Although experiences of racial and ethnic discrimination have been found to be prevalent among Asians living in the United States (Cheryan & Monin, 2005
; Gee, 2002
; Goto, Gee, & Takeuchi, 2002
; Rumbaut, 1994
), the research on the psychological ramifications of these experiences remains scarce. Using the first-ever nationally representative sample of Asian adults in the United States, we can examine systematically how ethnic identity influences the association between discrimination and distress across age and nativity status. This goal is achieved in three steps. First, the current study examines differences in reports of discrimination and ethnic identity by age and nativity. Second, we examine the association between discrimination and mental health while taking into account respondents' age and nativity. It is expected that across the sample, discrimination will be associated with increased distress; however, for older individuals there may be an even stronger association between discrimination and outcomes due to the accumulation of life events over time. Further, it is hypothesized that individuals born in the United States will exhibit a weaker association between discrimination and health because being raised in the United States has likely included experiences that prepared them for experiences of discrimination. Third, in order to examine the role of ethnic identity in the association between discrimination and psychological outcomes, we will explore evidence for the buffering and centrality hypotheses while stratifying the sample by age and nativity. It is expected that during developmental periods when one's life is relatively stable, such as in midlife, centrality may be a buffer for discrimination. In contrast, at the youngest and oldest ages of our sample, when one's life circumstances are more likely to be changing, centrality may be more likely to exacerbate discrimination.
The data come from the Asian respondents in the National Latino and Asian American Study, a household survey conducted between 2002 and 2003. In order to obtain a nationally representative sample, sampling strategies included three components. First, core sampling of primary sampling units (metropolitan statistical areas and counties) and secondary sampling units (from continuous groupings of Census blocks) were selected using probability proportion to size, from which housing units and household numbers were sampled. Next, high density supplemental sampling of census-block groups with greater than 5% density of target ethnic groups was included. Finally, second respondent sampling to recruit participants from households from which a primary respondent had already been interviewed was employed. Weights were developed to account for the joint probabilities of selection for these three components of the sampling design, allowing the sample estimates to be nationally representative. Specifically, the post-stratification weights adjust for the age and gender distribution of Asians living in the United States to match the Census.
Respondents were Asian adults ranging from 18 to 95 years of age and residing in the United States (N
= 2,095). In order to create age groupings that were consistent with other life span research (Mroczek & Kolarz, 1998
), the current study examines only participants 18 to 75 years of age (N
= 2,047; ). The sample includes 586 Chinese, 491 Filipino, 508 Vietnamese, and 462 others (140 South Asians, 105 Japanese, 81 Koreans, 39 Pacific Islanders, and 97 others). Approximately 43% of the sample reported graduating from high school, and another 42% completed college or had an advanced degree, and the remaining 15% reported not completing high school. The majority of the sample (64%) was employed, but 18% reported living in poverty.
Sample Characteristics and Descriptive Statistics
Interviews were conducted using computer-assisted software and administered by trained interviewers with linguistic and cultural backgrounds similar to those of the respondents. Interviews were conducted in the respondent's choice of English, Cantonese, Mandarin, Tagalog, or Vietnamese. All instruments were translated using the standard techniques of translation and back-translation. The median length of the interview was 2.4 hr. As a measure of quality control, a random sample of participants with completed interviews was recontacted to validate the data. The overall response rate was 66%. Participants were compensated for their participation. Missing data for the variables reported here varied between .05 and .08%.
The Kessler Psychological Distress Scale (K10) measured distress (Furukawa, Kessler, Slade, & Andrews, 2003
). This 10-item inventory assesses the prevalence of negative feelings in the past 30 days (e.g., “how often did ….you feel depressed?, …hopeless?”). Respondents reported frequency on a 5-point scale (1 = none of the time
, 2 = a little of the time
, 3 = some of the time
, 4 = most of the time
, and 5 = all of the time
) and responses were summed (range = 10 to 44; α = .83; ). Because this variable was positively skewed, skewness = 2.37 (.05), kurtosis = 7.10 (.11), it was log transformed (range = .00 to .63, M
= .09, SD
Perceptions of Racial and Ethnic Discrimination
Three items were averaged to assess the frequency of discrimination: “How often do people dislike you because you are (self-described ethnic/racial group)?,” “How often do people treat you unfairly because you are…?,” and “How often have you seen friends treated unfairly because they are…?” (Vega, Zimmerman, Gil, Warheit, & Apospori, 1993
) on a 4-point scale (1 = never
, 2 = rarely
, 3 = sometimes
, 4 = often
; α = .83; ).
A single item, “How close do you feel, in your ideas and feelings about things, to other people of the same racial and ethnic descent?,” measured ethnic identity centrality. Respondents indicated their agreement on a 4-point scale (1 = not at all, 2 = not very close, 3 = somewhat close, and 4 = very close; ).
Consistent with life span research involving national samples (Mroczek & Kolarz, 1998
), age was categorized into four groups by decades: below 30 years old (n
= 575), 31 to 40 years old (n
= 516), 41 to 50 years old (n
= 476) and 51 to 75 years old (n
= 490). These groups approximate age groupings described as “young adults,” “adults,” “middle-aged adults,” and “older adults” (Diehl, Coyle, & Labouvie-Vief, 1996
Nativity was measured with a single item indicating whether the respondent was foreign born or U.S. born.
Demographic and Control Variables Socioeconomic status
Socioeconomic status has been observed to be related to reports of discrimination and distress (Clark, Anderson, Clark, & Williams, 1999
) and thus, an important covariate. Household income was derived from seven questions tapping sources of income (respondent, spouse, social security, government, family, and other), which were summed. Income was categorized by quartile, 1 = $0–10,999
(25.9%), 2 = $11,000–39,999
(27.2%), 3 = $40,000–74,999
(22.3%), 4 = $75,000
Age of immigration
Age of immigration has been found to be predictive of the prevalence of mental disorders (Takeuchi, Hong, Gile, & Alegria, 2007
). As such, “How old were you when you first came to this country?” was included as a control for analyses conducted with individuals born outside of the United States (range = 0–70 years old).
Studies suggest that men often report more racial discrimination than women, although women often report more mental health problems than men (Kuo, 1995
). Gender was included as a control variable (0 = male
; 1 = female
Asian ethnic groups in the United States vary in reports of discrimination (Gee et al., 2007
; Kuo, 1995
). Each of the four ethnic groups (i.e., Chinese, Filipino, Vietnamese, and other) was dummy coded.
Experiences of discrimination (Gee, 2002
), ethnic identity (Umana-Taylor & Shin, 2007
), and psychological well-being (Markus, Plaut, & Lachman, 2004
) vary according to geographic location. Four regions were included as a control variable, Northeast (n
= 152), Midwest (n
= 91), West (n
= 1707), and South (n
Research among Asians in North America finds that support from family and same-ethnicity others buffers the effects of discrimination (Noh & Kaspar, 2003
). To control for this influence, a 10-item scale assessing respondents' perception of family closeness was included (Olson, 1986
). Respondents rated their agreement (1 = strongly disagree
, 2 = somewhat disagree
, 3 = somewhat agree
, and 4 = strongly agree
) with items such as “Family members feel very close to each other” and a mean was estimated (α = .93).
The bivariate unweighted correlations between discrimination, ethnic identity, distress, family cohesion, and social desirability were estimated using SPSS (). There was a significant correlation between all of the variables with the exception of discrimination and ethnic identity. Discrimination was positively correlated with distress, such that more reports of discrimination were associated with more distress. Ethnic identity was found to have a negative correlation with distress, stronger ethnic identity being associated with less distress.
Correlation among Study Variables
For the remaining analyses, the statistical software, Stata 9.2 (StataCorp, 2005
), was employed to adjust for complex survey design weights. The weighted and unweighted means are shown in . Before proceeding with the primary analyses, differences in distress, discrimination, and ethnic identity by age and nativity were examined. Analyses of variance were conducted and found no differences in reports of distress by age group or nativity status (). The omnibus test for age differences in discrimination was significant with post hoc contrasts indicating that individuals 41 to 50 years of age reported significantly more discrimination than individuals over 51 years of age. Nativity was also associated with discrimination, such that individuals born internationally reported more discrimination than those born in the United States. As well, there were differences in reports of ethnic identity and family cohesion by age and nativity status. Post hoc contrasts revealed that individuals below 30 years of age reported significantly lower ethnic identity and family cohesion than the other three groups. In addition, individuals born outside the United States reported significantly stronger ethnic centrality and family cohesion than their United States–born counterparts. There were also age differences in reports of social desirability, the youngest age group reporting significantly lower levels than individuals over age 41. Immigrants scored higher on social desirability than U.S.-born individuals.
Comparison of Means Across Age Groups and Nativity Status (Weighted)
Analyses of variance were also conducted to examine the means for key variables, taking both nativity and age into account. Among individuals born overseas (), there was a significant difference in reports of discrimination by age, such that individuals 41 to 50 years of age reported significantly more than individuals over age 51. Differences in ethnic identity were also observed, those below age 30 reporting weaker identification than the two older age groups. Family cohesion also differed by age, such that the youngest group reported significantly lower cohesion compared to the other three age groups. Finally, there were also significant differences in reports of social desirability, the two younger groups reporting lower levels than the older two groups. Among United States–born individuals (), there was a significant difference in distress by age group, such that individuals 41 to 50 years of age reported less distress than individuals over 51 years of age. No differences were reported for discrimination, ethnic identity, or family cohesion; however, the oldest age group reported significantly higher levels of social desirability than the other three age groups.
Means by Age for Immigrants (Weighted)
Means by Age for U.S.-Born Individuals (Weighted)
Because age and nativity are seen to represent different developmental contexts, regression analyses were first conducted examining the effect of nativity status separately. Next, regression analyses were stratified by both variables (4 age groups × 2 immigration statuses = 8 analyses). With psychological distress as the outcome, the main effects of discrimination and ethnic identity as well as their interaction were included as predictors. In addition, socio-economic status, gender, ethnicity, region, family cohesion, and social desirability were included. Age of immigration was included for analyses involving immigrant individuals. Before examining the effects of nativity status, the association between discrimination and ethnic identity and their interaction was estimated in the full sample. There was a main effect of discrimination on distress, such that more frequent reports of discrimination were associated with increased distress (b = .02, SE = .01, p < .01).
Examining just the subsample of immigrants, similar results were observed, such that more frequent reports of discrimination were associated with increased reports of psychological distress (). To explore differences by age group, four regressions were conducted among the immigrant subsample (). Although the coefficients for the main effect of discrimination differ slightly across the age groups, the four models were found to be statistically equivalent, suggesting no differences between the groups (Chow, 1960
). Moreover, ethnic identity did not moderate the effects of discrimination and distress in any of the four age groups.
Regressions of Distress by Nativity Status (Weighted)
Regressions of Distress by Age, in Years, for Immigrants (Weighted)
To examine possible age differences among individuals born in the United States, another four regressions were conducted (). For individuals below 30 years of age, only a main effect of discrimination was observed, such that increased discrimination was associated with increased distress. For individuals between 31 and 40 years of age, a main effect of discrimination as well as an interaction of discrimination and identity were observed. To explore the nature of this interaction, tests of simple effects were conducted (Aiken & West, 1991
). Consistent with the centrality hypothesis, individuals with strong (1 SD
above the mean) and moderate levels of centrality reported a significant positive association between discrimination and distress (b
= .05, SE
= .02, p
< .05) and the slopes were significantly different from the low centrality group (). For individuals between 41 and 50 years of age, there was also an interaction between discrimination and identity; however this interaction was consistent with the buffering hypothesis (). Specifically, individuals with strong ethnic identity reported lower distress when faced with discrimination, whereas individuals with weaker ethnic identity reported the opposite pattern. Tests of simple slopes found that the high and low centrality groups differed from one another (b
= −.06, SE
= .03, p
< .05). Finally, consistent with the centrality hypothesis, a significant interaction was also observed for those over 51 years of age (). The slopes for the groups were significantly different (b
= .04, SE
= .02, p
< .05), and individuals with high centrality reported more distress when they reported discrimination compared to those with low centrality.
Regressions of Distress by Age, in Years, for U.S.-Born Individuals (Weighted)
U.S.-born 31- to 40-year-olds: The moderating effects of ethnic identity on the association between discrimination and distress.
U.S.-born 41- to 50-year-olds: The moderating effects of ethnic identity on the association between discrimination and distress.
U.S.-born 51- to 75-year-olds: The moderating effects of ethnic identity on the association between discrimination and distress.
In sum, we find that discrimination was associated with increased distress for the entire sample. Moreover, among individuals born in the United States, the effect of discrimination varied by age and ethnic identity. Specifically, among individuals in their 30s and over 51, consistent with the centrality hypothesis, discrimination was exacerbated by ethnic identity. However, for individuals in their 40s, the effects of discrimination were buffered by ethnic identity.