Reliability and Validity of the Multidimensional Scale of Perceived Social Support Among Women and Adolescent Girls With Disabilities in Selected Sub-districts of Bangladesh

Background Adequate community-based or societal collaboration and cooperation are considerably important for the overall welfare of women and adolescent girls with disabilities. “The Multidimensional Scale of Perceived Social Support (MSPSS)” has not been evaluated for reliability and validity amid women and adolescent girls with disabilities in the Bangladeshi context. Methods A Bangla-translated form of the MSPSS was constructed, and the survey was conducted among 152 women and adolescent girls with disabilities who were purposefully recruited from Bogura Sadar and Chapainawabganj Sadar sub-districts of Bangladesh. Results The Cronbach's alpha of the entire scale was 0.868, indicating high internal consistency. Cronbach’s alpha for the family sub-scale was 0.763, the friends sub-scale was 0.820, and the significant others scale was 0.776. The composite reliability for the family sub-scale was 0.849677, the friends sub-scale was 0.881248, and the significant others sub-scale was 0.859668. Convergence reliability was established following sub-scale-wise scores. It affirms the consistency of measurements. The content validity score was >0.62, following the Lawshe approach. The three-factor model was adopted during confirmatory factor analysis when the three-factor model run in SPSS Amos (version 21) CFI (comparative fit index) was 0.919. Conclusions In Bangladesh, to the best of our knowledge, our study is initially to calculate the perceived societal assistance of women and adolescent girls with disabilities. We validated the Bangla-translated form of the MSPSS from the Bangladeshi perspective. Researchers and clinicians may rely on our accurate and validated MSPSS translation into Bangla when working with this group. Based on our findings, this study endorses implementing the MSPSS for assessing professed community-based collaboration using the three-factor model, especially among women and adolescent girls with disabilities.


Introduction
Within the complex and ever-evolving societal framework of Bangladesh, the well-being and social integration of marginalized populations, specifically women and adolescent girls grappling with physical disabilities, represent a critical concern.The influence of social support in shaping their resilience and overall quality of life cannot be overstated, as it provides them with a fundamental cornerstone for navigating the intricate challenges associated with disability and societal expectations.The Multidimensional Scale of Perceived Social Support (MSPSS) [1][2][3][4], developed by Gregory Zimet and his team, is a widely used instrument designed to assess an individual's perception of the availability of social support.This scale measures support from three sub-distinct sources: family, friends, and a significant other.Participants respond to items reflecting perceived emotional, informational, and instrumental support from each category.The MSPSS provides a comprehensive understanding of an individual's social support network, offering a nuanced assessment of the various dimensions of support received.Its simplicity and ability to capture diverse sources of support make it a valuable tool in research and clinical settings for evaluating perceived social support across multiple dimensions [1][2][3][4].Nevertheless, the inquiry into the reliability and validity of the MSPSS when explicitly applied to women and adolescent girls with disabilities in selected sub-districts of Bangladesh demands meticulous exploration.This paper will hopefully serve as a gateway to a comprehensive investigation into the dependability (constancy) and reasonableness (effectiveness) of the MSPSS within the distinctive sociocultural landscape of Bangladesh.Recognizing the central responsibility of community-based collaboration and cooperation in enhancing the breaths of womenfolk and adolescent girls with disabilities, this research endeavors to bridge the existing gaps in our realization of the intricate dynamics of societal accomplishments within this demographic.
Our study aspires to unearth the subtleties of social support within this unique context, recognizing the MSPSS as a potential catalyst for improving their overall well-being and fostering their integration into society.Against the backdrop of selected sub-districts in Bangladesh, this research can potentially significantly subsidize the enduring dialogues on disability rights and social equity in the region, ultimately advancing the cause of empowerment and inclusion for women and adolescent girls with disabilities.
Subsequent chapters in this research will thoroughly examine the research methodology employed.Furthermore, we will explore the implications of our findings for policymakers, practitioners, and the wider academic community.This research may have the potential to influence policy decisions, inform social interventions, and guide further research initiatives, all of which play a pivotal role in shaping the future landscape for women and adolescent girls with disabilities in Bangladesh and, potentially, in similar contexts worldwide.

Total participants
One hundred fifty-two women and adolescent girls with disabilities (physical disability 93, partial verbal 24, partial vision 24, partial hearing 11, and aged 11 to 43 years) were included.Physical disability means the person has at least a damaged limb; partial verbal disability means the person can speak but not fluently or clearly like a person without disabilities; partial vision disability means the person's vision is partially impaired; and partial hearing disability means the person's hearing capability has not entirely ceased.Their responses in respective domains in the Washington Group short-set questionnaire were utilized to conduct statistical analysis.

Study places
The study was conducted in two sub-districts of Bangladesh, namely Bogura Sadar and Chapainawabganj Sadar Upazila.The sub-districts have been selected purposefully.Both sub-districts are sub-urban, disasterprone, agriculture-based societies and relatively disadvantaged sub-districts.Additionally, purposive sampling was adopted because the principal investigator (PI) had friends in these sub-districts.These individuals agreed to support PI in collecting data with pleasure.PI is currently a Ph.D. scholar at Hiroshima University without a scholarship.

Study type
The cross-sectional research was performed in April 2023.The "Multi-dimensional Scale of Perceived Social Support" examined recognizes community-based assistance among women and adolescents with disabilities.The reliability and validity process of our utilized tool, the MSPSS English version developed by Zimet et al., Zimet et al., Canty-Mitchell and Zimet, and Dahlem et al. [1][2][3][4], was followed (Table 1).There are a few more similar scales, e.g., the Social Support Scale (SSS) [5], the Online Social Support Scale [6], the Family Support Scale (FSS) [7], the Family Resilience Assessment Scale [8], the Family Connectedness Scale [9], the Perceived Support Network Inventory [10], the Sexual Relationship Power Scale [11], and the Social Support Questionnaire -Short Form (SSQ6) [12].Furthermore, there are different tools to assess social support available at the following link: https://elcentro.sonhs.miami.edu/research/measures-library/socialsupport-relationship-construct/index.html.Nonetheless, in this study, we are concerned about MSPSS only.

Item description
Response level  The translated Bangla MSPSS scale version has been uploaded in appendices and in jpeg format, as shown in Figure 2. The image format has been utilized to facilitate Bengali-speaking authors.The image will retain authentic Bangla font.To the best of our knowledge, this is the first Bangla translation of the MSPSS scale [1][2][3][4].
There is no significant difference in the reliability and validity scores of the MSPSS among women and adolescent girls with disabilities in the selected sub-districts of Bangladesh.The "multi-dimensional scale of perceived social support" [1][2][3][4] is divided into three sub-scales: "family, friends, and significant others" [1][2][3][4].

Ethical considerations
The Institutional Review Board of the National Institute of Preventive and Social Medicine (NIPSOM), Dhaka, Bangladesh, has approved the study protocol, Reference No.: NIPSOM/IRB/2023/07, Dated: February 9, 2023.During recruitment, participants provided their informed written consent.Before receiving informed written permission, participants were briefed about the study objectives, procedures, and measures to ensure privacy and anonymity.They are also well-versed in the fact that they could revoke this research regardless of the survey period.The legal guardian provided informed consent on the participant's behalf when she was under 18.All quotes throughout the paper belong to references [1][2][3][4].

Results
The total number of study participants was 152.They were women and adolescent girls with disabilities from selected subdistricts of Bangladesh.There were no missing values in the dataset.

Language validity
The Bangla-translated scale was given to a group of academicians to assess its cultural appropriateness in Bangla and the level of comprehension of each item.The literature mentioned the content validity criteria as 0.62 [14].Based on these academic experts' opinions and the Lawshe approach, it was concluded that all the scale's items were more than 0.62.No item was removed.
Cronbach's Alpha for Friends Sub-Scale was 0.820.Cronbach's alpha based on standardized items was 0.819.The total number of items was 4: item 06, item 07, item 09, and item 12 (Table 2).Cronbach's alpha for the significant other sub-scale was 0.776.Cronbach's alpha based on standardized items was 0.782.The total number of items was 4: item 01, item 02, item 05, and item 10 (Table 2).The MSPSS English version developed by Zimet and his team during the reliability and validity process was followed [1][2][3][4].

In our study -Cronbach's alpha
Among the 12 items, the high response mean was 5.64 in item 01, and the lowest was 4.74 in item 07.Based on the standard deviation score, the most widely spreading response was found in item 05 (SD 0.868), and the least spreading response was in item 01 (SD 0.615).The total number of responses was 152 (Table 2).The mean score of the 12-item scale was 61.36, the variance was 33.065, and the standard deviation was 5.750 (Table 4).The MSPSS English version developed by Zimet and team was followed during the reliability and validity statistical analysis process [1][2][3][4].

Mean
In the correlation matrix, it was found that items are mostly significantly correlated with each other.The significance level of the total score was 0.000 (two-tailed).The Pearson correlation of the total score was 0.540.Critical values for Pearson's correlation coefficient: 0.3104 (significance level 0.01, df=66).The observed value of the total score is greater than the critical value.It denotes the significant validity of the items (Table 6).

Factor analysis with scale items
Principal component analysis was used as an extraction method.In this factor analysis with 12 items, it was found that only one item can explain the 40.916% variance.That is not satisfactory.When we adopted a higher-order three-factor model, it was found to best fit in the case of factor analysis (Table 7).The Kaiser-Meyer-Olkin (KMO) measure of sampling acceptability or sufficiency [19] for the family sub-scale was 0.770, the friends sub-scale was 0.776, the significant others sub-scale was 0.650, and the total scale was 0.821 (Table 8).The KMO gauge of sampling ampleness is a statistic that shows the proportion of variance in research variables that might be caused by rudimentary features.High values (around 1.0) denote that factor analysis has the benefit of interpreting this research data.If the value is below 0.50, the results of the factor analysis may not be very beneficial [19,20].Bartlett's test of sphericity was utilized.Maurice Stevenson Bartlett, an English statistician, first developed this test [21,22].Bartlett's test of sphericity is cast off to analyze the null proposition that the correspondence matrix is a uniqueness matrix.An identity association matrix interprets surveyed variable stars as orthogonal, discrete, and imperfect for factor analysis [23].

Initial eigenvalues Extraction sums of squared loadings Rotation sums of squared loadings
Additionally, Bartlett's sphericity tests revealed that this dataset is seemingly for factor analyses [24].Thus, Bartlett's test of sphericity assesses whether the correlation matrix of a set of variables helps determine if there are significant relationships among the variables, validating the appropriateness of conducting factor analysis on the dataset.The chi-square for the family sub-scale was 144.543, the friends sub-scale was 219.797, the significant others sub-scale was 215.760, and the total scale was 800.261 (Table 8).

Factor analysis with sub-scales
When factor analysis runs with a family sub-scale, it was found that component 01 (item 03) has an eigenvalue of more than one and can explain 58.659% of the variance (Table 9).In the factor analysis, when running with the "friends' sub-scale," it was found that component 01 (item 06) has an Eigenvalue of more than one and can explain 65.105% variance (Table 10).In the factor analysis, when running with significant other sub-scales, it was found that component 01 (item 01) has an Eigenvalue of more than one and can explain 60.520% variance (Table 11).

FL
In the current study, a three-factor model was adopted.When the factor model was run using IBM SPSS Amos version 21, (IBM Corp. Released 2012, IBM SPSS Statistics for Windows, Version 21.0 (Armonk, NY: IBM Corp.), CMIN (chi-square value) was 111.576,DF (degree of freedom) was 50, CMIN/DF was 2.232, CFI (comparative fit index) was 0.919, and SRMR (standardized root mean squared residual) was 0.068.Figure 1 illustrates a three-factor model analysis.

Descriptive statistics
In our study, the highest response mean among the 12 items was 5.64 in item 01, and the nethermost was 4.74 in item 07.Based on the standard deviation score, the most widely spreading response was found in item 05 (SD 0.868), and the least spreading response was in item 01 (SD 0.615).The total number of responses was 152.
The study conducted by Ebrahim and Alothman [25] in Saudi Arabia revealed that among the 12 items, the high response mean was 3.89 in item 03 ("My family really tries to help me") [1][2][3][4].The lowest mean was 3.14 in item 12 ("I can talk about my problems with my friends") [1][2][3][4].Based on the standard deviation score, the most widely spreading response was found in item 12 (SD 1.37), and the least spreading response was in item 03 (SD 1.1).The total number of responses was 606 [25].
The study conducted by Zimet and his team (1998, 1990, 2000, and 1991) found that among the 12 items, the high response mean was 6.22 in item 03, and the lowest mean was 5.38 in item 08 [1][2][3][4].Based on the standard deviation score, the most widely spreading response was found in item 08 (SD 1.51), and the least spreading response was in item 09 (SD 1.01).The total number of responses was 275 [1][2][3][4].
Wongpakaran et al. [28] tested MSPSS in Thailand.Among the student group full scale, the Cronbach's alpha was 0.91, and among recruited patients with major depressive disorders, it was 0.87.They tested the scale among 310 medical pupils from Chiang Mai University and 152 individuals with psychological, emotive, and social ailments [28].
Sharif et al. [29] tested MSPSS in Pakistan.The scale was tested among 1154 pregnant women (older than or equal to 18 years).The MSPSS scale showed excellent internal consistency.The total scale's Cronbach's alpha level was 0.933.Confirmatory factor analysis retained the three-factor model [19].
One more study conducted among breast cancer survivors by Kim et al. in Korea reported that the Koreantranslated MSPSS scale had a Cronbach's alpha score of 0.90.This research also stated the internal consistency was 0.91 for the complete instrument, and the three realms considering partner, household, and acquaintances were 0.96, 0.90, and 0.90, respectively [30].

Convergent validity
The research performed by Wang et al. [21] found that the three-factor model showed the best fit in their analysis.They found that composite reliability for the "family sub-scale" was 0.87, for the "friends' subscale" was 0.88, and for the "significant other's support sub-scale" was 0.89 [1][2][3][4].This survey was conducted among 487 Chinese paternities of offspring with cerebral palsy [31].
One multicenter study was conducted in Spain among cancer patients (995) from 13 diverse hospitals.The Spanish-language MSPSS instrument had outstanding projected reliability, with scores surpassing 0.90 [23].Subsequently, the researchers concluded that the Spanish-transformed form of the MSPSS was an acceptable and dependable scale to appraise perceived community assistance among cancer fighters [33].
Our study ran a convergent validity test after adopting a higher-order three-factor model.The composite reliability found for the "family sub-scale" was 0.849677, the "friends sub-scale" was 0.881248, and the "significant others sub-scale" was 0.859668 [1][2][3][4].

Limitations of this study
One limitation is that the study focuses only on women and girls with disabilities in selected sub-districts of Bangladesh, limiting the ability to generalize the findings to the broader population of women and adolescent girls with disabilities.There is a chance of social desirability bias in the study.To reduce the limitations in the future based on available funding, more study sites with larger sample sizes will be recruited.A pleasant environment will be ensured to minimize social desirability bias, so female counterparts can respond freely without feeling coercion.

Conclusions
This study provides valuable insights into measuring social support among this demographic.The findings offer a preliminary understanding of the perceived social support levels, which can be a stepping stone for further research and the development of targeted interventions.To strengthen the validity and reliability of the scale in this context, future research should aim for more extensive and diverse samples, consider cultural nuances, and employ longitudinal approaches to track changes over time.Overall, this study lays the foundation for a more comprehensive understanding of social support among women and adolescent girls with disabilities in Bangladesh.

FIGURE 2 :
FIGURE 2: The multidimensional scale of perceived social supportBangla version by the principal investigator and his teamWe follow the exact scale scores as in the original paper.

TABLE 4 : Scale statistics
Notes: Test: Reliability analysis; Model: Two ways mixed; Type: Consistency; Confidence interval: 95%.Total item number in full (total scale): 12, Subscales item number: 4.The item-total statistics table found that overall, Cronbach's alpha level decreased if any item was deleted (Table5).

mean if item deleted Scale variance if item deleted Corrected item- total correlation Squared multiple correlation Cronbach's alpha if item deleted Item
1: "There is a special person who is around when I am in need."