three adult
respondents were interviewed. Our analysis
was limited to adult respondents younger
than 65 years—the age of eligibility for
Medicare. Twenty-three hundred respondents
had health insurance information and were
under 65 years of age. The analysis sample
size was reduced to 2130 after we excluded
respondents with missing information on the
independent variables.
More than half of the LAFANS-1 sample
was Latino (principally of Mexican origin),
and the sample included sizable numbers of
first- and second-generation immigrants (Latino,
White, and Asian) as well as nonimmigrants.
For more details, see Sastry et al
Analyses
Statistical analyses were done in Stata.25
Multivariate logit models predicting gaining
and losing insurance were used to obtain relative
risks adjusted for socioeconomic characteristics.
Variables controlling for the oversampling
of poor households and households
with children and variables related to nonresponse
were included in the model.26