Introduction
Social support is a multifaceted and complex concept that refers to the amount of assistance an individual can get through interpersonal interactions [
1,
2], including family, friends, peers, and members of a community [
3]. As opposed to received social support (i.e., the actual support that one receives), perceived social support infers the beliefs and/or perceptions of the already-present support provided by the social network when needed [
4‐
7]. Social support has an important role in human health [
8]. Its adequate availability seems essential to provide a buffer for stressful physical and psychosocial events through greater resilience [
9,
10], promote self-esteem [
11], and mitigate the effects of psychological distress [
12]. In this regard, perceived social support has been argued to have a more significant impact on health determinants compared to actual received social support [
13,
14]. However, the existing literature on the effect of social support on physical, mental health and quality of life has led to mixed findings; which is partly due to the use of different measurement instruments [
15].
A number of social support measures have been developed and tested in various groups and populations [
16]. Out of these measures, the Multidimensional Scale of Perceived Social Support (MSPSS) is one of the most widely used worldwide [
3,
6‐
8,
11,
12,
18]. The MSPSS is a 12-item brief, freely available, easy to administer, self-report scale designed by Zimet et al. to subjectively assess “the adequacy of received emotional social support” from three different sources (family, friends, and the significant other) ([
17], p. 186). The original English version of the MSPSS consists of a three-factor construct which had high internal consistency and test-retest reliability, as well as construct validity [
17]. The scale has been translated in many languages (e.g., Italian [
18], Swedish [
19], Polish [
20], Portuguese [
21], Greek [
22], South Korean [
23], Turkish [
24], Persian [
25], Indian [
26], Urdu [
27], Thai [
28], Hausa (Nigerian) [
29], Ugandan [
30], Malawi [
31], Malay [
32]) and countries (e.g., high- [
19,
21‐
23], middle- [
24,
25,
28,
32,
33], and low-income countries [
29‐
31]) across the globe. The psychometric properties of the MSPSS have proven their appropriateness in individuals from a variety of cultural backgrounds, ages, and clinical profiles. For instance, the Swedish version of the MSPSS reproduced the original three-factor structure and supported the good validity and internal consistency (α = 0.91–0.95) of the scale in samples of women with hirsutism and nursing students [
19]. The Polish validation also supported the factorial validity of the MSPSS among university students, as well as good reliability (α = 0.89-0.94) and concurrent validity as evidenced through adequate patterns of associations with psychological indicators (anxiety, loneliness, life satisfaction and current involvement in a romantic relationship) [
20]. Several other translated versions of the MSPSS provided support to the expected three-factor structure, and showed strong internal consistency, including Malawi [
31], Nigerian Hausa [
34], Thai [
35], and Spanish [
36]. Tonsing et al. [
37] found that the Nepali translation of the MSPSS confirmed the original three-factor model, whereas the Urdu version could retain only two factors, with the subscales of Significant Others and Family being combined into one factor and cultural factors suggested to partly explain these discrepancies. Both versions revealed high internal reliability and construct validity in two samples of South Asians living in Hong Kong [
37]. Measurement invariance across gender groups was verified and established in different MSPSS linguistic versions and populations, including and Chinese parents of children with cerebral palsy [
38], Spanish patients with cancer [
36], Romanian elite athletes [
39], as well as Chinese [
40], Nigerian [
41] Romanian [
42], and Indonesian adolescents [
43].
While there is evidence asserting the psychometric strength of the MSPSS across different contexts [
44], literature has also documented a substantial impact of culture on social support access and sources [
45‐
47]. For instance, collectivist cultures promote social cohesion and parenting/family relationships quality; therefore, individuals from such cultures expect extended family than other sources to supply them with any needed social support [
48]. Despite these data, the largest amount of research on social support has emerged from the Western/Eastern world. In addition, the original validation study and subsequent studies examining the MSPSS psychometric quality have been mostly performed in these cultural backgrounds. This limits our knowledge about the pathways linking social support to mental health and prevents evidence-based policymaking in the under-researched contexts, including Arab countries and communities from the Middle East and North Africa (MENA) region.
There are a total of 22 Arab countries geographically distributed over two continents (i.e., Africa and Asia), most of them defined as lower-middle-income economies, traditional, religious and collectivist societies [
49,
50], and having a current population estimated at greater than 450 million people [
51]. Arabic is thus spoken by hundreds of millions of people in both Arab and non-Arab countries. Over the last decades, Arab countries have faced a series of revolutions, armed conflicts, terrorist attacks, widespread violence, traumatic wars, and economic recessions, which have negatively affected their local communities’ mental health [
52‐
54]. At the same time, Arab countries suffer a substantial lack of information, mental health legislation and policy [
54]. One of the main factors that impede access to evidence-informed care and mental health research in Arab countries is the shortage of valid and reliable assessment tools [
55]. As we specifically focus on perceived social support in the present study, we point to the little information available on this construct in Arab contexts. We could find only a few studies among Arab people using the MSPSS in specific populations (e.g., Arab American adolescents [
56] and women [
57], Arab immigrant women [
58], refugees in Jordan [
59], mothers of children with developmental disabilities [
60]); which are far from being representative of the Arab general population. All these observations highlight the strong need for an Arabic valid tool to evaluate social support.
A systematic review published in 2018 by Dambi et al. [
61] investigated the psychometric properties of the non-English translations of the MSPSS found only one Arabic version available (i.e., [
33]). The authors described its methodology as “poor” based on poor internal consistency and validity (no confirmatory factor analysis performed). Dambi et al. [
61] also estimated that this version had “unknown evidence for construct validity” and provided “scanty details” for the adaptation process; which may in turn lead to the risk of misleading findings and negatively affect policy formulation. These potential methodological flaws encouraged our team to translate and validate the MSPSS to the Arabic language, in order to address the identified gaps of the previous Arabic version and provide a psychometrically sound social support scale for the Arabic-speaking researchers, clinicians, patients and the broad community. Our main objective was therefore to examine the psychometric properties of an Arabic translation of the MSPSS in a sample of Arabic-speaking Lebanese adults from the general population. We expect that the Arabic MSPSS will (1) replicate the original three-factor structure; and (2) yield good internal consistency, convergent validity, and measurement invariance across gender groups. We expected to demonstrate measurement invariance at the configural, metric, and scalar levels. Given that strict invariance is very hard to meet, and thus rarely hold [
62]; and since this form of measurement invariance is acknowledged to be overly restrictive [
63], we did not expect to be capable to show invariance at this level. Convergent validity was tested through demonstrating that MSPSS scores correlate to other relevant constructs (here, resilience and posttraumatic growth) in the theoretically expected way [
64]. We chose these correlates because social support has consistently been demonstrated to promote behaviours that improve stress-regulation, such as enhancing resilience and growth [
65]. Indeed, a strong evidence exists supporting that social support is a potential attribute of resilience [
66] and post-traumatic growth [
67,
68]. We thus expect to establish convergent validity of the MSPSS by demonstrating positive correlations between social support and both resilience and post-traumatic growth scores.
Discussion
The purpose of the present study was to rigorously test the psychometric characteristics of the Arabic MSPSS in a non-clinical sample of Lebanese adults; more specifically, to fill the gaps of the first Arabic validation attempt by Merhi et al. [
33,
61]. As hypothesized, the Arabic MSPSS maintained excellent psychometric properties in our sample in terms of factor structure, internal consistency, gender invariance and convergent validity. The findings, therefore, provided strong evidence that the Arabic version of the MSPSS is a suitable self-report measure of social support for use in research and clinical practice in Arab settings.
Our analyses indicated an adequate fit indices for the three-factor model, which is in line with the original development study [
17], and several other previous language adaptations in non-clinical samples [
91‐
93]. Given that perceptions of social networks are shaped by culture [
94‐
96], MSPSS may show inconsistent structural validity across many various cultural settings. For instance, certain previous validation studies, mainly in Asian countries, yielded a unidimensional factor structure (e.g., [
27,
28,
97‐
101]). Other psychometric studies reported a 2-factor structure, with either Family and Significant others [
89] or Friends and Significant others [
90] subscales merging into one factor. Hence the importance of performing CFA in addition to EFA which, when used alone, could result in inappropriate conclusions [
102]. Beyond factor structure, our results demonstrated that the Arabic MSPSS and its subscales have a high internal consistency with McDonald’s ω values between 0.94 and 0.97 [
103]. A good internal consistency of the MSPSS has been consistently supported through high Cronbach’s alphas coefficients in the original English [
17] as well as other language versions [
26,
27,
32,
104,
105]. While prior evidence has shown that Cronbach’s alpha is inappropriate in estimating the internal consistency of multidimensional instruments [
106], and that McDonald’s ω is particularly more advantageous [
106], very few studies used McDonald’s ω when assessing the internal consistency of MSPSS [
41,
107].
Both genders exhibited no significant difference in all MSPSS dimensions. Our analyses also established measurement invariance across gender at the configural, metric, and scalar levels. These findings support the appropriateness of the MSPSS in measuring identical constructs with the same 3-factor structure across genders, and its usefulness in comparing the mean scores of the three social support subscales (Family, friends, and the significant other) between males and females. In the original validation by Zimet et al. [
17], male undergraduates displayed lower perceived support from friends and the significant other, as well as lower total support than females. Other studies documented mixed patterns of gender differences (i.e., greater support in all dimensions among females [
20], more support from family and friends among females [
24], greater support from family among males [
108], no gender difference [
109]). These discrepancies may be explained by variations in socialized gender-roles across societies and cultures; which calls for further cross-cultural research on gender differences in perceived social support to confirm our assumptions. Finally, convergent validity was supported by showing that all three MSPSS sub-scores and total score correlated significantly and positively with resilience and PTG scores. These expected patterns of associations are in agreement with previous studies in which social support has consistently been found to be significantly related to both resilience [
110] and PTG [
111,
112] constructs. Resilience has been suggested to enable more positive perceptions of resources, which in turn acts as a stress buffer [
113,
114]. Similarly, greater interpersonal resources were found to be amongst the positive changes undertaken by individuals after trauma [
115].
Study limitations
First, our sample lacks diversity, comprised exclusively of Arabic-speaking Lebanese people. Future research should consider validating the MSPSS in other Arab countries and communities. Second, we adopted a cross-sectional design and a self-report method to collect our data; longitudinal validation studies are still required to assess predictive validity. Third, while some fit indices did not show adequate values (such as RMSEA value), some psychometric properties were not examined in the context of the present study, such as responsiveness and test-retest reliability, and should be subject of additional research to confirm temporal stability of the MSPSS. The forward and backward translations should have been made by at least two independent translators each time [
116]. The number of participants in the pilot study might be small [
117]. A selection bias might be present because of the snowball sampling technique method used to collect the data and the unknown refusal rate. Information bias might be present since the answers were self-reported by participants and not evaluated by a healthcare professional. Finally, due to the web-based and self-report nature of the data collection, we were not able to exclude respondents who did not understand the questionnaire due to cognitive impairment or other diseases.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.