Background
Nowadays, mental health services in the community are required to develop empowerment and recovery-oriented approaches that challenge traditional structures and processes in mental health practice [
1,
2]. This transformation is intended to facilitate community integration, recovery, and to strengthen the participation, as well as the social and political power of people with mental health issues that use those services [
3‐
6].
Community-based research suggested that community and organizational participation fosters the developmental process of personal, social and civic empowerment [
7]. Research conducted by consumer-run organizations also revealed increased personal empowerment from participating in strength-based challenging initiatives [
8‐
10].
In the mental health field the concept of empowerment was introduced by the mutual help and advocacy movement [
4] and also studied in community psychology [
11]. Definitions of empowerment address the multidimensionality and the multi-level aspects of concept as it comprises individual, social and political components [
12,
13], whether referring to individuals, groups or communities [
14].
Previously published accounts [
15], considered it a continuous process of individual development of personal capacity and of community participation: individuals have awareness, optimism for the future, and confidence about decisions, thus revealing agency, and the ability to contribute to collective goals. Authors such as MA Zimmerman and J Rappaport [
16] or KI Maton and AE Brodsky [
7] viewed empowerment as a principle for action involving participation in groups, increased individual’s control over their life-course and the potential to access and to change community resources.
MA Zimmerman and S Warschausky [
17] reviewed a number of studies in rehabilitation literature revealing increased skills and awareness which improve individuals’ sense of control and participation in other community activities. Those studies supported the idea that the core component of empowerment of exerting control over one’s life is a vital step towards improving individual-level outcomes in rehabilitation. Mental health systems are required to develop opportunities for people’s participation in decision-making and in service policy and evaluation.
Adopting an empowerment approach also implies the development of appropriate and valid outcome measures to evaluate practice in similar mental health settings, accordingly. The Empowerment Scale (
ES) [
18,
19], also known as “Making Decisions Scale”, is one of the few existing empowerment outcome measures in the mental health field and therefore widely used and validated across countries and contexts [
20‐
25] to evaluate the efficacy of interventions.
In Portugal, parallel to being the de-institutionalization objective, the reformed mental health policy (Plano Nacional para a Saúde Mental (PNSM) - 2007–2016) created a new law (DL 8/2010) for the implementation of integrated care in the community context, which reviewed a previous framework (DC 407/98) for psychosocial rehabilitation and community support services in the mental health field. The PNSM states the mental health system must address the need for the development of mental health care in the community, the users’ participation and their involvement in the recovery process and personal achievements (the Portuguese plan can be retrieved from one Health Ministry website at
http://saudemental.pt/wp-content/uploads/2011/02/relatorioplanoaccaoservicossaudemental.pdf).
The empowerment concept is a relevant principle to respond to transformative changes in the mental health system [
26,
27] with implications for both individual and community quality of life [
17]. Based on empirical evidence, community mental health organizations (CMHO) may operate as mediating resources to foster individuals’ empowerment [
28‐
30]. Therefore, the adaptation of the ES to the Portuguese context and language is of relevance in the context of current reform and policy change towards an empowerment and recovery-oriented mental health system.
Development and adaptation of the ES cross-culturally
ES Rogers, J Chamberlin, ML Ellison and T Crean [
18], developed the ES in a participatory study with 261 participants from self-help groups, and established the five-factor structure for the ES: esteem and efficacy, power and powerlessness, optimism and control over the future, community activism and autonomy, and righteous anger. The authors also reported a satisfactory degree of internal consistency (α = .86) for the scale. The ES five-factor solution was further validated [
19] in a study with a large number of participants (
N =1827) from a multi-site consumer-operated services research project. The confirmatory factor analysis identified 3 items to be removed from the ES, in which the model with 28items showed a fair fit, and thus provided a revised 25-item version for the empowerment measure. The study also examined the relation of personal empowerment with personal recovery (
r = .67) and psychiatric symptoms (
r = -.39). The shortened version, which is detailed in the method section in this article, produced better confirmatory fit statistics (
CFI =0.835,
GFI =0.878,
RMSEA = .070, and
NNFI =0.835), and maintained good reliability for the overall scale (α = .82) in terms of internal consistency. However, the subscales scores varied from a modest to an excellent internal consistency (esteem-efficacy, α = .82; power and powerlessness, α = .59; community activism and autonomy, α = .59; optimism and control over future, α = .45; righteous anger, α = .64).
PW Corrigan, D Faber, F Rashid and M Leary [
25], with a group of individuals released from an inpatient service and from a partial hospitalization programme (
N = 35), used an earlier unpublished version of the ES with seven factors to test a model that included two super ordinate factors: the dimensions of self- and community orientation to empowerment. Previous reliability analysis for the subscales showed a low righteous anger (α = .38) and this factor was removed from the analysis. The remaining subscales showed good reliability (α > .75): self-efficacy, powerlessness, self-esteem, optimism and control over the future and group/community action. The Empowerment Scale was also tested with participants (
N =283) from an outpatient public mental health service bySA Wowra and R McCarter [
20] that confirmed its reliability (
α = .85) and its five-factor model. Interestingly, the authors found that respondents with full-time jobs and college experience scored higher in the overall empowerment.
A review of existing literature showed the 28-item version of the Empowerment Scale had also been translated and subjected to psychometric analysis across the counties of Sweden, Japan and the Netherlands [
21‐
24]. L Hansson and T Björkman [
21]highlighted the validity of the ES in the course of a follow-up study among participants from case management services in Sweden (
N = 92) and found a very satisfactory Cronbach’s alpha for the overall scale’s internal consistency (
α = .84), and of .64 to .90 coefficients levels for the subscales except for the power-powerlessness subscale (
α = .45). The study also supported the second-order factor structure proposed in previous research [
25]. The Swedish scale presented significant and positive association with quality of life, size and quality of social network and psychological functioning and associated negatively with psychiatric symptoms, needs for care, and with the negative stigmatizing attitudes.
The Empowerment Scale was also adapted for the Japanese context [
24] and used in a second study with 72 respondents from one mental health day and vocational service [
22] to determine their level of empowerment and to examine the ES results with social adjustment and attitudes towards negative circumstances. In both studies, significant correlations between the factors and the overall score were found, except in the case of righteous anger. Likewise, this subscale had inverse correlations with power and with optimism for the future. S Yamada and K Suzuki [
22] accounted for the significance of the righteous anger subscale in the measure. S Castelein, M van der Gaag, R Bruggeman, JT van Busschbach and D Wiersma [
23], in an outpatient service in the Netherlands (
N = 50) compared the properties of three empowerment measures, including the Empowerment Scale. They reported satisfactory internal consistency for the ES (α = .82) and its sensitivity to the symptom scores.
Aim of the study
Considering the need for the development of outcome measures to evaluate empowering interventions with people who experience mental illness, the current study aims to assess the reliability and validity of the Portuguese version of the Empowerment Scale. Regarding the construct validity, it was also hypothesized that empowerment would be positively correlated with personal recovery and negatively correlated with psychiatric symptoms.
Discussion
To the best of our knowledge, this is the first empirical study to use the 25-item Empowerment Scale short version [
19]. The current study replicated the proposed five-factor model to assess reliability and validity of the Portuguese version with a sample of 213 participants from five representative CMHO psychosocial/community support programmes. Our respondents’ sample represents an important portion of users taking advantage of these services in Portugal. A report of the “Carta Social” (Social Chart) from 2012 showed that 800 people were supported by this type of programme. This data is available at a website from the Social Security Institute (
http://www.cartasocial.pt/pdf/csocial2012.pdf).
Observing the findings from reviewed studies [
10,
12,
25], they suggested the usefulness of empowerment as a psychological construct for people who experience mental illness in diverse clinical and social environments, as well as cultures. Thus, the processes of empowerment are relevant features in current mental health interventions [
1,
28] and throughout worldwide mental health policy reform and transformation [
2,
26]. Therefore, it is important to have construct-related validity of translated measures to facilitate the study and comparison of intervention efficacy across different countries’ mental health systems.
From our factorial validity analysis, in the face of preliminary unsatisfactory fit statistics which were nonetheless not so wrong as to be inadmissible, the ES model required its refinement to achieve a better and reasonable adjustment to the data. The adjusted model yielded a better 20-item solution [
39]. The original validation study from ES Rogers, RO Ralph and MS Salzer [
19], revealed similar quality issues.
Confirmatory factor analysis performance is impacted by circumstances that may affect the validity of outcome measurements such as the hypothesised model, the measurement instrument itself (e.g. number of items per latent factor and its feasibility), the sample size, multivariate normality and the parameter estimates [
38,
40]. Concerning the theoretical model, the ES is a consumer-constructed scale strongly anchored in the mental health advocacy consumer movement [
4,
18] and that background foundation is consistent and relevant for the ongoing transformative changes in mental health systems [
2]. The measure is based on an empowerment definition that incorporates process components such as being hopeful, learning and thinking critically in terms of personal agency and efficacy and decision making, which are psychological-related dimensions; and group/community-oriented dimensions such as the relationship to the institutionalized power, including learning about expressing righteous anger, feeling part of a group, increased capacity to act, and effecting change in one’s community [
4,
9,
14,
15]. Therefore, one may consider that the model under consideration captures essential empowerment domain criteria in the mental health field and in users’ experience.
Current factorial validity of the Portuguese version of ES revealed respect for non-severe violation of multivariate normality and presented reliable parameter estimates [
40]. On the other hand, the refined measurement confirmed two latent factors (optimism and power) with less than three items in the model which maybe is considered an impairment in the performance of an outcome measure [
42]. Also the five excluded items appeared to be theoretically related with personal empowerment developmental processes, excepting just two of them (“Most of the misfortunes in my life were due to bad luck”, and “I feel I have a number of good qualities”) that may not demonstrate an empowerment feature but just general qualities. According to current analysis, though consistent with empowerment theory, the excluded items may reflect a different portion of empowerment processes not sufficiently pertinent to the factors presented in the model [
40,
43,
44].
The present study also found strong to moderate correlations between the overall empowerment scale and its subscales with the exception of the “anger” subscale, which showed no significant association. This non-significant correlation was also mirrored in the “anger” subscale item-total results, underlining the specificity of the variable within the measure. These factorial-related findings are close to what was found in other studies [
19,
21,
22,
25] with few items systematically weighted at latent factors different from the original ones. S Yamada and K Suzuki [
22] also highlighted the significance of the righteous anger subscale when applying the ES cross-culturally, namely how respondents perceive anger behaviours, attitudes and judgements. In our case, questions of how to translate “anger” and, “angry” into the Portuguese language, may be pertinent [
42,
43].
L Hansson and T Björkman [
21], considered that the inherent contradiction in subscale items, some addressing perceived power and others addressing perceived powerlessness may affect the internal consistency of the “power” subscale. In the present study, most items in those subscales presented lower item-total correlations although they still loaded enough in the respective factors and, coincidently most inversely-stated items dropped from the adjusted model. While aiming for response accuracy, that approach may have been somewhat confusing for the respondents [
45].
For the current study with this sample of community mental health users, both circumstances, such as the comprehensiveness of the model of measure in terms of manifest and latent theoretical components across stages of personal empowerment and; the accurateness of the measurement in terms of some items inversely stated, the number of items per factor and the cultural or context specificity of the righteous anger factor, may have affected significantly ES factorial validity estimates [
40,
44].
The exclusion of items is not considered a sufficient reason for model improvement, rather it had the purpose of finding the better adjustment of the model in relation to the data with the current sample of participants [
38,
40,
41]. In order to obtain a stronger ES there is a need for further investigation to improve its less robust aspects. Issues of ES content validity should be substantiated with constituent involvement in qualitative methods. Small group with people who experienced mental illness at different stages and from diverse contexts of participation (
eg. psychosocial, community integration supports and advocacy), are a way of exploring arenas of personal empowerment across mental health system settings. Content validity must also combine
thinking-aloud with
verbal probing techniques for cognitive item evaluation, particularly in the cases of “righteous anger” and “power-powerlessness” items to verify potential issues of lexical accuracy and cultural or contextual-related aspects [
43,
45].
According to ES Rogers, RO Ralph and MS Salzer [
19], empowerment is a construct that can be positively affected by settings characteristics and thus may be a modifiable psychological outcome in mental health. The construct-related validity is, therefore, a continuing process; it cannot be proved definitively [
43]. This empowerment assumption is of relevance to its relation with the values and empowerment-recovery orientation of the current mental health systems as they change worldwide [
2].
Another assumption of empowerment is that an individual does not have to display every quality specified by the definition because it is not an a defining “status” but rather a process of growth and change through participation [
4,
15]. The results from our study demonstrated that the overall mean score for the sample was above the midpoint for the instrument, which indicates a high level of empowerment for the current study participants from community programmes.
In terms of reliability analysis, the refined ES achieved an overall satisfactory internal consistency level, which parallels the Rogers’ study. Consistent with factorial data, current results for the subscales’ internal consistency varied from good to excellent in the esteem-efficacy and activism components but less satisfactory for perceived power, optimism and anger subscales, as reported by Cronbach’s alpha correlation levels.
Likewise the authors of the original study [
19], due to identical subscale reliability issues, proposed solely the use of the overall ES as a valid and reliable measure. The 20-item Portuguese ES also proved reliability for its use as an overall empowerment measurement which permitted the use of the conducted convergent and discriminant validity analysis with the concurrent measures of recovery and symptoms.
Consistent with conceptual and empirical assumptions, empowerment and recovery showed themselves to be strongly associated, as empowerment is considered an important mediator for mental health recovery [
6,
8,
9,
27,
30]. The concurrent analysis for the overall scale confirmed the hypothesized results. Findings indicate that the ES is measuring a defined psychological construct that is qualitatively-related in the same direction with personal recovery; and inversely with manifest psychiatric symptoms. Our results were similar to those from the reviewed studies with the same or equivalent measures [
19,
21,
25].
The current study determined satisfactory reliability of the overall ES for its use in community mental health organizations. Validity was also assessed and ensured by the convergent and discriminant analysis in terms of construct validity, being that the inputs of the factorial analysis highlighted the need for improvements to the model in order to achieve a stronger empowerment measurement in the context of the mental health system.
Strengths and limitations
The current study is, to the best of our knowledge, the only psychometric study using the ES shortened version [
19] with participants from community mental health organizations. While parameter estimates were reliable, with the sample size being an important condition performing structural equation modeling (number of cases per estimated parameter) the ratio of 4.17 needs to be reported here [
38,
46]. Facing the scarcity of empowerment measures in the mental health field, the development of a reliable Portuguese measure of personal empowerment is a fundamental requirement for there to exist empowerment/recovery-oriented measures in the mental health services. The translated equivalent ES also fosters the capacity to compare results on empowerment across different countries. This study also added evidence of the need for future factorial evaluation of the ES scale.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
MFJM and JHO designed the study. MFJM interviewed and collected the data as a member of the research project, performed the statistical analysis and drafted the original paper. MFJM and JHO revised the paper and approved the final version.