Construct validity of the Empowerment Scale in patients with schizophrenia

Corresponding Author:

En-Chi Chiu, OTD, PhD
Assistant Professor at the Department of Long-Term Care
National Taipei University of Nursing and Health Science
Taipei 112, Taiwan, ROC
Tel: 886-2-28227101
Fax: 886-2-23891464

Abstract

The Empowerment Scale contains eight domains that are divided into three subscales. The sum score of the three individual subscales represents three different aspects of empowerment (i.e., second-order models), such as personal, interpersonal, and social political aspects. The total score of the eight domains reflect the overall empowerment status (i.e., third-order factor). However, factor structures of the Empowerment Scale have not been evaluated in patients with schizophrenia in Taiwan. Therefore, the main purpose of this study was to examine construct validity (i.e., one 8-factor model [first-order], three second-order models, and one third-order model) of the Empowerment Scale using confirmatory factor analysis (CFA).

Keywords

Empowerment, Schizophrenia, Construct validity

Introduction

“Empowerment” is defined as “both individual determination over one’s own life and democratic participation in the life of one’s community, often through mediating structures” or as “a process by which people, organizations, and communities gain mastery over their affairs” [1]. The concept of empowerment is a process of obtaining control over one’s life and people should have power to influence their personal and social lives [2,3]. Similarly, the concepts of empowerment are applied as guiding principles to facilitate health care promotion and maintenance of high quality health care standards. Some examples are found in the eHealth services that assist patients with diabetes, and cardiovascular diseases [4-6].

Empowerment is one of the main beneficial factors assisting the recovery and achievement of treatment goals for patients with schizophrenia [7,8]. Research on a group of 157 schizophrenia or schizoaffective disorder patients measured the impact of social network, stigma and empowerment on quality of life. Path analysis results found that the recovery program improved patients’ social networks, enhanced empowerment and reduced stigma. This recovery program also reduced depression in patients with psychosis and improved their QOL [9].

In another recent study, Štrkalj-Ivezic and colleagues found that with intervention, all schizophrenia participants showed a positive trend for empowerment and were assisted in recovery [10]. These two research studies provide strong additional evidence that empowerment is an aid to recovery in schizophrenia.

Empowerment enables patients to improve their ability to cope with symptoms and to increase their capacity to have control over their life [11]. Patients gain benefits from empowerment such as increased self-confidence and improved of quality of life [12,13]. Therefore, both clinicians and researchers need to address and accurately measure empowerment issues in order to optimize treatment plans.

The Empowerment Scale formulated in Chinese by Dr. Song conforms to Taiwanese Culture and is designed to measure empowerment in multidimensional constructs [14]. Dr. Song used the concepts of personal, interpersonal, and political dimensions as a base for her Empowerment Scale. She then added the following elements: self-esteem, self-efficacy, power, community action, autonomy, optimism, control of the future, legitimate anger [15], interactive knowledge/skill, selfaffirmation, perceived influence, partnerships, perceived support from others [16,17], and empowerment in Chinese culture [18]. Song consulted with experts and used exploratory factor analysis (EFA) to construct the items and domains of the Empowerment Scale. The resultingEmpowerment Scale consists of 8 subscales and 34 items [14]. The Empowerment Scale has eight domains that profile the patients’ strengths and weakness in all domains. These eight domains were divided into three subscales that represent a second-order factor (Figure 1). The three second-order factors are three different aspects of empowerment (i.e., personal, interpersonal, and social political empowerment). The total score of eight domains reflects overall empowerment status (third-order factor), which is useful to express the patients’ empowerment status in a comprehensive manner.

The Empowerment Scale has been examinedd for construct validity, discriminative validity, internal consistency, and test-retest reliability in healthy population and clients with domestic violence problems in Taiwan [14,19]. However, to our knowledge, no studies have evaluated the construct validity (i.e., first-, second-, third-order factor structures) in patients with schizophrenia using confirmatory factor analysis (CFA), which limits the explanations of the scores of the Empowerment Scale in Taiwan.

The purpose of this study was to apply CFA to evaluate construct validity for the first-, secondand third-order models of the Empowerment Scale for schizophrenic patients. In addition, we examined floor/ceiling effects and internal consistency of each domain in the Empowerment Scale in patients with schizophrenia.

Methods

▪ Participants and ethics

This study used a secondary dataset of schizophrenic patients that was part of “the Examination on the Unity Model of Recoverythe Development of the Stage of Recovery Instrument and the Application of Strengths Perspective (E98010)”located in the Survey Research Data Archive [20]. The Institutional Review Board at Chang Gung Memorial Hospital approved the study.

Data from 339 schizophrenic patients with a mean age of 36.6 years were used in this study. 57.8% were males. The average age of onset of schizophrenia was 24.0 years. 72.6% of the patients had at least a high school education. Additional demographic characteristics are shown in Table 1.

Table 1: Characteristics the patients with schizophrenia (n=339).

The inclusion criteria of recruiting the participants in this study were as follows: (1) diagnosis with schizophrenia; (2) aged over 18 years; and (3) hospitalized at least once for treatment of schizophrenia. We excluded any patient who had also been diagnosed with dementia or substance abuse. The participants were recruited from both past and present in-patients of 24 psychiatric rehabilitation centers. All participants filled out the Empowerment Scale with the assistance from a research assistant.

▪ Instrument

The Empowerment Scale is a questionnaire with 34 items. A 4-point Likert scale is used for each item. Six items were reverse coded, including item 9, 10, 11, 12, 21, and 27. A higher score designates the greater function. The Empowerment Scale contains eight domains as follows: (1) self-efficacy and internal control (8 items); (2) external control (6 items); (3) interpersonal communication skills (3 items); (4) interpersonal assertiveness (3 items); (5) social assertiveness (3 items); (6) social political resources and influence (5 items); (7) social political power (3 items); and (8) social political action (3 items) [14]. The personal subscale includes two domains (i.e., “self-efficacy and internal control” and “external control”). The interpersonal subscale consists of three domains, such as “interpersonal communication skills”, “interpersonal assertiveness”, and “social assertiveness”. The social political subscale contains three domains, such as “social political resources and influence”, “social political power”, and “social political action”.

▪ Data analyses

CFA was performed using the LISREL 8.8 software. We first examined the first-order factor structure of the Empowerment Scale (i.e., 8-factor model). A diagonally weighted least square method was used to estimate CFA parameters because the scale was ordinal [21,22]. We used four goodness-of-fit indices that were considered as a whole to evaluate the level of fit between the overall model and data. The goodness-of-fit is the ratio of the chi-square value to the degrees of freedom (χ²/df), comparative fit index (CFI), Tucker-Lewis index (TLI), and the root mean square error of approximation (RMSEA). The criterion of χ²/df was <2.00 [23]. A value of CFA and TLI values ≥0.95 is recognized as a good fit [24]. A value lower or equal to 0.05 for RMSEA indicated a good fit [25]. Moreover, we examined the factor loadings of all items. Items with low factor loading (<0.40) were removed [26] and the first-order CFA was reconstructed.

After the first-order factor structure was supported, we conducted three second-order models (i.e., personal subscale, interpersonal subscale, and social political subscale). If the three second-order models showed a good model fit, we further verified the third-order model (i.e., empowerment as the three-order factor). The second- and third-order models were also evaluated by the four goodness-of-fit indices (i.e., χ²/df, CFI, TLI, and RMSEA).

After deleting items under validation with CFA, we further examined floor and ceiling effects and internal consistency reliability for each domain. Floor and ceiling effects were calculated by the percentage of participants with lowest and highest scores, respectively. 20% or higher represented a noticeable floor and ceiling effect [27]. Cronbach’s alpha (α) was used for examining internal consistency. The standard for the reliability coefficient was greater or equal to 0.70 [28]. We also examined corrected itemtotal correlation for the deleted items. Corrected item-total correlation is the correlation between an item and sum score of the other items in the corresponding subscale. The standard of corrected item-total correlation is more than 0.3 [29].

Results

The CFA results of the first-order factor structure met the preset criteria of four fit indices (χ²/df=1.13, CFI=1.00, TLI=1.00, and RMSEA=0.019). However, one item (item 27: Social reality cannot be changed by people) showed low factor loading (0.06). We deleted item 27 and reconstructed the first-order CFA and the results then showed a good model fit (Table 2). The retained items (i.e., 33 items) demonstrated sufficient factor loadings (0.63- 0.85) (Table 3).

Table 2: Results of fit indices for the first-, second-, and third-order models.

Table 3: Factor loadings of the first-order model of the Empowerment Scale with 33 items.

Regarding the higher order CFA models of the Empowerment Scale with 33 items (ES-33), the three second-order models showed good model fits (χ²/df=1.09-1.77, CFI=0.99-1.00, TLI=0.99- 1.00, and RMSEA=0.017-0.048) (Table 2). The third-order model also demonstrated a good model fit (χ²/df=1.25, CFI=1.00, TLI=1.00, and RMSEA=0.027).

Table 4 illustrates the floor and ceiling effects and Cronbach’s α of each domain of the ES- 33. No obvious floor and ceiling effects were found in the eight domains (0.3%-3.5% and 1.5%-18.0%, respectively). These domains had sufficient internal consistency (α=0.72-0.90), except for the social assertiveness domain (α=0.68). The ES-33 all items met the criteria of corrected item-total correlation with the exception of item 27 that measured 0.3.

Table 4: Floor/ceiling effects and internal consistency of the Empowerment Scale with 33 items.

Discussion

This study employed CFA to determine the factor structures of the Empowerment Scale in patients with schizophrenia. We examined the 8-factor model (i.e., first-order factor structure) to determine its robustness for further application and validation. Then, we examined the higher order models (i.e., second- and third-order factor structures) to verify appropriateness of using the sum score of each subscale and the total score of the scale. The CFA results displayed that the factor structures of theES-33 were well supported in the first-, second, third-order models.

The 8-factor model of the ES-33 met the criteria of the four CFA fit indices, indicating the ES- 33 is useful to measure multidimensional empowerment in patients with schizophrenia. The ES-33 with eight domains is applicable to express and identify the multiple influences of empowerment in patients with schizophrenia [30]. Item 27 of the Empowerment Scale was deleted because of low factor loading. Corrected item-total correlation of this item was also low. A possible reason could be low consistency of subject emphasis for item 27 with other items in the external control domain. The content of item 27 is whether “general people” could change the current situations. However, the other items focus on feelings and perceptions of “self” for the current situations. The retained 33 items of the Empowerment Scale had sufficient factor loadings, which indicate each item was a suitable indictor of its corresponding domain [31]. For the clinical implication, the ES-33 could be used as patient-reported outcome to reflect patients’ perceptions of multiple influences of empowerment in the process of recovery at clinical and research settings [32].

The higher order models of the ES-33 showed good model fits. For one of the three secondorder models, which have a single secondorder factor (i.e., personal, interpersonal, and social political factors), each CFA model fit the data well. Our results indicate that the sum score of domains for each second-order model can be used to represent its subscale-aspect of empowerment. For example, the scores of two domains (i.e., “self-efficacy and internal control” and “external control”) of the personal subscale can be summed up to reflect the empowerment in the personal aspect of empowerment. Moreover, the CFA results of the third-order model were satisfactory, indicating that a total score of the ES-33 is an appropriate representation of the overall empowerment status in patients with schizophrenia. Clinicians and researchers can apply the total score as an outcome indicator.

The eight domains did not have floor and ceiling effects. These domains can distinguish patients’ domain-specific functions among the low and high score ranges. However, two domains (i.e., “interpersonal communication skills” and “social political power”) had higher percentage of patients within highest score (17.4% and 18.0%, respectively). A possible explanation may be that the sample recruited from past and present rehabilitation center patients who may have already reached a certain level of functioning. The acceptable ceiling effects might restrict the ability of these two domains to distinguish their domain-specific functions. Further studies are needed to recruit patients with a variety of severities to evaluate the ceiling effect of these two domains. Regarding the internal consistency, our results were similar to a previous study [14], except the “interpersonal assertiveness” domain (α>0.7 in this study; α<0.7 in the previous study) and the “social assertiveness” domain (α<0.7 in this study; α>0.7 in the previous study). The reason may be that the characteristics of the sample in this study were different from those in the previous study (e.g., healthy population and clients with domestic violence problems). However, indicators of the psychometric property (e.g. reliability, discrimination, difficulty) are “sample dependent” [33]. A questionnaire needs to be validated in several different samples before the questionnaire can be applied. Our findings provide further evidences of psychometric properties of the ES-33 in patients with schizophrenia.

Two limitations of this study should be noted. First, our data were taken from another’s’ dataset. The original data collectors did not provide any information of the severity of the psychiatric condition or the cognitive level. This in turn may restrict our ability to accurately predict if severity of symptom level of patients affects the benefits of using the ES-33. Further studies are needed to assess these variables, and should be conducted in diverse populations

A second limitation for generalized use of the ES-33 arises from its development using only Taiwanese Chinese speaking patients. Our results from the Taiwanese population may be affected by cultural, linguistic or other factors not present in other countries, especially when compared to Western populations. Further studies could translate the ES-33 into different languages and further examine the factor structures.

Construct validity of the ES-33 was well supported from the first-, second-, and thirdorder CFA models. The sum score of the three individual subscales is appropriate to represent different aspect of empowerment (i.e., personal, interpersonal, and social political aspects). Moreover, the total score of the ES-33 is able to explain the overall empowerment status in patients with schizophrenia. The ES-33 is recommended for use to capture the multiple dimensions of empowerment and overall empowerment status on perspective of patients with schizophrenia.

Acknowledgements

We would like to thank the Survey Research Data Archive (SRDA) at Academic Sinica in Taiwan that provided the dataset for this research. We confirm that no organization funded our` research.

References

1. Rappaport J. Terms of empowerment/exemplars of prevention: Toward a theory for community psychology. Am. J. Community. Psychol 15(2), 121-148 (1987).
2. Wallerstein N, Bernstein E. Empowerment education: Freire's ideas adapted to health education. Health. Educ. Behav 15(4), 379-394(1988).
3. Segal SP, Silverman C, TemkinT. Measuring empowerment in client-run self-help agencies. Community. Ment. Health. J 31(3), 215-227 (1995).
4. World Health Organization. The Jakarta declaration on leading health promotion into the 21st Century. WHO, Geneva (1997).
5. Andersson J, Kjerrman V. Patient empowerment and user experience in eHealth services: A Design-Oriented Study of eHealth Services in Uppsala County Council (2013).
6. Wong CKH, Wong WCW, Wan YF, et al. Patient Empowerment Programme in primary care reduced all-cause mortality and cardiovascular diseases in patients with type 2 diabetes mellitus: a population-based propensity-matched cohort study. Diabetes. Obes. Metab 17(2), 128-135(2015).
7. Warner R. Recovery from schizophrenia and the recovery model. Curr. Opin. Psychiatry 22(4), 374-380 (2009).
8. Jana AK, Ram D, Praharaj SK. Empowerment and its associations in schizophrenia: a cross-sectional study. Community. Ment. Health. J 50(6), 697-701 (2014).
9. Sibitz I, Amering M, Unger A, et al. The impact of the social network, stigma and empowerment on the quality of life in patients with schizophrenia. Eur. Psychiatry 26(1), 28-33(2011).
10. Štrkalj-Ivezic S, Sesar MA, Mužinić L. Effects of a group psychoeducation program on self-stigma, empowerment, and perceived discrimination of persons with schizophrenia. Psychiatr. Danub(2016).
11. Kilian R, Lindenbach I, Löbig U, et al. Indicators of empowerment and disempowerment in the subjective evaluation of the psychiatric treatment process by persons with severe and persistent mental illness: a qualitative and quantitative analysis. Soc. Sci. Med 57(6), 1127-1142 (2003).
12. Linhorst DM, Hamilton G, Young E, et al. Opportunities and barriers to empowering people with severe mental illness through participation in treatment planning. Social. work 47(4), 425-434 (2002).
13. Sibitz M, Amering A, Unger ME, et al. The impact of the social network, stigma and empowerment on the quality of life in patients with schizophrenia. Eur. Psychiatry 26(1), 28-33 (2011).
14. Song LY. The development and validation of an empowerment scale. Soc. Policy. Soc. Work 10(2), 49-86 (2006) .
15. Rogers ES, Chamberlin J, Ellison ML, et al. A consumer-constructed scale to measure empowerment among users of mental health services.Psychiatr. Serv (Washington, D.C.) 48(8), 1042-1047 (1997).
16. Gutierrez LM, Parsons RJ, Cox EO. Empowerment in Social Work Practice: A Sourcebook, Brooks/Cole Publishing Company, Pacific Grove, CA (1998).
17. Miley KK, O’Melia M, DuBois B. Generalist social work practice: An empowering approach. Boston: Allyn & Bacon (2001).
18. Yip K. The empowerment model: a critical reflection of empowerment in Chinese culture. Social. work 49(3), 479-487 (2004).
19. Lin PC. The development and validation of the Empowerment Scale: Short Form. Taiwan. United. Way. Review 3(1), 51-69 (2014).
20. Song LY. The examination on the unity model of recovery - the development of the stage of recovery instrument and the application of strengths perspective (E98010) (2013 ).
21. Forero CG, Maydeu-Olivares A, Gallardo-Pujol D. Factor Analysis with Ordinal Indicators: a Monte Carlo Study Comparing DWLS and ULS Estimation. Struct. Equ. Modeling 16(4), 625-641 (2009).
22. Minrdila D. Maximum Likelihood (ML) and Diagonally Weighted Least Squares (DWLS) Estimation Procedures: A comparison of estimation bias with ordinal and multivariate non-normal data.Int. j. digit. Soc 1(1), 60-66 (2010).
23. Tabachnick BG, Fidell LS. Using Multivariate Statistics. 5th ed., Allyn and Bacon, New York (2007).
24. Hu LT, Bentler PM. Cutoff criteria for fit indexes in covariance structure analysis: conventional criteria versus new alternatives. Struct. Equ. Modeling 6(1), 1-55(1999).
25. MacCallum RC, Browne MW, Sugawara HM. Power analysis and determination of sample size for covariance structure modeling. Psychol. Methods 1(2), 130-149 (1996).
26. Nunnally JC. Psychometric theory. McGraw-Hill, New York (1978).
27. Holmes W, Shea J. Performance of a new, HIV/AIDS-targeted quality of life (HAT-QoL) instrument in asymptomatic seropositive individuals. Qual. Life. Res 6(6), 561-571 (1997).
28. Aaronson N, Alonso J, Burnam A, et al. Assessing health status and quality-of-life instruments: attributes and review criteria.Qual. Life. Res 11(3), 193-205(2002).
29. Knapp TR, Brown JK. Ten measurement commandments that often should be broken. Res. Nurs. Health 18(5), 465-469 (1995).
30. Yamada S, Suzuki K. Application of Empowerment Scale to patients with schizophrenia: Japanese experience. Psychiatry. Clin. Neurosci 61(6), 594-601 (2007).
31. Wang RH, Lin LY, Cheng CP, et al. The psychometric testing of the diabetes health promotion self-care scale. J.Nur. Res 20(2), 122-130 (2012).
32. McCabe R, Saidi M, Priebe S. Patient-reported outcomes in schizophrenia. Br. J. Psychiatry 191(50), S21-28 (2007).
33. Hambleton RK, Van der Linden WJ. Advances in item response theory and applications: An introduction. Appl. Psychol. Meas 6(1), 373-378 (1982).