Identify the Independent Variable and the Dependent Variable This assignment requires you to review the five scholarly articles attached to this assignment and identify the following: 1)

Depression and Anxiety among Asian Americans: The Effects of Social Support and

Strain Cindy C. Sangalang and Gilbert C. Gee

It is almost taken for granted that social relationships benefit mental health, yet these rela- tionships may not always be protective. This study examines how the support and strains individuals derive from family and fiiends may be related to depression and anxiety among Asian Americans. Data come from the 2002-2003 National Latino and Asian American Study, the first nationally representative study of mental health outcomes among Asian Americans (rt = 2,066). Results indicate that farrdly support was associated with decreased odds oí Diagnostic and Statistical Manual of Mental Disorders (4th ed., text rev.) criteria for both major depressive disorder (MDD) and generalized anxiety disorder (GAD) among men and women. In addition, family strain was associated with increased odds of GAD equally among men and women. However, fiiend strain was associated with increased odds of GAD among women but not men, and faniüy strain was marginally associated with increased odds of MDD for women but was unrelated for men. The findings affirm the need to consider social strain along with social support, as well as their sources, with attention to the potentially stronger effects of strain for women. Implications for social work practice are discussed.

KEY WORDS; Asian Americans; mental health; social strain; social support; women

C onventional wisdom indicates that receipt of social support is universally beneficial. Social support from network

ties is beheved to be a fundamental aspect of psy- chological well-being and the therapeutic process (Kawachi & Berkman, 2001); however, social strain rooted in the same support networks may contribute to poorer mental health outcomes (Rook, 1992). Furthermore, research suggests support and strain may differentially affect men and women (Umberson, Chen, House, Hopkins, & Slaten, 1996). Recent research suggests that social networks and the ways individuals use these networks may also vary by culture, yet studies on minority communities, such as Asian Americans, are limited (H. S. Kim, Sherman, Ko, & Taylor, 2006).

Asian Americans are one of the fastest-growing minority groups in the United States (U.S. Census Bureau, 2011). Although prior research suggests that Asian Americans have lower rates of psychological disorder and use of mental health services than other racial and ethnic groups (Matsuoka, Breaux, & Ryujin, 1997), recent

studies indicate a substantial need for mental health care among Asian Americans (ManhaU, Schell, Elliott, Berthold, & Chun, 2005; Mui & Kang, 2006; Takeuchi et al., 2007). Shame and stignia associated with mental ulness may discour- age people from using formal mental health ser- •vices. This may be particularly true among Asian Americans, who often rely on famüy and members of their informal networks for support of their mental health problems (Leong & Lau, 2001; Spencer & Chen, 2004).

The purpose of this study is to investigate the effect of social support and strain on mental health among Asian Americans. This study has several unique features. First, in contrast to prior studies that only focus on either support or strain, we examine both factors simultaneously whüe distin- guishing between support and strain from friends and from farrdly. Second, we investigate major depressive disorder (MDD) and generalized anxiety disorder (GAD) because these are clinical outcomes that contribute to considerable disabuity (World Health Organization, 2001). Third, we explore possible gender differences in these

doi: 1O.1O93/sw/swrOO5 ® 2012 National Association of Social Workers 49

relations. Finally, we examine these issues among an understudied population and use a nationally representative sample.

BACKGROUND

Social Support, Social Strain, and Well-Being A large body of hterature has described the salu- brious effects of social support (Berkman 8c Glass, 2000; House, Landis, & Umberson, 1988). Berkman and Glass described social support as emotional, instrumental, tangible, and informa- tional resources that individuals obtain firom their network ties. These resources are known to promote mental health by enhancing self-esteem, strengthening effective coping, and mitigating dis- tress. In studies on Asian Americans, social support has been negatively associated with accul- turative stress and depression (Mui, 2001; Yeh & Inose, 2003). Social support also buffered the effect of Stressors on mental and physical health among diverse Asian American groups (Choi, 1997; Gee et al., 2006).

Other research suggests that Asian cultural values related to collectivism can limit the health- ful benefits of social support. In contrast to Western individualism, a coUectivist orientation may hinder disclosure of personal problems for fear of burdening others, disrupting group harmony, or losing face (B.S.K. Kim, Atkinson, & Umemoto, 2001; H. S. Kim et al., 2006). Indeed, some studies have found that compared with white Americans, Asian Americans appear less hkely to seek social support in response to stress (Taylor et al., 2004) or after breast-cancer treat- ment (WeUisch et al., 1999). H. S. Kim et al. (2006) found that Asian American students sought less support firom family and friends and, further, perceived support to be less helpful in deahng with stress than white students. Taken together, the conflicting findings regarding social support's effectiveness for Asian Americans demonstrate a need for further investigation.

An emerging body of research suggests that the very relationships that provide support can also engender strain. Social strain includes conflicts, feelings of undue obhgation, and demands from one's friends and family (Lincoln et al., 2010; Lincoln, 2000; Mittelmark, 1999; Rook, 1992). Because greater attention is given to positive

rather than negative components of social interac- tions, only a handful of studies have simultaneous- ly examined both social support and social strain (Lincoln, 2000).

Some studies have found that strain and support are independently associated with well-being, such that strain is associated with diminished mental health while support is associated with improved mental health (Revenson, Schiaffino, Majerovitz, & Gibofsky, 1991; Stephens, Kinney, Norris, & Ritchie, 1987). Other studies have found that social strain exerts more robust effects on psychological well-being than social support (Fiore, Becker, & Coppel, 1983; Rook, 1984). Although promising, these studies have tended to focus on specific Stressors (for example, caregiving to spouses with Alzheimer's disease), which can obscure the presence of multiple straining rela- tionships and contexts (Fiore et al., 1983; Mittelmark, 1999). Moreover, these studies used speciahzed, nonrepresentative samples, raising questions about generalizabOity to other popula- tions (Monette, Sullivan, & Dejong, 2005).

The coUectivist orientation that characterizes many Asian cultures can also trigger social strain rooted in obhgation, expectations, and norms of reciprocity (B.S.K. Kim et al., 2001). For example, Asian Americans who are family oriented may feel burdened with obhgations that confhct with American norms of individuahsm (Yee, Huang, & Lew, 1998). Children and parents of Asian American famihes may have friction over parental expectations of educational and occupa- tional achievement (Yee et al., 1998). Individuals may face pressure to fulfill unspoken expectations of reciprocity and indebtedness to friends and other network des, such as the cultural value of utang na hob (feeling of mutual obligation) among Füipino Americans (Nadal, 2009). These aspects of cultural norms that can constrain social relations may not be captured in existing studies on social support for Asian Americans.

Social Support and Social Strain from Family versus Friends Family members and friends may not present the same types of support and strain. Family relation- ships are often of longer duration and can include legal responsibüities and benefits. Studies have produced equivocal evidence for family versus friend support, with some studies showing that

50 Social Work VOLUME 57, NUMBER I JANUARY 2011

famüy support may buffer health more (Procidano & HeUer, 1983; WeUisch et al., 1999), whereas other studies have found the convene (Walen & Lachman, 2000). These mixed findings may be due, in part, to the sampling of different cultural populations, in which the bonds of family may be more or less sahent than the bonds of fdendship.

With regard to Asian Amedcans, scholan have noted that the family is often the central social unit, not the individual (H. K. Kim & McKenry, 1998). Indeed, research suggests that many Asians avoid bdnging shame to their families (B. S. K. Kim et al., 2001) and have a strong sense of family obhgation (Fuhgni, 2001). Furthermore, family is often the fint line in seeking help for mental health problems (Lin, Inui, Kleinman, & Womack, 1982). Hence, family members may be more aware of one's mental health needs, and their support may play a greater protective role than support from one's fdends. Yet, for these same reasons, the strain from famuy may be even more salient than similar stresson coming from fdends.

Social Support, Social Strain, and Gender Gender may also influence how strain and support are related to mental health. Umbenon et al. (1996) descdbed the ways in which social relation- ships may have greater demands and responsibil- ities for women than for men. One factor relates to increased exposure to stress due to role strain. Compared with men, women tend to have add- itional demands from vadous relationships, par- ticularly for those who have chüdren, are manried, or care for Ul family memben (Umbenon et al., 1996). Another factor involves differential sensitiv- ity to social strain. Several scholan have suggested that social relationships are more important for women than men and that disordered relation- ships may be particularly harmful for women (BeUe, 1982; Miller, 1988). Accordingly, this sug- gests that social strains may place a greater psycho- logical burden on women than men.

Studies indicate that social support for Asian Amedcan women is pdmarily family based (Meemeduma, 1992; WeUisch et al., 1999). WeUisch et al. found that Asian Amedcan women survivon of breast cancer relied less on social support compared to Anglo Amedcan women, in hne with previous work that suggests Asian Amedcan women prefer to provide rather than

receive care and are concerned with maintaining interpenonal harmony. Taken together, the afore- mentioned hnes of research suggest that social support may be less protective of mental illness for women than men and, further, that social strain may be more strongly related to mental Ulness for women than for men. Moreover, the effects of family support and strain should be stronger among women for these same reasons.

Based on the hterature, we examine the foUow- ing four hypotheses:

1. Social support is related to lower dsk of de- pressive disorder and generahzed anxiety disorder.

2. Social strain is related to higher dsk of de- pressive disorder and generahzed anxiety disorder.

3. Support and strain firom family wül be more strongly related to these mental health pro- blems than support and strain from fdends.

4. Gender wül moderate these associations, such that famuy or fdend support wül be more protective for women and strain wül be a greater dsk for women than men.

Because depressive disorder and generahzed anxiety disorder (CAD) often co-occur (Bakish, 1999), we anticipate that the relationship between both social support and social strain wül have par- aUel effects across both disorden.

METHOD Data and Sampling This study is based on data from the National Latino and Asian Amedcan Study (NLAAS), a psychiatdc epidemiologic survey conducted between 2002 and 2003 across the United States (Alegda et al., 2004a, 2004b; Heerenga et al., 2004). Eligible respondents were age 18 yean or older and were not in the muitary or institutiona- hzed. The current analysis is restdcted to Asian Amedcan respondents.

The NLAAS used a three-stage stratified prob- abüity sampling design. Fint, pdmary stage units of single counties or groupings of adjacent coun- ties were devised into strata on the basis of size, location, and population charactedstics. Second, within primary stage units, census blocks compds- ing area segments were stratified by geographic

SANGALANG ANO G E E / Depression and Anxiety among Asian Americans 51

location and the race/ethnicity of household respondents. High-density supplemental samples were obtained firom areas with five percent or greater of targeted ethnic groups. Sampling was performed within probabilities proportionate to census blocks for each area segment. Third, sys- tematic random sampling was used to select housing units within area segments. Within con- senting households, eligible main respondents were randomly selected, and then secondary respondents from the same household were also recruited. Interviews were conducted in the respondents' choice of Cantonese, English, Mandarin, Spanish, Tagalog, or Vietnamese using computer-assisted survey instruments. The re- sponse rate was 69.3 percent for main respondents and 73.7 percent for secondary respondents. Sample weights were developed to account for the complex sampling design and to allow esti- mates to be nationally representative (Heerenga et al., 2004).

The NLAAS is designed to be valid and reliable for diverse Asian and Latino communities. Briefly, this involved five steps: (1) identification of con- structs that are relevant to the target population; (2) collaboration among a multicultural study team that included researchers with ethnic back- grounds similar to those of the target population; (3) translations and back-translations of research materials, and by extension, the availability of the study in sbc languages; (4) use of key-informant interviews with members of the target population across several states (and in Puerto Rico) to assess materials; and (5) pilot testing and revision of the materials as needed. Further details are available elsewhere (Alegria et al., 2004a, 2004b).

Participants The Asian Ametican subsample includes 2,095 respondents. The current analysis is restricted to individuals with no missing values for variables of interest (n = 2,066). Weighted sample characteris- tics are presented in Table 1.

Measures MDD and GAD. The dependent variables were based on measures from the World Health Organization's expanded version of the Composite International Diagnostic Interview (WHO-CIDI) (World Mental Health Survey Consortium, 2004). Diagnostic outcomes were

derived from algorithms of participant responses to the WHO-CIDI that were then used to classify an individual as meeting criteria for MDD or GAD, respectively, as defined by the Diagnostic and Statistical Manual of Mental Disorders (4th ed., text rev.) (DSM-IV-TR) (American Psychiatric Association [APA], 2000). These dichotomous

Table 1: Sample Characteristics of Asian Americans in the National Latino and

Asian American Study, 2002-2003 (n = 2,066)

DSM-IV—TR major depressive disorder (past 12 months)

DSM-IV—TR generalized anxiety disorder (past 12 months)

Family support (range: 1—4)

Friend support (range: 1—4)

Family strain (range: 1—4)

Friend strain (range: 1-4)

Gender

Male

Female

Age (range: 18-95 years)

Ethnicity

Chinese

Filipino

Vietnamese

Other Asian

Education

Less than high school

High school graduate

Some college

College graduate

Work status

Employed

Unemployed

Not in the labor force

Marital status

Married

Divorced/separated/widowed

Never married

Nativity status

U.S. born

Foreign born

Region

West

Northeast

Midwest

South

4.6

1.4

47.41

52.59

28.51 21.58

12.87 37.04

14.29 17.77 25.07 42.87

63.82 6.40

29.78

68.66

8.31 23.03

23.7 76.30

68.01

15.53 8.56 7.89

2.73 2.66 1.96 1.68

41.27

0.03 0.03 0.02 0.02

0.70

Notes: Data are weighted to reflect population percentages. DSM-IV-TR = Diagnostic and Statistical Manual of Mental Disorders {4th ed., text rev.) (American Psychiatric Association, 2000).

52 Social Work VOLUME 57, NUMBER I JANUARY 2012

variables indicated the presence or absence of the disorder within the past 12 months (0 = absent, 1 = present).

Social Support among Family and Friends. Social support was measured with a brief six-item scale previously examined in the NLAAS with Latinos by Mulvaney-Day, Alegria, and Sribney (2006). Three questions asked about the firequency of support (1 = none to 4 = a lot) by three specific means: (1) talking on the phone or getting to- gether with relatives, (2) relying on relatives for help with a serious problem, and (3) relying on relatives to talk about worries. Family support was assessed by averaging across these three items (Cronbach's alpha = 0.71); higher values indicated greater social support. The same three questions were repeated, but with regard to friends, to create an analogous measure of friend support (Cronbach's alpha = 0.77).

Social Strain among Family and Friends. Two items measured family strain, or the firequency of respondents' confiicts and demands with family members. Respondents were asked two questions: (1) how often their relatives and children make too many demands, and (2) how often they argue with family. Responses ranged from 1 = not at all to 4 = often. Items were averaged to create an index in which higher scores reflected greater strain (Cronbach's alpha = 0.59). Friend strain was measured with two similar items (Cronbach's alpha = 0.60).

Control Variables. The multivariate analyses control for the following sociodemographic char- acteristics: gender, age, education, work status, marital status, ethnicity, and nativity (immigrant versus nonimmigrant). These characterisdcs have all been previously related to psychological well- being and disorder (Aneshensel, Rutter, & Lachenbruch, 1991; Dohrenwend et al., 1992; Mirowsky & Ross, 1992; Peadin & Johnson, 1977; Stronks, Van De Mheen, Van Den Bos, & Mackenbach, 1997; Takeuchi et al., 2007).

Analysis We first examined bivariate relations between variables and then proceeded with multivariate models using logisdc regression for each depend- ent variable. First, we examined the associations between family support and the dependent vari- able (MDD or GAD within the past 12 months) and between friend support and the dependent

variable, controlling for covariates. Second, we examined the associations between faniuy strain and the dependent variable and between friend strain and the dependent variable, controlling for covariates. The final model included family support, friend support, family strain, friend strain, and covariates simultaneously. We then examined the effect of gender on the associations between support and strain and dependent variables with interaction terms. These analyses centered con- tinuous variables at their mean to reduce multi- coUinearity and to facilitate interpretation (Aiken & West, 1991). Finally, significant interacdons were graphed to aid interpretation. All analyses were weighted to account for complex sampling design and to make the estimates nationally repre- sentative. We used the Stata software program, version 10.0 (StataCorp, 2007).

RESULTS

Of the respondents, 4.6 percent could be classified as having MDD and 1.4 percent as having GAD within the past 12 months (see Table 1). Respondents as a whole reported moderate levels of social support; a mean of 2.73 from family and 2.66 from friends suggests that respondents relied on these people "sometimes." A relatively lower level of social strain was reported; a mean of 1.96 from family and 1.68 from friends indicates that conflicts and demands fix)m these people occurred "rarely."

The bivariate associadons between measures of social support and strain from family and friends and the odds of meeting criteria for MDD and GAD within the last year are presented in Table 2. Unadjusted bivariate analyses showed that family strain was positively associated with

Table 2: Bivariate Odds Ratios (ORs) of 12-Month Major Depressive Disorder

(MDD) and Generalized Anxiety Disorder (GAD) (n = 2,066)

IMDDJ [GÀDl

Family support

Friend support

Family strain

Friend strain

• O R S 0.75

1.35

1.83***

1.64

K.95%tCll 0.56, 1.00

0.91, 2.00

1.42,2.37

0.94, 2.88

• P R l 0.55*

1.06

1.90*

1.76

•95yo'[Cj 0.32, 0.94

0.59, 1.89

1.10,3.28

0.94, 3.28

Note: CI = confidence interval. *p<.05. ***p<.001.

SANGALANG AND GEE / Depression and Anxiety among Asian Americans 53

higher odds of MDD. Specifically, a one-unit in- crease in family strain resulted in 1.83 greater odds of MDD (95 percent confidence interval [CI] [1.42, 2.37]). Unadjusted bivariate analyses also showed that family support was negatively asso- ciated with the odds of meeting criteria of GAD (odds ratio [OR] = 0.55; 95 percent CI [0.32, 0.94]) and family strain was positively associated with the odds of meeting criteria of GAD (OR = 1.90; 95 percent CI [1.10, 3.28]).

Associations between Support and Strain and MDD Models la to 5a in Table 3 show the associations between support and strain firom family and friends and MDD, controlling for covariates (tables omit the covariates for parsimony, but they are available from the authors). These models indicated that both famuy support and family strain were associated with MDD within the last year. In model la, family support was associated with decreased odds of MDD (OR = 0.63; 95 percent CI [0.48, 0.83]), whereas in model 3a, family strain was associated with increased odds of MDD (OR = 1.63; 95 percent CI [1.28, 2.07]). We found no significant associations between friend support and MDD or fiiend strain and MDD (models 2a and 4a). With the inclusion of all of the support and strain variables, model 5a shows a similar pattern of relations, in which greater family support was associated with reduced odds of MDD (OR = 0.63; 95 percent CI [0.48, 0.82]) and greater family strain was associated with greater odds of MDD (OR = 1.15; 95 percent CI [0.64, 2.05]).

Associations between Support and Strain and GAD Models lb to 5b in Table 3 show the associations between support and strain firom family and fiiends and GAD, controlling for covariates. Model lb indicated that increased family support was protective against GAD (OR = 0.44; 95 percent CI [0.26, 0.74]). In model 3b, a one-unit increase in family strain was associated with 1.78 higher odds of meeting criteria for GAD (95 percent CI [1.04, 3.05]). Models 2a and 4a show that neither fiiend support nor friend strain was associated with GAD. In model 5b, family support and strain variables exerted similar effects on GAD with the inclusion of all independent

9

^H <

M w

'S* m

1

W Es

Q

a

1̂ •̂~

9; 9̂ 1

^̂ 1̂1

13 in 01

• 0 0

s

d el

<

0 g

o

el 3

1

1 S

_̂ (N

o d

el

S

•3 • 0 0 S

IS m 'S 0

s

S

ie

0

m m

10

IS

de l

0

S

IS

o d

el

g

G

m01

K 0

G

95 °/

ce 0

G

Sf

ce 0

G -5 m 0)

5% C

l O

R

01

ce 0

G

m01

ce 0

G

Si 01

S G

Si 01

ce 0

G

m 01

ce 0

G

in O)

ce 0

£ .2

oi^ m— o o " ^ " ^ — lA Off)

o — "

"O

I (N — CN >o u~ '«r i n ' 0 0 *r -^ (N — "O o ; Ö d -^ -¡ (N d (N

Ö — -^

• = = 1 =

I Q.

il Ï V

ê?-

54 Social Work VOLUME 57, NUMBER I JANUARY 2012

variables. Specifically, family support was nega- tively associated with meeting criteria for GAD (OR = 0.42; 95 percent CI [0.27, 0.65]). Further, in the presence of famüy support, family strain continued to be related to higher odds of meeting criteria for GAD (OR =1.69; 95 percent CI [1.13,2.51]).

The Moderating Effect of Gender Our analyses suggested that gender moderated strain. Specifically, we found a significant inter- action between gender and friend strain with GAD {p < .05) and a marginally significant inter- action between gender and family strain with MDD (p = .06). The interaction between friend strain and gender with GAD, wherein friend strain was associated with increased log odds of GAD among women but not among men is illu- strated in Figure 1. Among men, friend strain was associated with decreased log odds of GAD. In a similar manner. Figure 2 presents the interaction between family strain and gender with MDD. For both genders, family strain was associated with increased log odds of MDD, but this association appears attenuated among men compared with women. Finally, gender did not moderate the effect of support from family or friends on MDD or GAD.

Figure 1: Gender Differences in the Influence of Friend Strain on Generalized

Anxiety Disorder (GAD)

0-

– 1 –

-2

§ – 3 (9

; -4

§ s -s

– 6 •

-7

– 8 –

—

1 2 3 Strain from Friends

MEN

WOMEN

4

Figure 2: Gender Differences in the Influence of Family Strain on Major

Depressive Disorder (MDD)

0-

-o.s-

– 1 –

-3-

-3.5-

-4-

•

•

1

'

2 3 Strain from Famiiy

MEN

WOMEN

.—

4

DISCUSSION The aims of this study were to examine the main effects of social support and social strain firom family and fiiends on mental health variables (MDD and GAD) and to investigate possible gender differences in these relationships among a nationally representa- tive sample of Asian Americans.

Our data showed that support from famüy, but not friends, was associated with decreased odds of MDD and GAD, after controlling for sociodemo- graphic characteristics. Many studies combine support fi'om family and friends (for example, Noh & Kaspar, 2003; Taylor et al., 2004; Yang & Clum, 1995). Our study indicates that it is im- portant to consider who is providing this support.

Furthermore, strain from family, but not friends, was related to increased odds of MDD and GAD. These results are aligned with other studies that demonstrate greater levels of social strain are hnked to psychological distress and dis- order (Revenson et al., 1991; Walen 6¿ Lachman, 2000). In particular, measures of familial strain related to confiicts and demands such as those reported presently have also been captured in earher studies on Asian Americans that examined issues such as intergenerational confHct and care- giver burden (Adams, Aranda, Kemp, & Takagi, 2002; Ying& Han, 2007).

SANGALANG AND G E E / Depression and Anxiety among Asian Americans 55

Moreover, this finding demo