Adaptation and Psychometric Evaluation of the Multidimensional Scale of Perceived Social Support for Arab Immigrant Women

Abstract

We adapted the Multidimensional Scale of Perceived Social Support (MSPSS) for use with Arab immigrant women (MSPSS-AW) and estimated the psychometric properties of the adapted version with a sample of 539 Arab immigrant women living in the United States. Confirmatory factor analysis (CFA) supported the proposed three-factor solution. Internal consistency reliability coefficients for the three subscales ranged from good to very good. Additional evidence for construct validity of the MSPSS-AW subscales was demonstrated through relationships with theoretically related measures. We conclude that the MSPSS-AW is reliable and valid for use with Arab immigrant women.

View correction statement:

Erratum

Women are a large and growing component of international migration, representing approximately half the world's estimated 191 million immigrants (United Nations, 2008). Social support is an important resource for these women (Aroian, Norris, González de Chávez Fernández, & Averasturi, 2008; Hiott, Grzywacz, Arcury, & Quandt, 2006; Remennick, 2005). Investigating social support for immigrant women is hampered, however, by the availability of culturally sensitive measures of social support. We adapted and evaluated a measure of social support, the MSPSS (Zimet, Dahlem, Zimet, & Farley, 1988), for use with Arab Muslim immigrant women to the United States. This adapted version is hereafter referred to as the MSPSS for Arab women (MSPSS-AW). Specific study aims were to establish the factor structure and concurrent validity of the MSPSS-AW with theoretically related measures of emotional distress and coping.

According to two extensive reviews about the salutary functions of social support, the vast majority of studies testify to the benefits of social support (Cohen, Underwood, & Gottlieb, 2000; Taylor, 2007). Theoretical explanations for the relationship between social support and psychological health posit that social support provides a generalized source of positive affect and security, a sense of purpose and belonging, and information from well-meaning others that can help people avoid or minimize stressful or high-risk situations (Cohen et al., 2000; Taylor, 2007). Stress buffering models of social support include the notion that social networks provide the resources to bolster coping, either by increasing one's perceived ability to cope and dampening affective reactions to stressful life events or by providing more tangible kinds of coping assistance to solve problems (Lepore, Silver, Wortman, & Wayment, 1996; Thoits, 1986, 1995). Social support also is linked closely with research and theory on stress and coping (Cohen et al., 2000; Lazarus & Folkman, 1984). In fact, coping by seeking social support is considered one of the more productive coping strategies. Social support also can lead to other constructive kinds of coping strategies, such as problem solving, and can prevent maladaptive coping, such as self-blame and avoidance. Avoidance, for example, is maladaptive if it makes a situation worse or leads to a reoccurrence of the problem. (See Skinner, Edge, Altman, and Sherwood (2003) for a comprehensive review of coping strategies and health outcomes.)

Immigrant women are at risk for social support deficits. They leave extended family and close friends behind in the homeland, and their husbands are not likely support substitutes, particularly if husbands are struggling to cope with their own immigration difficulties (Aroian, 1992; Aroian, Spitzer, & Bell, 1996; Llacer, Zunzunegui, del Amo, Mazarrasa, & Bolumar, 2007; Menjivar, 1995; Simich, Beiser, & Mawani, 2003; Waters, 1997). Social support deficits may be even greater for immigrant women from traditional cultures who do not work outside of the home and lack work-related social networks from which to draw support. These risks, however, may not be longstanding. New friends eventually become “fictive kin,” providing support similar to what used to be provided by extended family members (Ebaugh & Curry, 2000). The marital relationship also may become closer and more supportive over time. The MSPSS (Zimet et al., 1988) shows promise as a measure for use to answer these kinds of research questions.

The Multidimensional Scale of Perceived Social Support

The MSPSS is a 12-item measure of the adequacy of social support from three specific sources: family, friends, and significant other. Items measuring support from a significant other refer to a “special person,” which may be interpreted variously to mean a romantic or other particularly close relationship. The Family, Friends, and Significant Other subscales each have four items, all rated on a 7-point scale (very strongly disagree [1] to very strongly agree [7]). Higher subscale scores indicate greater perceived adequacy of social support from each of the three respective sources of support. Reading level of comprehension is the third grade.

There is ample reliability and validity evidence supporting the three-scale factor structure of the MSPSS, including extensive psychometric data from adults, adolescents, and undergraduates as well as psychiatric in- and outpatients (Canty-Mitchell & Zimet, 2000; Cecil, Stanley, Carrion, & Swann, 1995; Clara, Cox, Enns, Murray, & Torgrudc, 2003; Dahlem, Zimet, & Walker, 1991; Edwards, 2004; Eker & Arkar, 1995; Kazarian & McCabe, 1991; Ramaswamy, Aroian, & Templin, 2009; Zimet, Powell, Farley, Werkman, & Berkoff, 1990). Some of these psychometric evaluation studies also include different racial and ethnic groups, specifically Muslim adults in Turkey (Eker & Arkar, 1995); Mexican American adolescents (Edwards, 2004); urban, primarily African American adolescents (Canty-Mitchell & Zimet, 2000); and Arab American adolescents (Ramaswamy et al., 2009). The psychometric evaluation with Arab American adolescents, however, was with a slightly different version of the MSPSS, which included assessing support from school personnel.

The MSPSS also has been used with Chinese immigrant mothers and their children (Short & Johnston, 1997), adults in Italy (Prezza & Pacilli, 2002), and adult Russian immigrants (Ponizovsky & Ritsner, 2004) and native-born Jewish and Palestinian undergraduates (Ben-Ari & Gil, 2004) in Israel. These were not psychometric evaluation studies, however, and psychometric information was limited to reports of high internal consistency reliability.

Although the MSPSS appears to be robust with populations that are different from the U.S. mainstream population from which it was developed, more evaluation is needed with specific ethnic groups (Kim, Sherman, & Taylor, 2008). Arab Muslim immigrant women, for example, may have different social networks and social support expectations than the samples in the psychometric evaluation studies noted above.

Another yet to be resolved issue regarding the MSPSS pertains to the authors’ choice of the term “special person” to describe a source of social support. This term was designed specifically to allow respondents to interpret items in ways most relevant to themselves (Zimet et al., 1988). The authors were of the opinion that the use of a more specific term could weaken the scale by presuming the existence of particular relationships (Canty-Mitchell & Zimet, 2000). Our concern, however, was that using a term that allows respondents to construe items idiosyncratically could lead to ambiguous interpretations.

The potential for ambiguity was demonstrated in a study by Prezza and Pacilli (2002). Immediately after they collected data with the MSPSS, respondents were asked to specify the role relationship of the person they were referring to when they responded to items about a special person. Married respondents most frequently but not exclusively identified their spouses. Separated, divorced, or widowed respondents most frequently but not exclusively identified their children. For never-married respondents, the most frequently identified special person was almost equally divided between close friend or parent; a few but not many identified a romantic partner. Without the level of probing undertaken by Prezza and Pacilli (2002), researchers using the MSPSS in its present form would have to make assumptions about who respondents are classifying as a special person. Such assumptions are problematic when researchers are interested in the specific distribution of social support by source.

Adaptation of the MSPSS for Studying Arab Immigrant Women

To address this issue and accommodate our specific interest in spousal support as distinct from family support, we modified the wording of MSPSS items on two of the original three MSPSS subscales. Instead of measuring perceived adequacy of support from significant other, the MSPSS-AW measured perceived adequacy of support from husbands. With regard to the Family subscale, respondents were instructed that husbands were to be considered separately from other family members. Since the term “special person” on the Significant Other subscale was originally intended to be interpreted as a particularly close relationship that could be of a romantic nature, asking about husbands instead of a special person also was consistent with the traditional cultural expectation that Muslim women refrain from romantic relationships outside of the context of marriage (Kulwicki, 2008). The resulting MSPSS-AW, like the original MSPSS, had 12 items, or four items for each of the three sources of support.

We also collapsed the 7-point rating scale (very strongly disagree [1] to very strongly agree [7]) to a 3-point rating scale. The rationale was that Arabs, similar to other non-European-American groups, are less likely to use middle response categories when presented with this many options (Hui & Triandis, 1989; Marin & Marin, 1991). The three points were coded disagree (1), neutral (4), and agree (7) in order to maintain comparability with prior versions of the scale.

The following validation hypotheses about the MSPSS-AW were tested:

1. Like the three-factor structure of the original MSPSS according to important dimensions of people's social lives, we anticipated obtaining a three-factor structure, specifically family, husbands, and friends.

2. Consistent with literature about direct and indirect effects of social support (Cohen et al., 2000; Taylor, 2007), we hypothesized that perceived support from family, husband, and friends will be inversely related to emotional distress.

3. Based on research and theory about maladaptive coping (Skinner et al., 2003), perceived support from family, we hypothesized that perceived support from family, husband and friends will be inversely related to self-blame and avoidance coping.

4. Consistent with theories about constructive coping (Skinner et al., 2003), we hypothesized that perceived support from family, husband, and friends will be positively related to assistance seeking and problem solving coping.

We noted after testing hypotheses 2, 3, and 4, however, that the Husband subscale was performing differently than the other two MSPSS-AW subscales. Thus, we added a discriminant validation hypothesis to further investigate the construct validity of the Husband subscale:

1. We hypothesized that daily hassles specific to husbands will be negatively correlated with support from husbands and this correlation will be stronger than the correlations between daily hassles specific to husbands and support from family and friends.

METHOD

Participants

The sample for this psychometric evaluation consisted of 539 Arab Muslim immigrant women residing in metropolitan Detroit. They were part of a larger study on mother–child adjustment in Arab Muslim immigrants to the United States. Due to the focus of the larger study, the sample included only women who had one or more children between the ages of 12 and 15. Unmarried women were included in the larger study but omitted from this psychometric evaluation. Questions on the MSPSS about spousal support were not applicable to unmarried women, and their small number in our study sample did not permit a separate evaluation of the structure of their social support. Reading ability and English ability were not criteria for study participation because data collection occurred verbally and respondents had the choice of English or Arabic. Most women (97%) chose Arabic as the language for data collection.

The sample was recruited through verbal advertisement by Arabic-speaking research assistants. The research assistants also were immigrants and representative of the countries of origin in the local Arab community. Of note is that the local Arab immigrant community is the largest one in the United States, with a conservative estimate of 200,000 to 250,000 Arab immigrants living in ethnic enclaves in metropolitan Detroit (Schopmeyer, 2000; Zogby, 1998).

Almost half of the women (43.8%) were born in Iraq, about a third (33.6%) were born in Lebanon, and 13.4% were born in Yemen. The remaining 10.2% were from one of 11 other Arab countries in the Middle East or North Africa. Most women (84.2%) were homemakers who were not looking for work. More than half of the women (58.9%) reported an income of less than $20,000. A majority of the women (64.7%) had less than a high school degree. Only 8.9% had a college degree. The average length of time living in the United States was 8.24 (SD = 4.20) years. The average age was 40.20 (SD = 6.45) years. Almost equal numbers of the women entered the United States as immigrants (47.7%) or refugees (44.5%), with the remaining 7.8% entering on tourist, student, or work visas before seeking permanent residence. Close to half of the women's husbands were employed full time (46.2%) and almost a fifth (18%) were employed part time. The low rate of employment among husbands was due to the number of refugees in our study sample. Women with refugee status were significantly more likely than women without refugee status (p < .01) to have husbands who were unemployed, presumably due to disability from premigration trauma.

Data Collection

Research assistants contacted interested study participants and arranged a time for data collection. Data collection occurred in the study participants’ homes and involved obtaining consent and verbally administering the measures. The women were given $30 for their time.

The measures for psychometric evaluation included the measure of interest, the MSPSS-AW, and the measures used for concurrent and discriminant validation, which were the Profile of Mood States (POMS; McNair, Lorr, & Droppelman, 1981); the Seeking Social Support, Problem Solving, Blaming Self, and Avoidance scales from the Revised Ways of Coping Checklist (RWCCL; Vitaliano, Russo, Carr, Maiuro, & Becker, 1985); and a single item from the Daily Hassles Scale (DHS; Kanner, Coyne, Schaefer, & Lazarus, 1981). Following procedures outlined by Werner and Campbell (1970), a team of bilingual experts developed an Arabic language versions of the measures through translation and back-translation, and discrepancies were resolved through discussion.

The POMS, our indicator of emotional distress, is a 65-item 5-point adjective rating scale of mood disturbance, identifying Tension–Anxiety, Depression–Dejection, Anger–Hostility, Vigor–Activity, Fatigue–Inertia, and Confusion–Bewilderment. A total mood disturbance score is obtained by reverse coding the Vigor–Activity items and summing the scores across 58 of the items that make up the total score (seven items from a Friendliness scale are excluded from the total score). High scores indicate greater mood disturbance. The highest maximum score is 232. The POMS is used extensively and has satisfactory reliability and validity data from a wide variety of studies. The internal consistency coefficients obtained from various studies range from 0.84 to 0.95 for the five subscales. Statistically significant negative relationships between total Mood Disturbance and effective coping strategies also have been obtained and provide evidence for the validity of the measure (McNair & Heuchert, 2003). We used the total mood disturbance score rather than subscale scores because of high factor correlations obtained with Arabs (Aroian, Kulwicki, Kaskiri, Templin, & Wells, 2007). Cronbach's alpha for the POMS with this study sample of Arab Muslim immigrant women was 0.95.

The RWCCL is a widely used multidimensional coping measure with satisfactory reliability and validity (Vitaliano, Wolfe, Mauiro, Russo, & Katon, 1990; Vitaliano et al., 1985). The internal consistency reliability of the scales obtained from various samples range from 0.72 to 0.88 (Vitaliano). Support for concurrent validity includes significant correlations in expected directions with measures of anxiety and depression (Vitaliano). Respondents are asked to focus on a recent primary stressor and respond to each item according to the degree to which a particular coping strategy was used to deal with the identified stressor. Each scale assesses a different type of coping, including seeking social support (six items), problem focused (15 items), blaming self (three items), and avoidance (10 items). Example items are “Talked to someone to find out about the situation” (Seeking Social Support), “Made a plan of action and followed through with it” (Problem Focused), “Realized I brought the problem on myself” (Blames Self), and “Went on as if nothing had happened” (Avoidance). Response options ranged from never used or not applicable (0) to regularly used (3). Relative rather than raw scores were calculated to represent the percentage of efforts made on specific strategies and not bias scores by total number of efforts endorsed on the scales. High scores indicate more coping for each type of coping. Internal consistency reliabilities of the RWCCL scales for this study were 0.75 for Seeking Support, 0.88 for Problem Focused, 0.59 for Blames Self, and 0.69 for Avoidance.

The single item from the DHS (Kanner et al., 1981) was “Problems with your spouse.” This item was rated on a 4-point scale from “Did not happen or was not annoying” (0) to “Extremely annoying” (3). A high score (maximum 3) indicates greater problems with one's spouse.

RESULTS

Testing the Proposed Factor Structure

Consistent with the three-factor structure of the original MSPSS according to important dimensions of people's social lives, we anticipated obtaining a three-factor structure of the MSPSS-AW, specifically, family, friends, and husbands. Confirmatory Factor Analysis (CFA) was used to test this proposed factor solution using Analysis of Moment Structures (AMOS) Version 7 (Arbuckle, 2006). A model was constructed that represented the items and the corresponding three factors. The latent factors were allowed to covary, and items were allowed to load only on one factor. The scale for the factors was determined by fixing one loading on each factor to unity. The adequacy of the model fit was evaluated using chi square, the comparative fit index (CFI), and root mean square error of approximation (RMSEA) to assess the degree of fit between estimated and observed variance/covariance matrix.

The initial model was evaluated and modified in a random split of the cases, and cross validation was achieved using a multigroup method. First, the total sample (N = 539) was randomly split into two groups (n ₁ = 269, n ₂ = 270), and the initial model was simultaneously estimated in both groups independently. This resulted in the following fit: χ² (102, N = 539) = 271.30, p < .01, CFI = .93, and RMSEA = .06. This model was improved by adding a path to show covariance between the error terms on two of the Family subscale items, talking to the family about problems and willingness of the family to help with decisions. The improvement in fit was significant (Δχ² (2) = 65.93, p < .01); χ² (100, N = 539) = 205.37, p < .01, CFI = .96, and RMSEA = .04. Next, the factor loadings of the independently estimated models were constrained to be equal across groups and reestimated. The resulting difference in fit was not significant, indicating good stability of results across independent samples (Δχ² (9) = 6.42, p > .05); χ² (109, N = 539) = 211.79, p = .00, CFI = .96, and RMSEA = .04. The difference in unstandardized factor loadings for the independently estimated groups ranged from −.06 to .09, with a mean difference of .01 (SD = .04), which suggested that unstandardized estimates were very similar for the two samples. Thus, for the final analysis the total sample was used to obtain the final parameter estimates.

Table 1 displays the standardized factor loadings obtained with the total sample for the three subscales of the MSPSS-AW. The standardized factor loadings for the 12 items ranged from 0.41 to 0.85, and all were statistically significant (p < .01). The most salient item for the Husband subscale was “My husband is around when I am in need.” For the Friends subscale, “I can count on friends when things go wrong” was the most salient item. For the Family subscale, the items “I get the emotional help and support I need from my family” and “My family really tries to help me” were the most salient. The overall fit of the CFA model for the total sample was good: χ²(50, N = 539) = 148.97, p < .01, CFI = .96, and RMSEA = .06 with respect to acceptable standards for these fit indices (Kline, 1998).

TABLE 1 Standardized Factor Loadings From Confirmatory Factor Analysis of the 12 MSPSS-AW Items by Subscale

MSPSS-AW by subscales	Factor loadings
Husband
My husband is around when I am in need.	0.85
I can share my joys and sorrows with my husband.	0.81
My husband is a real source of comfort to me.	0.78
My husband cares about my feelings.	0.83
Friends
My friends really try to help me.	0.75
I can count on friends when things go wrong.	0.77
I have friends with whom I can share joys and sorrows.	0.71
I can talk about my problems with my friends.	0.60
Family
My family (other than husband) really tries to help me.	0.78
I get the emotional help and support I need from my family (other than husband).	0.78
I can talk about my problems with my family (other than husband).	0.41
My family (other than husband) is willing to help me make decisions.	0.49
Note. MSPSS-AW = Multidimensional Scale of Perceived Social Support-Arab Women.
All items were significant at p < .001.

The three factor scores were not highly correlated (range from −.02 to .26). Family factor scores correlated significantly (p < .01) with both Friends factor scores (r = .26) and Husband factor scores (r = .15). The correlation (r = −.02) between Friends factor scores and Husband factor scores was not significant.

Means, Standard Deviations, and Internal Consistency of the MSPSS-AW

The highest mean score was 6.43 (SD = 1.36) for the Husband subscale followed by a mean score of 5. 75 (SD = 1.63) for the Family subscale and a mean score of 4.34 (SD = 2.04) for the Friends subscale, indicating that the husband was perceived to be highest on social support followed by family and friends, respectively (see Table 2). The mean for the total scale was 5.51 (SD = 1.10), indicating that perceived social support scores were in the higher range given the possible range of 1 to 7.

TABLE 2 Means, Standard Deviations, and Reliability Estimates for Each of the Subscales and the Total MSPSS-AW

	Husband	Friends	Family	Total
M ^a	6.43	4.34	5.75	5.51
SD	1.36	2.04	1.63	1.10
(Cronbach's α)	0.89	0.80	0.73	0.74
Note. MSPSS-AW = Multidimensional Scale of Perceived Social Support-Arab Women.

Internal consistency reliability estimates for the three MSPSS-AW subscales ranged from good to very good: Cronbach's alpha for the Husband subscale was 0.89, for the Friends subscale was 0.80, and for the Family subscale was 0.73. The total MSPSS-AW had good internal reliability (Cronbach's α = 0.74).

Concurrent and Discriminant Validity

We hypothesized that perceived support from family, friends, and husband would be inversely related to emotional distress and two coping strategies—blaming self and avoidance—and positively related to two additional coping strategies—seeking social support and problem focused coping. We added an additional hypothesis that support from husband would be inversely related to hassles with spouse. We also anticipated that this correlation between hassles with spouse and support from spouse would be stronger than the correlations between hassles with spouse and support from family and friends.

To test these hypotheses, Pearson's r and differences in r (Δr) were used to obtain the correlations between the measures of emotional distress (POMS), coping (the Self-blame, Avoidance, Seeking Social Support, and Problem Focused Scales of the RWCCL), hassles with spouse (i.e., the single item from the DHS), and each of the MSPSS-AW subscales. A structural equation modeling (SEM) analysis procedure outlined by Cheung and Chan (2004) was used to statistically compare dependent correlations to test the discriminant validity of the Husband subscale.

Table 3 displays the correlations of the MSPSS-AW subscales with the POMS, the RWCCL scales, and the DHS item about problems with spouse. Scores from POMS, the RWCCL scales, and the DHS item about problems with spouse showed a different pattern of correlations with the Husband subscale than with the Friends and Family subscales. In almost every case, the differences in these patterns were statistically significant (Δr < .01). In partial support of our original concurrent validation hypotheses, social support from husband negatively was associated with the woman's POMS total score (r = −.34) and her use of self-blame (r = −.22) and avoidance (r = −.23) coping, indicating a strong negative relationship between support from husband and the women's emotional distress, self-blame, and avoidance coping. Also in partial support of our original concurrent validation hypotheses, there were significant positive correlations between the Friends and Family subscale scores and the women's use of seeking social support (r = .18 and .16, respectively) and problem focused coping (r = .16 and .17, respectively). Unexpectedly, however, the Husband subscale did not correlate significantly with the woman's seeking social support or problem focused coping. Also unexpectedly, the Friends and Family subscales did not correlate significantly with the women's emotional distress or self-blame or avoidance coping.

TABLE 3 Bivariate Correlations Between the MSPSS-AW Subscales and POMS Total Score, RWCCL Scales, and Spouse Daily Hassles Item and Discriminant Validity Coefficients (Δr)

	POMS		RWCCL Seeking Support		RWCCL Problem Focused		RWCCL Blames-Self		RWCCL Avoidance		Spouse Daily Hassles Item
MSPSS-AW Subscales	r	Δr	r	Δr	r	Δr	r	Δr	r	Δr	r	Δr
Husband	−.34∗∗		−.01		.08		−.22∗∗		−.23∗∗		−.58∗∗
Friends	−.06	−.28∗∗	.18∗∗	−.19∗∗	.16∗∗	−.08	.01	−.21∗∗	−.06	−.29∗∗	.06	−.64∗∗
Family	−.06	−.28∗∗	.16∗∗	−.17∗∗	.17∗∗	−.09	−02	−.20∗∗	−.06	−.29∗∗	−.12∗∗	−.46∗∗
Note. MSPSS-AW = Multidimensional Scale of Perceived Social Support-Arab Women, POMS = Profile of Mood States, RWCCL = Revised Ways of Coping Checklist. The discriminant validity coefficients were computed by taking the differences between correlations within each column; for example, Δr for Friends in the POMS column was computed by taking −.34 − (−.06) = −.28. The formula is Δr Friend = r Husband—r Friend.

As shown in Table 3, the item from the DHS about problems with spouse correlated −.58 with husband support, providing strong evidence for the construct validity of the Husband subscale. This is a particularly high correlation if one takes into account the unreliability of a single item. The correlation between scores from the DHS item about problems with spouse and the Husband subscale also was significantly stronger than the correlations between the DHS item about problems with spouse and the Family and Friend subscales, further indicating the validity of the Husband subscale.

DISCUSSION

Findings from this study demonstrate that the MSPSS-AW is a reliable and valid measure of social support for Arab immigrant women. The hypothesized three-factor structure was supported, as evidenced by the good fit indices and factor loadings. The Husband subscale, which was revised to specify husband rather than a “special person,” had excellent, in fact higher, factor loadings than the items from the other, unrevised subscales.

The concurrent validation hypotheses, however, were only partially supported. Scores from the Friends and Family subscales were almost totally unrelated to scores from the measure of emotional distress, specifically the POMS, and two of the coping scales from the RWCCL, Self-Blame and Avoidance. Conversely, scores from the Husband subscale were almost totally unrelated to the other two coping scales from the RWCLL, Seeking Social Support and Problem Focused coping. We investigated the validity of the Husband subscale further by determining how this and the other MSPSS-AW subscale scores were related to scores from an item about hassles with spouse. Our findings provided assurance of the concurrent and discriminant validity of the Husband subscale.

The striking pattern of the relationships we obtained for the original concurrent validation hypotheses as well as the low correlations among the husband, family, and friends factor scores suggests that social support from husbands functions differently than social support from family and friends. Husband support was related exclusively to measures of affect or emotional constructs, namely, emotional distress as measured by the POMS and the Avoidance and Self-Blame scales of the RWCCL. Avoidance, for example, regulates one's emotions by not thinking about the problem (Skinner et al., 2003). Similarly, attributing blame to one's self focuses on one's emotions and not on the problem (Skinner et al., 2003). In contrast, support from family and friends were related exclusively to measures of constructs that mostly were instrumental in nature, namely, the Problem Focused and the Seeking Social Support scales of the RWCCL. Although seeking support from others theoretically can result in instrumental or emotional support, four of the six items from the particular measure we used, the Seeks Social Support scale of the RWCCL, pertained to finding solutions or changing the problem (e.g., “Asked someone I respected for advice and followed it”). Only two items from the Seeks Social Support scale, “Talked to someone about how I was feeling” and “Accepted sympathy and understanding from someone” could be construed as seeking emotional rather than instrumental support.

The relationships between spousal support and emotional constructs also are consistent with literature about psychological, sociocultural, and marital adaptation as different domains of immigrant adaptation (Ataca & Berry, 2002; James, Hunsley, Navara, & Alles, 2004). Psychological adaptation pertains to feelings of well-being and satisfaction (Ward & Seale, 1991), a domain that clearly is affective in nature. In contrast, sociocultural adaptation involves gaining the knowledge and skills for everyday living in the new country (Ward & Seale, 1991). Gaining knowledge and skills are instrumental processes geared toward achieving pragmatic outcomes. Marital adaptation is specific to communication and maintaining a good marriage while accommodating as individuals and as a couple to the new country (Ataca & Berry, 2002; James et al., 2004). Although theorists posit that each of these domains are distinct, studies attempting to illustrate this distinction have found that marital and psychological adaptation share overlapping sets of predictors (Ataca & Berry, 2002; James et al., 2004). Consistent with the pattern of relationships we obtained between the MSPSS-AW subscales and affective/emotional variables, the overlap between marital and psychological adaptation suggests that the marital relationship is intertwined with emotional state in immigrants.

Another explanation for the lack of a significant relationship between husband support and coping by seeking social support may have to do with the wording of the items on the Seeking Social Support scale of the RWCCL. We asked cultural informants to review our findings and provide potential explanations. They attributed the lack of a relationship between coping by seeking social support and support from husbands to the reference point used for the Seeking Social Support items. All but one of the six items on the Seeking Social Support scale used the word “someone” as the reference point (e.g., “Asked someone for advice”). The remaining item pertained to getting professional help. The cultural informants doubted that Arab women would interpret someone as referring to their spouse. The term, they explained, is too impersonal and connotes someone other than husband in the Arab culture.

In our study, Arab immigrant women reported husbands as providing the most support, followed by family and friends. These findings are consistent with literature describing the central importance of marriage and family in the Arab culture (Carolan, Bagherinia, Juhari, Himelright, & Mouton-Sanders, 2000; Kulwicki, 2008). The high levels of support reported by our study participants may be because our study participants had lived in the United States an average of 8.2 years (SD = 4.20). This was likely sufficient time to rebuild social networks. Another possibility for the high level of support reported by our study sample may be due to technological changes in how people access social support. Mobile phones and the Internet potentially allow immigrants to communicate daily with family and friends in the homeland, thereby providing the opportunity to obtain social support from a distance. We were unable to explore this possible explanation, however, because we did not inquire about mode of contact or geographic location.

A particularly notable contribution from our study is that it developed a reliable and valid subscale, social support from husbands. This subscale offers researchers the potential to further explore distinctions among psychological, sociocultural, and marital adaptation and to focus in more depth on marital adaptation, which is a neglected domain in current research on immigrants (Ataca & Berry, 2002; James et al., 2004). Nonetheless, by replacing the Significant Other subscale with the Husband subscale, we rendered the MSPSS-AW inapplicable to unmarried Arab women. Although unmarried Arab women can complete the Family and Friends subscales and their total social support scores can be adjusted accordingly, this approach does not address the possibility that support from some other particularly close relationship (i.e., a special person) substitutes or compensates for the lack of spousal support. More comprehensive assessment of the configuration of social support networks for married and unmarried Arab immigrants would provide more information for researchers to explore how social support functions for this population.

One limitation of our study is that we did not randomly select our sample. Another limitation is that we did not include Arab men. Future research is needed to evaluate an Arab version of the MSPSS with Arab men. We also recommend that researchers include the original items from the MSPSS Significant Other subscale in future research, regardless of whether study participants are married or unmarried. Due to the ambiguity noted by Prezza and Pacilli (2002) about the Significant Other subscale, however, we suggest that researchers ask people to specify the role relationship of their so-called special person when responding to these items. Given that the mean score we obtained for husband and family support were close to the top of the range of possible scores, researchers also may want to reconsider our decision to use a 3-point rating scale. With more response options, the Arab women in our study may have rated their social support as moderately rather than very high. Last, we recommend asking about the location and mode of contact for support from various sources to explore the possibility that modern technology offers new ways of obtaining support, offsetting the tendency for immigration to result in restricted social support networks.

This research was funded by a grant from the National Institutes of Health, Mother-Child Adjustment in Arab Immigrants and Refugees, R01NR 008504, to Karen Aroian, Principal Investigator. We thank the author of the original measure, Dr. Gregory Zimet, for granting permission to adapt the MSPSS for use with Arabs.

Notes

^aTotal scale and subscale scores are averaged over items (range = 1 to 7).

∗∗p < .01; ∗p < .05.