Psychometric properties and cross-cultural equivalence of the Arabic Social Capital Scale: instrument development study

1. Background 

Children with special health care needs (SHCN) are defined as those who “have or are at increased risk for a chronic physical, developmental, behavioral, or emotional condition and who also require health and related services of a type or amount beyond that required by children generally” (Newacheck and Kim, 2005, p. 10). Parents of children with SHCN report both positive and negative influences of caregiving on their health (Esdaile and Greenwood, 2003, Khamis, 2007, Murphy et al., 2007, Trute et al., 2007). The caregiving and emotional challenges that are part of raising a child with a SHCN have been documented extensively over the past four decades, but there continues to be a gap in research that focuses on the societal conditions that affect families of these children (Ray, 2003) and how health care professionals can develop interventions that focus on the protective effects of certain social contexts.

According to the World Health Organization (WHO, 2006) 10% of the world's population – approximately 650 million people, of which 200 million are children – experience some form of disability. Most of the children who have a disability live in low- and middle-income countries, where research and progress in the area of childhood disability is “seriously lagging” (Maulik and Darmstadt, 2007, p. 52). There is a need for rigorous research in developing countries that focuses on childhood disability and that uses reliable, valid tools to assess the social context of health and support for families.

The social context of health and support for families of children with SHCN is a universally important area of health research. Moore et al. (2006) encouraged the development of research that views networks as social contexts that influence the behavior and practices of individuals. Furstenberg and Hughes (1995) emphasized the importance of examining the ways that social embeddedness in family and community contexts influence the course of human development. In the past few years, there has been increased interest in social capital in the health field and research studies have indicated that social capital is generally positively related to health.

Social capital is defined as an investment in relationships that facilitates the exchange of resources. This definition was “etymologically derived from the terms social, referring to interpersonal relationships, and capital, referring to assets that accrue through investment” (Looman, 2006. p. 325). Many definitions of social capital include trust as a component, consistent with Putnam's (1996) early work, where social capital is described in terms of three key features of social life: networks, norms and trust. In Putnam's later work, however, the focus on trust is shifted toward a focus on reciprocity, which emphasizes the social nature of social capital which relies on more than one individual. The definition of social capital for this study emphasizes this reciprocal nature of relationships and represents the family's investment in relationships in various social contexts (i.e. schools, faith-based organizations, and neighborhoods) as well as the respondent's perception of the investment by others in their relationships with the family.

Appreciating social capital and its multidimensional nature may help explain how certain types of protective factors influence psychological well-being (Furstenberg and Hughes, 1995). Poortinga, 2006a, Poortinga, 2006b analysed data from the 2002 Health Survey for England (N=7394) and found that social capital and support are important determinants of self-rated health and health behaviors. Studies have linked social capital to self-rated health (e.g. Hyyppä and Mäki, 2001, Kawachi et al., 1999); social capital has also been linked to cardiovascular and cancer mortality rates (Kawachi et al., 1997), child mental health (Caughy et al., 2003, Drukker et al., 2003), and general socioeconomic success of young adults (Furstenberg and Hughes, 1995).

Among families with children, social capital has been identified as a possible protective factor for the child in the context of risk factors such as poverty and low human capital (Harpham et al., 2006, Jones et al., 2002). Measuring social capital and conducting research in developed and developing countries is essential to understanding the mechanism underlying the relationship between social contextual factors and positive health outcomes for children. In a national longitudinal study of over 15,000 children in neighborhoods in Canada, Jones et al. found that certain aspects of neighborhoods related to social support and social capital have a statistically significant effect on the slope of the relationship between long-term poverty and children's physical health. Harpham, De Silva and Tuan, in a study of 2907 mother–child dyads in Vietnam, found that higher levels of cognitive social capital and social support, but not maternal membership of groups, were positively associated with a range of acute and chronic child health indicators among 1 year olds, and with child mental health among 8 year olds. Harpham, De Silva and Tuan hypothesized that high levels of maternal social capital permit mothers to access more services and assets, which in turn are associated with increased child health.

In a collectivist society, such as Egypt, the role of culture in health and illness is particularly important to understand. According to Hofstede (2001), an important dimension of culture is the extent to which people are expected to stand up for themselves (i.e. individualistic) or to act as a member of a cohesive group (collectivistic). Coker (2005) noted that, “in Egypt, one is part and parcel of the social environment,” which serves as a protective factor in light of “moral and religious factors to help the sick and infirm” (p. 928). While Arab families have traditionally been underrepresented in health research (Khamis, 2007, Maulik and Darmstadt, 2007), research that has been conducted in Arab countries suggests a protective role of social networks among families of children with disabilities. Duvdevany and Abboud (2003), for example, found that Arab mothers of children with intellectual disabilities reported a greater sense of well-being and lower stress if there was a broader informal support network for the family. The authors suggested that informal support served as a bridge to formal support and enabled mothers to redefine their roles and interactions with the child. A better understanding of social capital in Arab countries is necessary in order to design interventions that draw on the unique social context of Arab families raising children with special needs.

Few empirical studies measure social capital in developing countries such as Egypt. The purpose of this study was to evaluate the psychometric properties and cross-cultural equivalence of the Arabic translation of the Social Capital Scale (SCS; Looman, 2006). Carlson and Chamberlain (2003) recommended developing measures of social capital that are specific to health promotion. The SCS was developed with the assumptions that there are measurable indicators of investment by families and communities in their relationships with one another, and that these indicators are best assessed in a specified social context (i.e. raising children with SHCN) and with a specified outcome (i.e. the health of children with SHCN) in mind. The SCS was developed from a theoretical foundation of symbolic interaction (Blumer, 1969) and human ecology (Bronfenbrenner, 1979), using a process of item generation that included concept analysis and focus groups with parents of children with SHCN. For this reason, the SCS is unique in its applicability to the population of families raising children with SHCN and in its strong theoretical foundation.

The SCS, originally developed in English, was selected because it operationalizes social capital in the population of families with children with SHCN, and is the only known measure of social capital that has been psychometrically tested. Harpham et al. (2002) identified two important issues related to research on social capital and health. First, there is relatively little experience of measuring social capital in developing countries. Second, none of the new and emerging measures of social capital have measured reliability and/or validity. This study addresses the need for studies of social capital in developing countries by translating and evaluating the cross-cultural validity of the SCS in Egypt.

2. Instrument and translation 

Cross-cultural research is conceptually and methodologically complex. Beck et al. (2003) warn that translation of instruments should be prudent, noting, “Existing English-language instruments should not be blithely translated into other languages and assumed to have the same measurement properties across cultures” (p. 65). In addition to the potential problem for ethnocentricity when questionnaires are translated from English, most of the methodological problems with cross-cultural research relate to the quality of translation and the comparability of results in different cultural and ethnic groups (Sperber, 2004). The challenge for this study was to adapt the SCS in a way that would be culturally relevant while retaining the meaning of the original items as developed by Looman (2006), particularly due to the subtle nuances of language that may have different meanings outside the United States.

The original SCS is a set of 20 objective indicators of an investment by families and communities in their relationship with each other, as perceived by the parent or caregiver completing the survey (Looman, 2006). The SCS was originally developed and tested in four phases. In the first phase, 80 items were generated to represent the domain of indicators of social capital in the context of families raising children with SHCN. The domain of indicators emerged through both a concept analysis and a series of focus groups conducted by the first author. In the second phase, content validity of the items was determined by asking experts in the field of family science to rate the items for relevance and calculating a content validity index. Face validity was evaluated by pilot testing the items with a small group of parents. The pool of items thus was reduced to 38 in phase two. Phase three consisted of administering the 38 items in a mailed survey to 186 parents and caregivers of children with SHCN. In phase four, psychometric testing included item analysis, construct validity evaluation, and reliability analysis.

The intended respondents for the SCS are parents or caregivers who have a child with a SHCN. Items on the SCS are declarative statements followed by five response options indicating the respondent's degree of agreement or disagreement with the statement. Respondents are asked to “Read each statement and place an “x” in the column that best represents how well the statement describes your family and the community you live in.” Items are scored from 1 (strongly disagree) to 5 (strongly agree). Approximately one-third of the items are worded negatively to minimize acquiescence response set. The range of total scores on the instrument is 20–100. Higher scores suggest the presence of more social capital, or investment in the family–community relationship. Original psychometric testing of the SCS in the sample of 186 U.S. parents of children with SHCN demonstrated internal consistency (standardized item alpha reliability coefficient=.84), stability over time (2-week test–retest reliability correlation coefficient=.90, p<.01), and construct validity (as evaluated through factor analysis, correlation with existing measures, and hypothesis testing). Exploratory factor analysis suggested five subscales – Common Good, Sense of Belonging, Spiritual Community, School Connection, and Informing/Asking – that provide a means to assess family strengths and areas for improvement related to their investment in relationships that support health.

Social capital is distinct from social networks and support in that social capital is a feature of the social structure, not of the individual actors within the structure (Lochner et al., 1999). Original psychometric testing of the SCS demonstrated that the instrument measures something similar to but distinct from social support (Looman, 2006). According to House (1981), social support is the functional content of relationships and can be categorized as four types of supportive behaviors: emotional, instrumental, informational, and appraisal. Social capital, defined as an investment in relationships, is less about supportive behaviors and more about the relationships through which support is exchanged. An example of this distinction is an item in the SCS: “We ask for help when we need it”, which focuses on the family's actions that promote the exchange of support rather than the support itself. This distinction is important for families of children with SHCN because it acknowledges that families may have support available to them but they may not actually receive that support if they do not let others know that they need it. Similarly, the SCS assesses the respondent's perception of the availability of the support that is due to investments by others in their relationships with the family.

The Translation Monitoring Form (TMF; VanOmmeren et al., 1999) was used to guide the translation and adaptation of the SCS for use in an Arab population. The TMF requires the systematic use of strategies advocated by previous translation and adaptation researchers to enhance the methodical preparation of instruments for transcultural use (VanOmmeren et al.). Quality of the translation is enhanced by the translators’ ability to be consistent in identifying and correcting incomprehensible, unacceptable, incomplete and irrelevant translated items. The first step was translation of each item from English to Arabic by the second author, who is bilingual and a native of Egypt who works in health care systems with families of children with SHCN. Second, a second bilingual translator recorded a lexical back-translation of each item. Next, through an iterative process of translation and back-translation, a panel of both bilingual and monolingual experts (including the first and second authors) evaluated the comprehensibility, acceptability, relevance and completeness of each item using the TMF. The process was continued until the back-translation reflected congruence of meaning between the original and translated SCS.

In the process of translation, back-translation, and evaluation of face validity, some items were noted to have questionable relevance as translated directly. These items regarded religious or church-based activities. Specifically, 4 of the original 20 items were related to faith-based activities, and these items referred to church as a primary example of a place of worship (e.g. “We participate in activities through a church or place of worship”). Because 90% of Egyptians are Muslim (U.S. State Department, 2007), these items were revised so that the term mosque replaced church as the primary example. Additional minor changes to wording were made based on parent and expert feedback; these changes were revised until there was consensus among the experts that the meaning was clear and the items were comprehensible, acceptable, relevant, and complete.

3. Research design and methods 

This cross-sectional study consisted of psychometric testing of the translated SCS in a sample of Egyptian parents of children with SHCN. Egypt is one of several developing countries that have no established oversight agencies for human subject protection in research (Office for Human Research Protections, OHRP, 2007). Prior to data collection, approval to conduct this study in two metropolitan health clinics in Alexandria, Egypt was obtained by the second author through an Egyptian association for child mental health, and clinic nurses were trained to administer the written survey. Verbal informed consent was obtained from each participant before collecting data, and an introduction to the written survey included a statement describing the confidentiality procedure for maintaining privacy and the participants’ right to refuse to participate.

3.1. Sample and data collection 

A convenience sample of lay Egyptian parents of children with SHCN was recruited from two metropolitan health clinics in Alexandria, Egypt to complete the Arabic SCS. To be eligible to participate, respondents had to be a parent or caregiver of child with a chronic health condition between the ages of 1 and 18 years. The final sample included 117 respondents for whom there were sufficient data for analysis. Each respondent was from a different family (i.e. no two respondents represented the same household). The sample included primarily biological (91%) parents between the ages of 20 and 56 years of age (M=38, S.D.=8.5). Seventy percent of respondents were female, and 77% were married (Table 1). The target children were between the ages of 1 and 18 years of age (M=9.6, S.D.=3.7), and there were slightly more males (59%) than females. Based on parent report, 46% of these children had difficulty with mobility; 68% had difficulty with self-care, 68% had difficulty with communication, and 89% had difficulty with learning.

Table 1. General characteristics of survey respondents by respondent gender (n=117)

		Total		Female		Male
		N	%	N	%	N	%
	Relationship to child
	Biological parent	106	90.6	76	92.7	30	85.7
	Step parent	5	4.3	1	1.2	4	11.4
	Adoptive parent, foster parent, guardian, or other	6	5.1	5	6.1	1	2.9
	Educational status
	Some secondary school or less	33	28.2	24	29.3	9	25.7
	Secondary school completion or equivalent	38	32.5	27	32.9	11	31.4
	Junior or technical college	22	18.8	16	19.5	6	17.1
	College degree or higher	24	20.5	15	18.3	9	25.7
	Marital status
	Single/never married	1	0.9	1	1.2	0	0
	Married	94	80.3	64	78.0	30	85.7
	Widowed/divorced/separated	22	18.8	17	20.7	5	14.3
	Employment status
	Not working due to child's health	25	21.4	20	24.4	5	14.3
	Not working for other reasons	27	23.1	24	29.3	3	8.6
	Looking for work outside the home	9	7.7	5	6.1	4	11.4
	Working full- or part-time	39	33.3	19	23.2	20	57.1
	Full-time homemaker	17	14.5	14	17.1	3	8.6

4. Data analysis 

Data from the administration of the translated 20-item Arabic SCS were analysed to evaluate its factor structure, reliability and initial validity. LISREL 8.70 (Joreskog and Sorbom, 1996) was used for statistical analysis. Initial evaluation of items included examination of means, variances, and distributions of scores on the items. Using confirmatory factor analysis (CFA), we assessed the fit of 20-item Arabic SCS to the hypothesized five-factor model based on the EFA findings of Looman (2006) in the initial psychometric evaluation of the SCS. Indices of fit and a model chi-square statistic, with the null hypothesis of perfect model fit, were obtained. Due to a poor fit, six items were ultimately removed based on item analysis, and exploratory factor analysis was conducted on the resulting 14-item translated SCS. Since the goal was to attach meaning beyond the observed variables, principal axis factoring was used to generate a factor correlation matrix. Oblimin rotation was used to obtain a simple structure while allowing the factors to be correlated with each other. Cronbach's coefficient alpha was calculated as a measure of internal consistency for the scale.

Direct assessment of conceptual equivalence, according to Flaherty et al. (1988) is usually not possible on a subjective measure; other less direct techniques, such as analysing the relationship of responses to other variables in the study population, must suffice. As a measure of construct validity, hypothesis testing was conducted using an independent samples t-test to determine whether a significant difference exists between mean total social capital scores for two groups of respondents based on the parental rating of the child's overall health. The convenience sample of 117 parents of children with SHCN was used for hypothesis testing because an independent random sample was not available for separate hypothesis testing. Based on evidence in the literature cited previously, it was hypothesized that parents who rated their child's overall health as very good or excellent would have higher mean social capital scores parents who rated their child's health as good, fair, or poor.

5. Results 

Goodness of fit indices for the confirmatory factor analysis indicated that the original five-factor model provided a poor fit to the data, χ2(160)=450.33, p<0.001; RMSEA=0.12 (90% CI=0.11, 0.14); CFI=0.74; NNFI=0.70; SRMR=0.14. Coefficient alpha (standardized) was .69 for the 20-item scale. Due to the poor fit, we examined item-total correlations and factor loadings to identify candidates for item deletion. Six items were identified as particularly problematic, with very low item-total correlations and low factor loadings. Removal of these items enabled further analysis of the structure of the remaining items as a set, but it also meant that cross-cultural equivalence of the Egyptian and English version of the SCS was not possible, since it was not possible to achieve factorial invariance with a different set of items. We opted to continue with analysis of the smaller set of items rather than continue to try to fit the full set to the original five-factor model. Implications of this are described further in Section 7.

After removing the six problematic items, we used exploratory factor analysis (principal axis factoring with oblimin rotation and Kaiser normalization) to determine the underlying factor structure of the remaining set of items. Confirmatory factor analysis was not used at this point, since the original five-factor structure of the SCS was based on a 20-item scale. A four-factor solution resulted from principal axis factoring, with clean loadings of all items at .40 and above and with a load strength at least .14 higher on a primary factor when compared with the secondary factor (Table 2).

Table 2. Results of factor analysisa for the 14-item Egyptian Arabic Social Capital Scale

	Item	Communalities	Factor
			1	2	3	4
	We work with families like our own to help the community understand our needs	.66	.78	−.12	.20	−.10
	We usually ask for help when we need it	.57	.74	.14	−.09	.02
	We talk to others about ways to improve the community	.54	.53	−.04	.17	.25
	We do things with our neighbors to improve the neighborhood	.48	.51	.12	−.11	.34
	When our family is having a hard time, the community does not seem to noticea	.69	.06	.82	.18	-.06
	People in the health care system do not feel our child is importanta	.61	−.03	.77	−.07	.04
	Our child hardly spends time with people outside our familya	.38	.02	.62	.05	.03
	We participate in activities through a church or place of worship	.51	−.14	.21	.72	.02
	If we needed help from the school system, we know how to get it	.48	.32	.00	.55	.02
	If we wanted to make a change in our child's school we would know whom to talk to	.40	.20	−.10	.52	.05
	We work with others in the community to make it a good place to live	.46	.27	.20	.42	.15
	The health care system is set up to work for us	.61	−.15	−.06	.38	.60
	Our child's health is important to this community	.46	.24	.06	−.13	.58
	As parents, we are contributing to the community's well-being	.54	−.07	−.33	.28	.47
	Subscale Cronbach alpha		.79	.77	.71	.66
	Eigenvalues		4.20	2.42	1.59	1.02
	Percentage of explained variance		26.60	14.20	8.05	3.92
	Factor correlation matrix
	Factor 2		.02
	Factor 3		.19	−.13
	Factor 4		.38	−.15	.33

Note. Factor loadings above .40 are in bold. Factor 1=Engagement for the Common Good; Factor 2=Sense of Belonging; Factor 3=Systems Connection; Factor 4=Family Role in Community.

a Reverse-scored.

The four-factor solution explained 53% of the total variance with moderate sampling adequacy (Kaiser–Meyer–Olkin test=0.74), and with Bartlett's test of sphericity significant at the .001 level. Items in the first two factors were generally consistent with two of the original subscales of the English version of the SCS. The first factor explained 27% of the total variance and contained items describing parent engagement in actions that serve the common good. Three of the four items in this factor were in the Common Good factor in the original SCS. The first factor in the Arabic SCS was named Engagement for the Common Good. The second factor explained 14% of the total variance and included three of the items in the original Sense of Belonging Factor; this factor name was therefore retained. The third and fourth factors explained 8.1% and 3.9% of the total variance, respectively. The third factor contained items related to the connectedness in systems external to the family (e.g. places of worship, schools) and was named Systems Connection. The fourth factor, Family Role in Community, contained items related to the significance or role of families in the community and health care system.

6. Discussion 

There is a need for reliable and valid instruments that can be used to conduct research with families in the Middle East and in low- and middle-income countries. The majority of existing literature related to children with SHCN in these areas has been descriptive and epidemiological in nature and does not provide evidence needed to plan and implement interventions. This study was conducted to provide a culturally relevant and psychometrically sound measure of social capital for use in clinical and research contexts with Arab families of children with SHCN.

The 14-item Arabic SCS, translated and modified from the original 20-item English version of the SCS (Looman, 2006) was found to be reliable and valid in this sample, with four internally consistent factors. The significant relationship between (high) social capital and reported overall health of the children in this sample suggests that the instrument may be a useful tool for health-related research and practice with Arab families of children with SHCN. This instrument will enable clinicians and researchers to address an important gap in knowledge characterized by a paucity of research on childhood chronic illness in low- and middle-income countries such as Egypt.

Confirmatory factor analysis suggested a poor fit of the hypothesized five-factor model. There are a number of potential reasons for this poor fit. One explanation is that these items failed to translate conceptually due to social context terminology. For example, in the West, the word “community” has a number of connotations, including the geographic location shared by a group of individuals as well as the social connection of a group of individuals who share interests, beliefs, or political ties. The word “community” as translated into Arabic has the same meaning as the word “society”, which may connote a more highly structured system of human organization. Individuals responding to the items in Arabic may have interpreted the term “community” in a more broad sense than individuals completing an English version of the SCS.

Despite the poor fit with the original five-factor model, the four factors identified in the revised 14-item Arabic SCS are conceptually consistent with the content of the original model. The first factor, Engagement for the Common Good, contains items that describe activities (i.e. doing things with neighbors, talking to others, and working with families) that represent parent investment in their relationship with informal systems of support in the community. Similarly, the third factor, Systems Connection, represents engagement in more formal systems for support. In clinical practice or research with families, parent scores on these subscales might represent the degree to which the family is engaged with (or disengaged from) the informal and formal systems of support in the community. Practitioners might use this information to identify ways the family can either capitalize on this engagement or become more engaged.

The second and fourth factors in the 14-item Arabic SCS – Sense of Belonging and Family Role in the Community – also indicate the degree of investment in relationships between families and external systems. Sense of Belonging includes items that represent the respondent's perception of the investment by others in that relationship, whereas Family Role in the Community represents the family's investment in that relationship. In clinical practice or research with families, parent scores on these subscales may guide the development of interventions at either the family level or the systems level. Subscale scores and total scores on the 14-item Arabic SCS can also be assessed over time to evaluate the effectiveness of interventions aimed at increasing investment in the family–community relationship.

7. Limitations 

A primary limitation of the Arabic SCS is that it does not demonstrate measurement invariance across cultures. In order to compare two different cultural groups using a multi-item measure designed to assess individual differences, items must possess the property of factorial invariance (Beckstead et al., 2008). In our study, the purpose was specifically to establish a useful tool for use with Arab parents of children with special health care needs, not to document the properties of the SCS as it has been used in prior research. While structural equation modeling provides invaluable information regarding the measurement invariance and cross-cultural equivalence of a multi-item scale, analyses of our data from the Egyptian sample suggested that the model was a poor fit and subscales were not internally consistent (i.e. very low item-total correlations). Our conclusion was that there were problematic items that should not be retained, and that once these items were removed, it was clear that the SCS would not have factorial invariance. The 14-item Egyptian SCS will facilitate research to guide interventions for health within groups in Egypt, and future analyses comparing results between cultural groups in other countries will be important to qualify how well the shortened SCS functions in cross-cultural comparisons.

Future studies might focus on the effects of cultural background on the mean scores of the SCS and its items. VanHemert et al. (2001) present a method for evaluating the presence and size of cross-cultural item biases, particularly when a scale developed in a more individualist country (such as the United States) is used in a more collectivist country (such as Egypt). Such analyses could also be employed when using the Arabic SCS with Arab Americans in the United States. Future studies should replicate the factor analysis and reliability using an independent sample. Using a larger sample size would also increase the generalizability of our findings. Another limitation is that hypothesis testing was conducted using the convenience sample of parents of children with SHCN. This was due to limited resources for this study and it was not feasible to recruit an independent random sample. The reliability of the 14-item Arabic SCS was found to be acceptable for group-level comparisons, but reliability of subscales will need to be assessed in other samples. Finally, only 53% of the variance in the 14 items was accounted for by the factor solution; future studies will be necessary to explicate other sources of variance and possibly refine the factor structure. Future studies using the Arabic SCS will shed light on the construct validity of the Arabic SCS in larger samples.

Conflict of interest 

The authors have no conflict of interest to claim.

Ethical Approval 

The University of Minnesota Institutional Review Board approved this study. There was no formal IRB process in Alexandria, Egypt at the time of this study, but permission was obtained from the clinics and a representative of the Egyptian Association for Child Mental Health prior to conducting the study, and verbal informed consent was obtained from each participant.

Funding 

This study was not formally funded.