We selected Asian American data from the National Latino and Asian American Survey, which used a multi-frame, stratified probability sampling scheme. The scheme has been described in detail elsewhere (Heeringa et al., 2004
), but a summary follows. Samples were drawn using three methods. In the first, participants were recruited with a multistage stratified area probability sampling design: (a) city or contiguous census blocks were sampled based on population density in each neighborhood; (b) dwelling units were sampled within each block; (c) one adult was sampled within each selected dwelling unit. In the second method, census blocks with at least 5% of Asian households were over-sampled. In the third method, to increase the sample size, a second respondent from a previously sampled household was recruited. Weighting corrections were constructed to control for differences in selection probability.
Among the surveyed Asian Americans (N = 2095), 454 were US-born, 1639 were foreign-born, and 2 did not identify a place of birth. Interviews were offered in English, Mandarin, Cantonese, Tagalog, and Vietnamese. The three largest ethnic groups represented were Chinese (32%), Filipino (20%), and Vietnamese (16%). Participants self-identified with national origins in the regions of East Asia, Southeast Asia, South Asia, and Central Asia.
This analysis only included data from foreign-born Asian American participants aged 25 years and older (N
= 1451). Age at immigration, dichotomized at age 25 years, was tested as a potential moderator of the relationship between subjective social status and mental health. The age of 25 years was used as a psychosocial marker to distinguish between immigrants who experienced disruption during formative years (i.e., childhood, adolescence, early adulthood) versus later-adult years (i.e., middle- and late-adulthood). Recent psychological evidence suggests that social and cognitive development does not reach maturation until roughly the age of 25 years (Giedd, 2004
; Gogtay et al., 2004
). We theorize that the influence of age at immigration on the relationship between subjective social status and mental health is a psychosocial, as opposed to a biological, phenomenon. Age at immigration may capture a developmental context that includes influences of the family, US peers, and institutional practices (i.e., formal education). Therefore, age at immigration was treated as a categorical rather than continuous variable. This practice is consistent with related research on generations and immigrant health (Rumbaut, 2004
; Takeuchi, Hong et al., 2007
Our dependent variable was the presence of mood dysfunction in the 12 months prior to the interview (12-month period prevalence). Mood dysfunction was a composite formed by the presence of at least one clinical or sub-clinical symptom of anxiety or affective disorder, as measured by the Composite International Diagnostic Interview (CIDI; World Health Organization). The CIDI is the most widely used structured diagnostic interview and was designed to be used across cultures. A composite variable was used because separate analyses of anxiety (e.g., generalized anxiety disorder, agoraphobia, social phobia, post-traumatic stress disorder, and panic disorder) and affective disorders (e.g., dysthymia and major depressive episodes) produced similar results.
Subjective social status
Our independent variable was subjective social status
, as measured by a symbolic ladder with 10 rungs, where the first and tenth rung represent the lowest and highest social status, respectively (Adler et al., 2000
; Cantril, 1965
). Two dimensions of status were assessed. In one, subjective social status, respondents were asked to, “Think of this ladder as representing where people stand in the US
. At the top of the ladder are the people who are the best off, those who have the most money, most education, and best jobs. At the bottom are the people who are the worst off, those who have the least money, least education, and worst jobs or no job. What is the number to the right of the rung where you think you stand at this time in your life, relative to other people in the United States?”
In the second measure, csubjective social status, respondents were asked to, “Think of this ladder as representing where people stand in their communities. People define community in different ways; please de fine it in whatever way is most meaningful for you. At the top of the ladder are the people who have the highest standing in their community. At the bottom are the people who have the lowest standing in their community. What is the number to the right of the rung where you think you stand at this time in your life, relative to other people in your community?”
Two conventional measures of SES were analyzed. Educational attainment was measured by having respondents indicate the number of years of schooling they had completed; it was coded into several dummy variables in the analyses representing meaningful educational milestones. These dummy variables represent receipt of less than a high school education (0–11 years); high school graduate (12 years); some college (13–15 years); and college graduate and beyond (16 or more years). In analyses, we treat 16 or more years of education, the largest group, as the reference category. Household income is the sum of the midpoints of the following income measures: personal, spouse, other family members, social security, government assistance, and other sources. Because of a large number of missing values (270 missing), this variable was imputed using hot deck methods based on the variables of ethnicity, sex, age, education, household composition, and employment status. We divided household income by 1000 and collapsed it into four categories that we represent as a series of meaningful dichotomies, using $80,000 or more as the reference group. We control for the family size in order to make household income interpretable at the individual level.
Gender, ethnicity, US citizenship, marital status, English language proficiency, current age, and age at immigration were also examined. Ethnicity was categorized into one of four groups (Vietnamese, Filipino, Chinese, and Other Asian), where Chinese was the reference group as the largest ethnicity represented. US citizenship was dichotomized non-citizen versus citizen (reference group). Marital status was dichotomized married versus non-married (reference group). English language proficiency is a scale which asks the respondent to rank his or her ability to speak, read and write in English. For these three measures, response categories range from (1) poor to (4) excellent, yielding minimum and maximum scores from 3 to 12.
Current age and age at immigration were single-item measurements. Current age was divided by 10 in the multivariate regressions and descriptive statistics for easier interpretation, but still treated as a continuous variable. Age at immigration was dichotomized at age 25 years, and its influence on mood dysfunction was measured using a dummy variable for participants who moved to the US before the age of 25 years.
Our main analysis consists of a series of nested multivariate survey logistic regression models which assess the net effect of adult subjective social status on the presence of mood dysfunction. All of the analyses adjust for the hierarchical nature of the multistage survey data using SAS-callable SUDAAN procedures. The SUDAAN procedures allow for the incorporation of complex survey sampling methods, including designs with stratification, clustering, and unequal sampling weights, in the point and standard error (SE) estimation.
provides descriptive statistics for the total sample, and stratifies the sample by age at immigration. t-Tests were used to compare the weighted means for all variables in the age at immigration.
Weighted means for all the variables in the analyses, total sample and stratified by age at immigration
reports the results from a series of survey logistic regressions which focus on the relationship of subjective social status to the outcome of mood dysfunction. We report unstandardized maximum likelihood coefficients and their significance levels. We perform this analysis in six steps. Model 1 tests the unadjusted bivariate relationship of subjective social status with mood dysfunction. Models 2 and 3 sequentially add traditional SES and demographic measures to determine if the effects of subjective social status on mood dysfunction remain after their inclusion in the analyses. In Model 4, we test the direct effect of age at immigration on mood dysfunction. In Models 5 and 6, we assess whether the effects of subjective social status on mood dysfunction depend on current age or age at immigration by incorporating interaction terms; given the high correlation between current age and age at immigration, each variable was tested separately as a potential moderator.
The effects of subjective social status on mood dysfunction using survey logistic regression (N = 1451)
In this approach, outlined by Baron and Kenny (1986)
, current age
and age at immigration
are determined to moderate the relationship between subjective social status and mood dysfunction if there is a significant interaction. The analyses presented in replicate those presented in , but substituting csubjective social status for subjective social status in all analyses.
The effects of csubjective social status on mood dysfunction using survey logistic regression (N = 1451)