http://orcid.org/0000-0002-0386-1784
Figures
Abstract
Background
Social support plays an important role in adolescents’ mental health and well-being, and even more so for disaster survivors. To measure the level of social support, one needs an appropriate tool to produce valid and reliable results; therefore, we aimed to measure the invariance across gender groups, and analyze the construct validity and reliability of the Indonesian version of the Multidimensional Scale of Perceived Social Support (MSPSS), a social support measurement tool which was theoretically constructed and has been well validated in many countries with various cultures and backgrounds.
Methods
A school-based assessment was conducted in junior and senior high schools in a post-disaster setting in Yogyakarta Province, Indonesia. We analyzed 299 adolescent survivors of a volcanic eruption, aged 12~18 years who completed a 12-item Indonesian version of the MSPSS.
Results
The factorial validity confirmed the three-factor structure of the scale (Family, Friends, and Significant Others) which met all of the criteria of parameter indices and provided evidence of high internal consistency reliability. The three-level measurement of invariance, which consisted of configural, metric, and scalar invariance, also performed very well across gender groups with our data and corresponded to the recommended parameters. Our composite reliability values were all fine (>0.7) and indicated that the items in the same construct were strongly correlated and reliable.
Citation: Laksmita OD, Chung M-H, Liao Y-M, Chang P-C (2020) Multidimensional Scale of Perceived Social Support in Indonesian adolescent disaster survivors: A psychometric evaluation. PLoS ONE 15(3): e0229958. https://doi.org/10.1371/journal.pone.0229958
Editor: Kwaku Oppong Asante, University of Ghana, GHANA
Received: October 14, 2019; Accepted: February 17, 2020; Published: March 13, 2020
Copyright: © 2020 Laksmita et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the manuscript and its Supporting Information files.
Funding: The author(s) received no specific funding for this work.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Disasters have been reported to have long-term impacts on vulnerable groups, such as children and adolescents [1–3], and they deserve our attention. Vulnerability is indicated by a transition period from childhood to adulthood when they are physically and psychologically growing, and a disaster would definitely make them to more vulnerable. Social support is often disrupted in such populations after a disaster [4–6]. At the same time, social support plays an important role as a buffer against negative effects of disaster-related traumatic events. This protective effect of social support can help them recover from the event [2, 6, 7] and minimizes the potential of developing depression and posttraumatic stress disorder (PTSD) symptoms [8–11]. Therefore, evaluating social support among this population is very important and must be appropriately assessed using a robust, valid, and reliable measurement tool.
The Multidimensional Scale of Perceived Social Support (MSPSS) is a frequently used scale [12] and has been adapted by many studies in multicultural settings and diverse populations [13–16]. It has a three-factor structure which measures Family, Friends, and Significant Others. Social support from these groups is very meaningful for adolescents, and it gives them a sense that support is reliable and available when they need it [13, 17]. It was initially developed in US undergraduate students aged 17~22 years old [13] followed by other studies which confirmed the factorial validity of a three-factor structure for US adolescents and adults [17, 18], Turkish adults [19], Singaporean adults [20], and Chinese adults [21] and of a two-factor structure for Chinese adolescents [16], Hispanic adult immigrants [22], US older adults [23], and South Asian adults [24]. However, testing of the scale’s validity in disaster survivors is lacking, particularly in Indonesian adolescent populations. No psychometric evaluation of the MSPSS for use among Indonesian populations has been conducted. The results of our study can be used to develop social support research and interventions for Indonesian adolescent disaster survivors.
The literature suggests that gender differences exist in social support scales. However, there is a need to ensure the measurement invariance [21, 25]. Thus, it is very important to provide a prerequisite for meaningful comparisons across groups because measurement invariance examines whether populations of different genders interpret a given measure in a conceptually similar manner over several performance dimensions [26]. Measuring social support using a reliable instrument that produces valid research data across different groups of adolescent disaster survivors would be of critical importance in moving this research forward. Therefore, our purpose in this study was to establish valid comparisons of MSPSS item responses and its subscales across specific groups in Indonesia to confirm the construct validity and measurement invariance.
Materials and methods
Study design, setting and participants
A cross-sectional study design was applied in this study, and a stratified sampling method were used to select participants. The median sample size in Structural Equation Modeling is about 200 cases according to reviews of studies in different settings and disciplines [27, 28]; therefore, on September 2017 we invited the participation of 310 students who had experienced the 2010 Mount Merapi eruption, aged 12~18 years, and who study at schools (four senior and three junior high schools) in a post-volcanic eruption setting in Cangkringan District, the district most severely damaged by the eruption in Yogyakarta Province [29]. We distributed a parental consent form to be signed by the parents or guardian at home and also a student assent form to be signed by participants. The following day, participants who returned a completed parental consent and student assent form were then requested to complete a social support scale in their classrooms. Among the 310 participants, 299 of them met the inclusion criteria.
Instrument
Permission to use the MSPSS and its Indonesian version was granted by the original author [13] and translator [30, 31]. The copyright of the original version belongs to the original author [13], and that of the Indonesian version belongs to the translator [30, 31]. A reproduction of the scale was only obtained from those publishers. The MSPSS is a brief 12-item, self-administered measurement tool with three subscales: Family (items 3, 4, 8, and 11), Friends (items 6, 7, 9, and 12), and Significant Others (items 1, 2, 5, and 10). Every item uses a seven-point Likert scale ranging from 1 (very strongly disagree) to 7 (very strongly agree). A higher score indicates greater the social support perceived by an individual; the total possible score ranges 12~84, or it can be scored according to its subscales by adding the items in each subscale and then dividing by 4. Additionally, the scale is very easy to administer and is user friendly for both the administrator and participants [13, 15]. We tested the reading level of the Indonesian version for its face validity in 48 students from seventh to ninth grades, and it was found that all the words used in the scale were understandable.
The original version of the MSPSS has very good internal reliability with an α coefficient of 0.88 for the total scale, 0.87 for the Family subscale, 0.85 for the Friends subscale, and 0.91 for the Significant Others subscale [13]. In addition, its test-retest reliabilities after 2~3 months from the initial data collection were 0.85, 0.75, 0.72, and 0.85 for the Family, Friends, and Significant Others subscales, and overall scale, respectively. The construct validity between the scale and depression scores yielded an inverse correlation with an r value of -0.25.
The Indonesian version was translated using guidelines from Brislin [32] and first validated in 18~65-year-old Indonesian family caregivers of schizophrenia patients. The reliability value of the translated scale was 0.85, and the content validity index for the scale (S-CVI) of each content relevance and clarity was 1 [30, 31].
Data analysis
This study used statistical software of IBM SPSS version 21.0 and IBM SPSS for AMOS version 21.0 to analyze the data. We investigated participants’ demographic characteristics using a descriptive statistical analysis. No missing data were found in the analysis. We then replicated the original three-factor structure of the original MSPSS for use in our present study and examined whether the three-factor structure of the Indonesian version was consistent with the original one. We conducted a confirmatory factor analysis (CFA) because we wanted to assign the three factors (Family, Friends, and Significant Others) as a latent construct and measure each of them using four directly measurable items. This analysis determined the goodness of fit between the observed data and the hypothesized model through performing a first-order CFA with some model fit parameters including the ratio of the Chi-squared value to the degree of freedom (x2/df), root-mean-square error approximation (RMSEA), goodness of fit index (GFI), comparative fit index (CFI), Tucker-Lewis index [33], and standardized root mean square residual (SRMR). An acceptable goodness of fit of the model was determined by x2/df <5, RMSEA <0.08, GFI >0.90, CFI >0.90, TLI >0.90, and SRMR <0.08 as recommended by previous studies [34, 35].
We also investigated more details about the scale’s characteristics by evaluating three levels of the measurement invariance; the same construct was similarly measured across groups, and consisted of configural, metric, and scalar invariances across gender groups through a multi-group (MG)-CFA [36]. The MG-CFA is a very common procedure for measuring equivalence across groups, which are compared in terms of factor analytic parameters such as factor loadings, intercepts, error variances, factor variances, covariances, and factor means [26, 37, 38]. The first level, configural invariance [39], assumes that the same pattern of item-factor loadings exists across the groups being compared. It also suggests that the same items must have nonzero loadings on the same factors. Additionally, configural invariance demonstrates the same number of factors in every group, which then determines the requirement for the other tests. The second level, metric invariance [40], investigates the factor loadings’ invariance across groups. The last, scalar invariance which is the strongest level measurement of invariance [39, 41], examines the item intercepts across groups. However, scalar invariance must meet the assumptions of both configural and metric invariances.
We referred to some references as guides to perform the analysis and indices for tests of invariance. Most studies using the MG-CFA approach compared the fit of a baseline model to the fit of an increasingly constrained model. We employed the indices recommended by Cheung and Rensvold [42] and French and Finch [37] using differences in the CFI (ΔCFI) and the RMSEA (ΔRMSEA) for which values of ≤0.01 indicate strong invariance [37, 42].
In addition, we conducted a reliability test to evaluate the within-scale consistency of the responses to the items of the measure. We used Cronbach’s α (CA) and the composite reliability (CR) as the internal consistency of each factor [43]. CA and the CR are two different calculations, but may prove the same thing, although they might not have the same value. The initial method used to measure the reliability was CA; however, several researchers have argued that it has some limitations [44], i.e., it may over- or underestimate reliability [43, 45], and furthermore, underestimation of the true reliability may become a problem when the test is multidimensional. It might not be appropriate to use CA as an estimate of the reliability of a multidimensional composite scale score, even though the correlation between dimensions is strong [46]. In addition, the assumption of parallelity of CA indicates that all factor loadings and error variances are equally constrained. Hence, the CR is a better choice to measure the reliability due to its ability to draw on standardized regression weights and measurement correlation errors for each item, and a value of >0.7 is acceptable [47–49].
Results
Participant characteristics
In total, 310 questionnaires were distributed, and 299 (96.5%) valid questionnaires were analyzed. Table 1 demonstrates the participants’ characteristics.
Confirmatory factor analysis (CFA)
Results of the CFA indicated that the hypothesized three-factor structure demonstrated a very good fit to the data, because all of the parameters corresponded to the recommended criteria very well. The results indicated that x2/df = 2.468, RMSEA = 0.070, GFI = 0.935, CFI = 0.948, TLI = 0.933, and SRMR = 0.047. Using standardized estimates, our factor loadings ranged 0.49~0.80, and were significant (Fig 1). The results met the recommended criteria from Hooper, Coughlan (34).
Measurement invariance
Configural invariance.
Configural invariance was examined to determine whether the Indonesian version of the MSPSS was best described by a three-factor structure across the gender groups. Configural invariance was good (as evidenced by good model fit measures when freely estimating two groups, i.e., without constraints). The statistical analysis demonstrated that the configural invariance model (M1) fit the data very well with CFI = 0.931, RMSEA = 0.057, and all factor loadings being significant (p<0.05). The two groups were equivalent (Table 2).
Metric invariance.
A metric invariance model (M2) was also investigated to explore the invariance of factor loadings between the two groups. The model described a very good fit with CFI = 0.923 and RMSEA = 0.057. With regard to the ΔCFI and ΔRMSEA, significant changes occurred; ΔCFI = 0.008 and ΔRMSEA = 0.000 (Table 2).
Scalar invariance.
The last measurement invariance, a scalar analysis, was carried out to examine whether the intercepts and factor loadings were equal between the two groups. The model (M3) investigated if the factor loadings and intercepts were all constrained to be equal across groups, and the residual variances were freely estimated. The analytical results describing the data fit very well; CFI = 0.908 and RMSEA = 0.059. ΔCFI and ΔRMSEA values were 0.015 and 0.002, respectively, both of which were acceptable (Table 2).
Reliability
To assess the internal consistency reliability, CA and the CR were calculated for the total Indonesian version of the MSPSS and for each subscale. Results are presented in Table 2 in which coefficients for both the 12 items of the scale and across gender groups were considered acceptable (α >0.70) [50]; hence, results of our study met the recommended criteria. In addition, CR values ranged 0.74~0.83, indicating that the internal consistencies of these values were adequate, because they corresponded well with the suggested level of >0.70 [47, 48]. The literature states that CR values of > 0.70 indicate that items in the same construct are strongly correlated, and that all items are reliable [47] (Table 3).
Discussion
The present study is the first to investigate the psychometric properties of the MSPSS in adolescent survivors of a disaster in Indonesia. In this population, the three-factor structure of the MSPSS showed replicability. The version of the scale translated into Indonesian also demonstrated a very good construct validity and internal consistency reliability; therefore, it supported the factor structure of the original scale [13] and did other adolescent studies, such as of Hong Kong adolescents [51] and of urban, largely African-American adolescents [17]. The results of our study were dissimilar from the results from Chou [16], who found a two-factor structure, although both were tested in Asian adolescent populations. Ours also differed from findings of a two-factor structure from an adult population living in South Asia [24].
Moreover, the MG-CFA of the scale in the present study also demonstrated very good configural, metric, and scalar invariances across gender groups of adolescent disaster survivors. Results of the present study are similar to those of a study conducted by Cheng and Chang [51], which demonstrated that their initial three-level measurement of invariance across gender groups was equally valid for Hong Kong male and female adolescents. Our findings on the measurement invariance were also supported by another study [25], which investigated the MSPSS in samples of undergraduate men and women and reported that both the configural and metric tests were fully invariant, while its metric equivalence was partially invariant. Our study also obtained similar results from a recent study conducted in Chinese parents of children with cerebral palsy in which the MSPSS held the same constructs across gender groups [21]. Our findings described how the structure of the Indonesian version of the MSPSS measures similar constructs and item contents for both male and female adolescent disaster survivors. Therefore, in future studies, males and females can be put together without worrying about gender effects on the structure of the Indonesian version of the MSPSS.
In the present study, the lowest factor loading we obtained in the model was 0.49 (Friends subscale; item 7: ‘I can count on my friends when things go wrong.’). This might have been caused by how our respondents interpreted the item’s content of counting on their friends when they were experiencing a problem. In practice, the higher factor loading indicates that the item or indicator is highly associated with the factor, and vice versa. It connects the factor of theoretical interest with an empirical variable which attempts to measure the factor. Hence, the loading should be conceived as a validity coefficient if the factor indeed reflects the phenomenon of interest. Therefore, item 7 in our study was not that highly correlated with the Friends subscale. In the CFA, one needs to focus on the goodness-of-fit indexes because they reflect the construct validity. Since all of the indexes in our model met the recommended criteria, we did not need to delete any item. Additionally, for practical significance, a factor loading of either 0.45 with a sample size of 150, 0.40 with a sample size of 200, or 0.35 with a sample size of 250 is acceptable [52]. Tabachnick and Fidell [53] following the recommendations of Comrey and Lee [54] also described that a factor loading of 0.32 indicates poor, 0.45 is fair, 0.55 is good, 0.63 is very good, and 0.71 is excellent.
Regarding the reliability analysis, our statistical results supported that every construct in the scale being reliable. This was similar to results of a study of Chinese parents of children with cerebral palsy [21] and supported all three of the subscales achieving very good reliability values. The scale also demonstrated good internal consistency reliability with CR values of >0.7; therefore, the scale is considered a reliable and valid tool to measure social support as perceived by adolescent disaster survivors. Hair and Black [47] described how good CR values indicate that the items are reliable and have high correlations within the same construct. Our findings supported that the scale and its three subscales had excellent internal consistency reliability overall and across gender subgroups. The results are consistent with previous evaluations of the reliability of the MSPSS across diverse samples and are similar to other studies’ findings [17].
In this study, some values of CA and the CR were very close, and even most of the CR values were higher compared to CA. The CR is considered to produce a better estimate of true reliability than CA [55]. Although both of them indicated the consistency of the item measured the proposed construct, we employed both CA and the CR because CA assumes unidimensionality and indicates that the items are equally related to the construct (factor loadings are the same for all items). However, the CR does not assume this because it takes into consideration the factor loadings of every item. The structural equation modeling approach that we used in this study was empirically assessed, and it overcame some of the limiting assumptions of CA [56].
Although our present findings supported the original version of the scale, there were still some limitations that should be noted. First, we invited participation of Indonesian junior and senior high school students who had previously experienced a traumatic event; therefore, we might not be able to generalize whether the same factor structure and findings would be obtained in other adolescent populations, such as in normative or clinical settings. However, our findings contribute to understanding how social support might work and operate among individuals from different cultural backgrounds. Moreover, nonclinical samples can provide descriptive information on the MSPSS, as described [25]. Second, we used a cross-sectional research design; therefore, we were unable to confirm if the results would be consistent in different time periods. However, despite the limitations of the findings, our study contributes to the MSPSS literature. For example, our study employed measurement invariance through a multiple-group CFA which was one component used to determine the score validity evidence and evaluate the construct-irrelevant variance [37]. Building evidence to confirm the theoretical constructs and measurement invariance is important in establishing valid comparisons of MSPSS item responses and its subscales across specific groups [26].
Conclusions
In summary, the present study demonstrated that the Indonesian version of the MSPSS has unique psychometric characteristics, and the original version with a three-factor structure is applicable to Indonesian adolescent disaster survivors. Measuring social support in a specific population may have different expectations, because the nature of social networks and cultural backgrounds also differ [57]. The availability of the Indonesian version of the MSPSS will be very useful in helping academicians and researchers select an ideal social support scale. We conclude by noting that our findings provide strong support for the use of the MSPSS as a multidimensional-construct instrument. We suggest that future investigations using this scale should be conducted to complete the characteristics and add to the variability of analyzing the scale.
Acknowledgments
The authors would like to thank the study participants and school teachers for their cooperation in this study.
References
- 1. Terasaka A, Tachibana Y, Okuyama M, Igarashi T. Post-traumatic stress disorder in children following natural disasters: A systematic review of the long-term follow-up studies. International Journal of Child, Youth, and Family Studies. 2015;6(1):111–33.
- 2. Udwin O, Boyle S, Yule W, Bollton D, O’Ryan D. Risk factors for long-term psychological effects of a disaster experienced in adolescence: Predictors of post traumatic stress disorder. Journal of Child Psychology and Psychiatry. 2000;41:969–79. pmid:11099114
- 3. Furr JM, Corner JS, Edmunds JM, Kendall PC. Supplemental Material for Disasters and Youth: A Meta-Analytic Examination of Posttraumatic Stress. Journal of Consulting and Clinical Psychology. 2010.
- 4. Derivois D, Mérisier GG, Cénat J-M, Castelot V. Symptoms of posttraumatic stress disorder and social support among children and adolescents after the 2010 Haitian earthquake. Journal of Loss and Trauma. 2014;19(3):202–12.
- 5. Rubens SL, Vernberg EM, Felix ED, Canino G. Peer deviance, social support, and symptoms of internalizing disorders among youth exposed to Hurricane Georges. Psychiatry. 2013;76(2):169–81. Epub 2013/05/02. pmid:23631546
- 6. Agustini EN, Asniar I, Matsuo H. The prevalence of long-term post-traumatic stress symptoms among adolescents after the tsunami in Aceh. Journal of Psychiatric and Mental Health Nursing. 2011;18(6):543–9. pmid:21749561.
- 7. Jia Z, Shi L, Duan G, Liu W, Pan X, Chen Y, et al. Traumatic experiences and mental health consequences among child survivors of the 2008 Sichuan earthquake: A community-based follow-up study. BMC Public Health. 2013;13:1–9.
- 8. Banks DM, Weems CF. Family and peer social support and their links to psychological distress among hurricane-exposed minority youth. American Journal of Orthopsychiatry. 2014;84(4):341–52. Epub 2014/07/08. pmid:24999520.
- 9. Chen J, Wu X. Posttraumatic stress symptoms and posttraumatic growth in children and adolescents following an earthquake: A latent transition analysis. Journal of Traumatic Stress. 2017;30(6):583–92. Epub 2017/12/16. pmid:29244905.
- 10. La Greca AM, Silverman WK, Lai B, Jaccard J. Hurricane-related exposure experiences and stressors, other life events, and social support: concurrent and prospective impact on children’s persistent posttraumatic stress symptoms. Journal of Consulting and Clinical Psychology. 2010;78(6):794–805. Epub 2010/10/14. pmid:20939624.
- 11. Ma X, Liu X, Hu X, Qiu C, Wang Y, Huang Y, et al. Risk indicators for post-traumatic stress disorder in adolescents exposed to the 5.12 Wenchuan earthquake in China. Psychiatry Research. 2011;189(3):385–91. Epub 2011/02/08. pmid:21295350.
- 12. Bruwer B, Emsley R, Kidd M, Lochner C, Seedat S. Psychometric properties of the multidimensional scale of perceived social support in youth. Compr Psychiatry. 2008;49:195–201. pmid:18243894
- 13. Zimet GD, Dahlem NW, Zimet SG, Farley GK. The multidimensional scale of perceived social support. Journal of Personality Assessment. 1988;52(1):30–41.
- 14. Dahlem NW, Zimet GD, Walker RR. The multidimensional scale of perceived social support: A confirmation study. J Clin Psychol. 1991;47:756–61. pmid:1757578
- 15. Kazarian SS, McCabe SB. Dimensions of social support in the MSPSS: factorial structure, reliability, and theoretical implications. Journal of Community Psychology. 1991;19(2):150–60.
- 16. Chou K-L. Assessing Chinese adolescents’ social support: the Multidimensional Scale of Perceived Social Support. Personality and Individual Differences. 2000;28(2):299–307.
- 17. Canty-Mitchell J, Zimet GD. Psychometric properties of the Multidimensional Scale of Perceived Social Support in urban adolescents. American Journal of Community Psychology. 2000;28(3):391–400. Epub 2000/08/17. pmid:10945123
- 18. Cecil H, Stanley MA, Carrion PG, Swann A. Psychometric properties of the MSPSS and NOS in psychiatric outpatients. Journal of Clinical Psychology. 1995;51(5):593–602. Epub 1995/09/01. pmid:8801234.
- 19. Eker D, Arkar H, Yaldiz H. Generality of support sources and psychometric properties of a scale of perceived social support in Turkey. Social Psychiatry and Psychiatric Epidemiology. 2000;35(5):228–33. Epub 2000/08/15. pmid:10941998.
- 20. Vaingankar JA, Abdin E, Chong SA. Exploratory and confirmatory factor analyses of the Multidimensional Scale of Perceived Social Support in patients with schizophrenia. Comprehensive Psychiatry. 2012;53(3):286–91. pmid:21632040.
- 21. Wang Y, Wan Q, Huang Z, Huang L, Kong F. Psychometric properties of Multi-Dimensional Scale of Perceived Social Support in Chinese parents of children with cerebral palsy. Frontiers in Psychology. 2017;8:2020. Epub 2017/12/07. pmid:29209254.
- 22. Cobb CL, Xie D. Structure of the Multidimensional Scale of Perceived Social Support for Undocumented Hispanic Immigrants. Hispanic Journal of Behavioral Sciences. 2015;37(2):274–81.
- 23. Stanley MA, Beck JG, Zebb BJ. Psychometric properties of the MSPSS in older adults. Aging & mental health. 1998;2(3):186–93.
- 24. Tonsing K, Zimet GD, Tse S. Assessing social support among South Asians: the Multidimensional Scale of Perceived Social Support. Asian Journal of Psychiatry. 2012;5(2):164–8. Epub 2012/07/21. pmid:22813661.
- 25. Osman A, Lamis DA, Freedenthal S, Gutierrez PM, McNaughton-Cassill M. The multidimensional scale of perceived social support: analyses of internal reliability, measurement invariance, and correlates across gender. Journal of Personality Assessment. 2014;96(1):103–12. pmid:24090236.
- 26. Vandenberg RJ, Lance CE. A review and synthesis of the measurement invariance literature: Suggestions, practices, and recommendations for organizational research. Organizational Research Methods. 2000;3(1):4–70.
- 27.
Kline RB. Principles and Practice of Structural Equation Modeling. Little TD, editor. New York, NY: The Guilford Press; 2016.
- 28. Wolf EJ, Harrington KM, Clark SL, Miller MW. Sample size requirements for structural equation models: An evaluation of power, bias, and solution propriety. Educ Psychol Meas. 2013;76(6):913–34. pmid:25705052
- 29. Warsini S, Mills J, West C, Usher K. Living through a volcanic eruption: Understanding the experience of survivors as a phenomenological existential phenomenon. International Journal of Mental Health Nursing. 2016;25(3):206–13. pmid:26892390.
- 30. Winahyu KM, Hemchayat M, Charoensuk S. Factors affecting quality of life among family caregivers of patients with schizophrenia in Indonesia. Journal of Health Research. 2015;29(Suppl. 1):S77–82.
- 31. Winahyu KM, Hemchayat M, Charoensuk S. The relationships between health status, perceived control of symptoms, caregiver burden, perceived social support and quality of life among family caregivers of patients with schizophrenia in Indonesia. Journal of Prapokklao Hospital Clinical Medical Education Center. 2015;32:44–57.
- 32. Brislin RW. Back-translation for cross-cultural research. Journal of Cross-Cultural Psychology. 1970;1(3):185–216.
- 33.
Cohen S, Gottlieb BH, Underwood LG. Social relationships and health. In: Cohen BHG S., & Underwood L. G., editor. Social Support Measurement and Interventions: A Guide for Health and Social Scientist New York: Oxford University Press; 2000. p. 1–23.
- 34. Hooper D, Coughlan J, Mullen M. Structural Equation Modelling: Guidelines for determining model fit. Electronic Journal of Business Research Methods. 2008;6(1):53–60.
- 35. Lt Hu, Bentler PM. Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling: A Multidisciplinary Journal. 1999;6(1):1–55.
- 36.
Kline RB. Principles and practice of structural equation modeling. New York, USA: Guilford; 2015.
- 37. French BF, Finch WH. Confirmatory factor analytic procedures for the determination of measurement invariance. Structural Equation Modeling: A Multidisciplinary Journal. 2006;13(3):378–402.
- 38. King WC, Miles EW. A quasi-experimental assessment of the effect of computerizing noncognitive paper-and-pencil measurements: A test of measurement equivalence. Journal of Applied Psychology. 1995;80(6):643–51.
- 39. Meredith W. Measurement invariance, factor analysis, and factorial invariance. Psychometrika. 1993;58(4):525–43.
- 40. Horn JL, McArdle JJ. A practical and theoretical guide to measurement invariance in aging research. Experimental Aging Research. 1992;18(3–4):117–44. Epub 1992/01/01. pmid:1459160.
- 41. Bentler PM. Multivariate analysis with latent variables: Causal Modeling. Annual Review of Psychology. 1980;31:419–56.
- 42. Cheung GW, Rensvold RB. Evaluating goodness-of-fit indexes for testing measurement invariance. Structural Equation Modeling. 2002;9(2):233–55.
- 43. Raykov T. Coefficient alpha and composite reliability with interrelated nonhomogeneous items. Applied Psychological Measurement. 1998;22(4):375–85.
- 44. Shook CL, Ketchen DJ, Hult GTM, Kacmar KM. An assessment of the use of structural equation modeling in strategic management research. Strategic Management Journal. 2004;25(4):397–404.
- 45. Raykov T. Scale Reliability, Cronbach’s Coefficient Alpha, and Violations of Essential Tau-Equivalence with Fixed Congeneric Components. Multivariate behavioral research. 1997;32(4):329–53. pmid:26777071
- 46.
Kamata A, Turhan A, Darandari E. Estimating reliability for multidimensional composite scale scores. Annual Meeting of American Educational Research Association; Chicago2003.
- 47.
Hair JF, Black WC, Babin BJ, Anderson RE. Multivariate Data Analysis: A Global Perspective. 7 ed. Prentice Hall: Pearson Education International; 2010.
- 48.
Nunally JC, Bernstein IH. Psychometric Theory. 3 ed. New York: McGraw-Hill; 1994.
- 49. Fornell C, Larcker DF. Evaluating structural equation models with unobservable variables and measurement error. Journal of Marketing Research. 1981;18:39–50.
- 50. Cortina JM. What is coefficient alpha? An examination of theory and applications. Journal of Applied Psychology. 1993;78:98–104.
- 51. Cheng ST, Chan ACM. The Multidimensional Scale of Perceived Social Support: Dimensionality and age and gender differences in adolescents. Personality and Individual Differences. 2004;37(7):1359–69.
- 52.
Hair JF, Tatham RL, Anderson RE, Black W. Multivariate Data Analysis. 5 ed. London: Prentice-Hall; 1998.
- 53.
Tabachnick BG, Fidell LS. Using Multivariate Statistics. 5 ed. New York: Allyn and Bacon/ Pearson Education; 2007.
- 54.
Comrey AL, Lee HB. A First Course in Factor Analysis. 2 ed. Hillsdale, New Jersey: Lawrence Erlbaum Associates; 1992.
- 55. Peterson RA, Kim Y. On the relationship between coefficient alpha and composite reliability. Journal of Applied Psychology. 2013;98(1):194–8. pmid:23127213.
- 56. Raykov T. Bias of coefficient α for dixed congeneric measures with correlated errorrs. Applied Psychological Measurement. 2001;25:69–76.
- 57. Kim HS, Sherman DK, Taylor SE. Culture and social support. Am Psychol. 2008;25:1339–42. pmid:18793039