Systematic review of empowerment measures in health promotion

Empowerment, a multi-level construct comprising individual, community and organizational domains, is a fundamental value and goal in health promotion. While a range of scales have been developed for the measurement of empowerment, the qualities of these have not been rigorously assessed. The aim of this study was to evaluate the measurement properties of quantitative empowerment scales and their applicability in health promotion programs. A systematic review following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines was done to evaluate empowerment scales across three dimensions: item development, reliability and validity. This was followed by assessment of measurement properties using a ratings scale with criteria addressing an a priori explicit theoretical framework, assessment of content validity, internal consistency and factor analysis to test structural validity. Of the 20 studies included in this review, only 8 (40%) used literature reviews, expert panels and empirical studies to develop scale items and 9 (45%) of studies fulfilled ≥5 criteria on the ratings scale. Two studies (10%) measured community empowerment and one study measured organizational empowerment, the rest (85%) measured individual empowerment. This review highlights important gaps in the measurement of community and organizational domains of empowerment using quantitative scales. A priority for future empowerment research is to investigate and explore approaches such as mixed methods to enable adequate measurement of empowerment across all three domains. This would help health promotion practitioners to effectively measure empowerment as a driver of change and an outcome in health promotion programs.

empowerment, measurement, psychometric properties, health promotion

INTRODUCTION

Empowerment is recognized by the World Health Organization and health agencies around the world as a core concept in health promotion and integral to the achievement of social equity. Rappaport has articulated the relationship between empowerment and equity as follows: ‘To be committed to an empowerment social agenda and to be consistent with that agenda in one's approach to social science theory, research, and action is to be committed to identifying, facilitating, or creating contexts in which heretofore silent and isolated people, those who are ‘outsiders’ in various settings, organizations, and communities, gain understanding, voice and influence over decisions that affect their lives' [(Rappaport, 1981), p. 52]. However, while the central place that empowerment holds in health promotion is not contested, there are few studies that have measured empowerment as a process element or impact of health promotion programs, which has implications for evidence and theory building, and policy advocacy.

Use of the concept of empowerment across diverse fields, including social science, community development, community psychology and economics, has resulted in this taking on a range of meanings. Terms in wide use include psychological, economic, social, community and political empowerment. To date, theory development and testing has been most extensive in the area of psychological empowerment. Zimmerman has described psychological empowerment as a process of change that involves intrapersonal, interactional and behavioural components (Zimmerman and Rappaport, 1988; Zimmerman, 1990, 1995) and has applied this to promote healthy behaviour among at-risk populations (Perkins and Zimmerman, 1995). Others, such as Holden et al. (Holden et al., 2005), have operationalized psychological empowerment to improve program participation and delivery.

While important, psychological empowerment represents just one dimension of this concept, and programs that address this level only are at risk of neglecting social and political factors influencing health equity, that have a large bearing on the sustainability of health outcomes (Rappaport, 1987; Chavis and Wandersman, 1990; Fawcett et al., 1995; Kim et al., 1998). Organizational empowerment involves equipping individuals to exert control in achieving organizational effectiveness in service delivery and policy development (Zimmerman, 2000). Community empowerment refers to processes of interaction between individuals and organizations to enhance community living, thereby effecting changes in a larger social system. These levels of empowerment—psychological, organizational and community—are intimately linked and changes processed at one level have implications for the other levels. Furthermore, the interactions across these levels are culturally and contextually defined; therefore, processes of empowerment are likely to vary according to the community, organization or society where it is being operationalized (Rappaport, 1987).

Wallerstein (Wallerstein, 2006) has described empowerment as one of the prerequisites for health. Programs funded by the World Bank, USAID and WHO aim to build empowerment among youth, women and marginalized groups in communities and have applied community development, capacity building and policy change methods to achieve this (Amin et al., 1998; Romero et al., 2006; Wallerstein, 2006; World Bank, 2013a,b). Wiggins et al. (Wiggins, 2011; Wiggins et al., 2013) showed that popular education, also known as Freirian and empowerment education, is an effective method for improving empowerment among community members and health workers in health interventions. Studies show that initiatives to achieve empowerment can lead to health-related outcomes in a range of social and cultural contexts (Israel et al., 1994; Binka et al., 1995; Akey et al., 2000; Tsay Hung, 2004; Wallerstein, 2006). Laverack (Laverack, 2006) has demonstrated the link between empowerment and health outcomes using the ‘empowerment domains’ approach that clearly shows the influence of each of these domains on health outcomes. These outcomes include enhanced personal and coping skills, more effective use of health services (Dixon et al., 2001; Melnyk et al., 2004), reduced disparities in access to resources and improved implementation of public health policies (Binka et al., 1995; Rich et al., 1995).

There is scope to improve knowledge about the role of empowerment in community participation, capacity building and health improvement through the development of instruments that can measure these constructs in health promotion research and evaluation. These quantitative measures can complement the insights that can be gained from qualitative investigation of empowerment processes and outcomes, and can be used in evaluating the magnitude of project effects, assessing the relative impacts of different strategies and exploring the mediators and moderators of empowerment. Herbert et al. (Herbert et al., 2009) found that although there has been extensive literature on empowerment in the last 15 years, there is a dearth of studies reporting on the properties of scales for measuring empowerment. Furthermore, several studies have tested instruments measuring empowerment in relation to management of selected diseases, so their applicability to the wider population is limited. For example, Bakker and Van Brakel (Bakker and Van Brakel, 2012) reported on the measurement properties of 17 empowerment tools for use among people with disabilities in developing countries.

While empowerment has been extensively studied across these three domains, health promotion practitioners often view health programs as adopting either a ‘top–down approach’ wherein health agencies design programs with improvement in a particular health behaviour as the outcome, or a ‘bottom–up’ approach designed to improve power among groups or individuals as the outcome (Laverack and Labonte, 2000). Although a range of tools measuring individual empowerment as a process or outcome of health programs exist, the measurement of community empowerment entails greater complexity and challenges (Laverack and Wallerstein, 2001). Further, measuring these domains can be easier as a process, than as an outcome of health programs (Laverack, 1999).

Hence, greater scrutiny of the theoretical clarity, reliability and validity of empowerment tools is required to identify and address important questions on the role of empowerment in health promotion. It is also necessary to determine the potential for these instruments to be used in different population groups and settings. The purpose of this study, therefore, was to systematically review the measurement properties of quantitative empowerment tools. This has entailed investigation of the process of instrument development, the constructs that are measured, the population groups with whom testing has been undertaken, and statistical analysis of reliability and validity. The longer term purpose of this study is to facilitate research, evaluation and theory development concerning empowerment in health promotion.

METHODS

Search strategy

A systematic review has been carried out to evaluate the measurement properties of empowerment scales used in health programs. The conduct of the review adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (Liberati et al., 2009). Based on these guidelines, a comprehensive search of the following computerized bibliographic databases was conducted: MEDLINE, CINAHL, PsycINFO, IBSS, Pubmed and Embase covering the period from January 1990 to December 2012. The start date was chosen as 1990, because the literature on health-related empowerment has been on the rise since early 1990s (Herbert et al., 2009). Using relevant MeSH words or sub-headings, the following combination of key words was used for our search:

(Empowerment)

AND

(Questionnaire* OR tool* OR scale* OR survey*)

AND

(reliability OR validity)

The retrieved articles from each of the databases were imported into an Endnote library. The bibliographical references of retrieved articles were manually searched, complemented by citation tracking using Web of Science databases and Google scholar, to identify additional relevant studies.

Table 1:

Characteristic of the scales used in the review

Author, year	Country	Measurement aim	Study characteristics	No. of scale items	Methods of administration	Domains of empowerment	EFA and/or CFA	Internal consistency Alpha range
Akey et al. (2000)	USA	To investigate the measurement structure of the Psychological Empowerment Scale (PES) including the internal factor structure of the scale and the reliability of its scores.	N = 293 parent respondents of children with a disability. White: 89.1% Hispanic:3.9 Black: 3.6% Asian: 2.0% Native American: 1.4% Mean age: 35.42 years 9.9% male:	32	Self-administered	4 domains a) Attitudes of control and competence b) Cognitive appraisals of critical skills and knowledge c) Formal participation in organizations d) Informal participation in social systems and relationships	EFA CFA	Subscales 0.91–0.94
Bann et al. (2010)	USA	To develop and psychometrically evaluate scales to measure patients’ perceptions of provider support, patient-centered care, and empowerment in CAM (Complementary and Alternative Medicine) users	N = 216 patients 14.4% male Median age ≥21 years	5	Self-administered	1 domain	EFA CFA	Total Scale 0.85
Brookings & Bolton (2000)	USA	To validate the Personal Opinions Questionnaire(POQ) that measures 4 components of intrapersonal empowerment for people with disabilities	N = 473 53% female Median age:28 years	64	Self-administered	4 domains a) Personal competence b) Group orientation c) Self-determination d) Positive identity	EFA CFA	Subscales 0.73–0.88
Frans (1993)	USA	To describe development and validation of a scale designed to measure social workers’ perceptions of personal and professional power. (Social Worker Empowerment)	N = 520 social workers 74.5% female Mean age: 42 years	34	Self-administered	5 domains a) Collective identity b) Knowledge and skills c) Self-concept d) Critical awareness e) Propensity to act	EFA	Subscales 0.71–0.89
Gagnon et al. (2006)	Canada	To develop and validate the Health Care Empowerment Questionnaire(HCEQ)	N = 873 elderly (>75 years) Mean age: men 81.1 years women 82.4 years 61.4% female	10	Self-administered	3 domains a) Involvement in decision b) Degree of control c) Involvement in interactions	EFA CFA	Subscales 0.79–0.89
Haswell et al. (2010)	Australia	To develop and validate the Growth and Empowerment Measure (GEM) for Aboriginal Australians	N = 184 Indigenous Australians:100% 64.1% male Mean age: 39.9 years	26	Interview-administered	4 domains a) Emotional empowerment b) a) Inner peace c) b) Self-capacity Functional empowerment c) Healing d) Connection	EFA	Subscales 0.75–0.87
Holden et al. (2005)	USA	To describe the domains and attributes of psychological empowerment as an outcome of youth involvement in community-based tobacco control initiatives	N = 2,059 66.9% girls Mean age : 15.4 years White: 60.7% Other: 39.3%	15	Self-administered	6 domains a) Domain-specific efficacy b) Perceived socio-political control c) Participatory competence d) Knowledge of resources e) Assertiveness f) Advocacy	EFA CFA	Not reported
Johnson et al. (2005)	USA	To investigate a new measure of personal empowerment in women, the Personal Progress Scale-Revised (PPS-R).	N = 222 women Mean age: 25.15 years	28	Self-administered	7 domains a) Perceptions of Power and Competence b) Self-Nurturance and Resource Access c) Awareness of Cultural Discrimination d) Expression of Anger and Confrontation e) Autonomy f) Personal Strength and Social Activism	EFA	Total scale 0.88
Kasmel et al. (2011)	Estonia	To investigate the dimensions of Individual Community Related Empowerment (ICRE) scale and evaluate its measurement properties	N = 120 35% male Mean age: 43 years	20	Self-administered	5 domains a) Self-efficacy b) Intention c) Participation d) Motivation e) Critical awareness	EFA	Total Scale 0.86 Subscales 0.69–0.88
Koren et al. (1992)	USA	To describe the development and empirical examination of the Family Empowerment Scale for assessing empowerment in families whose children have emotional disabilities	N = 440 parents who reported having children under the age of 21 94% female Mean age: 40 years	34	Self-administered	3 domains a) Family b) Service System c) Community/Political	EFA	Subscales 0.87–0.88
Peterson et al. (2006)	USA	To evaluate the internal reliability and construct validity of a revised version of the Socio-Political Control Scale- Revised(SPCS-R)	N = 750 residents 58% female 39% 18–34 years; 22% 35–44 years; 31% 45–64 years; 10% 65+ years 46% African American 29% Hispanic	17	Interviewer-administered	2 domains a) Leadership competence b) Policy control	CFA	Subscales 0.78–0.81
Peterson et al. (2011)	USA	To investigate internal reliability and construct validity of a Sociopolitical Control Scale for Youth(SPCS-Y)	N = 865 students 60% female 55% Hispanic 37% African American 16% in 9th grade, 23% in 10th grade, 29% in 11th grade, 33% in 12th grade	17	Self-administered	2 domains a) Leadership competence b) Policy control	CFA	Total Scale 0.89 subscales 0.81 and 0.85
Rissel et al. (1996)	USA	To evaluate the internal reliability, construct and concurrent validity of General empowerment and Alcohol specific empowerment scales	N = 160 University employees 75% female 30% 18–29 years 27% 30–39 years 23% 40–49 years 9% 50+ years	General – 45 Alcohol specific - 26	Self-administered	4 domains a) Personal efficacy b) Role of group support and action c) Critical consciousness d) Support of social action	Neither	General 0.84 Subscales 0.57–0.87 Alcohol 0.78 Subscales 0.50–0.67
Rogers et al. (1997)	USA	To investigate the internal reliability, construct and predictive validity of an Empowerment Scale for users of mental health services	N = 271 members of self-help groups for people with mental illness	28	Self-administered	5 domains a) Self-esteem self-efficacy b) Power-powerlessness c) Community activism d) Optimism and control over the future e) Righteous anger	EFA	Total scale 0.86
Rogers et al. (2010)	US	To investigate the construct validity of an empowerment scale for people with mental illness	N = 1827 people with mental illness participating in consumer-operated programs 60% female Mean age: 43 years 57% White; 17% African American; 26% Hispanic or other	28	Interviewer administered	5 domains a) Self-esteem self-efficacy b) Power-powerlessness c) Community activism d) Optimism and control over the future e) Righteous anger	EFA, CFA	Total scale 0.82 Subscales 0.45–0.82
Segal et al. (1995)	US	To evaluate the internal and test-retest reliability of 3 empowerment constructs Personal Empowerment Scale, Organizational Empowerment Scale and Extra-organisational Empowerment Scale in persons with mental disabilities	N = 310 people with mental illness participating in self-help agencies	Not stated	Interviewer administered	3 domains a) Personal empowerment b) Organisational empowerment c) Extra-organisational empowerment	Neither	Subscales 0.73–0.87
Speer (2000)	US	To evaluate the content and construct validity of an empowerment scale for use in community organising contexts	N = 974 residents 60% female 60% aged 18–45 years 85% white	27	Interviewer administered	6 domains a) Power through relationships b) Political functioning c) Shaping ideology d) Perceived leadership confidence e) Political efficacy f) Behavioural empowerment	EFA	Subscales 0.47–0.78
Zimmerman and Rappaport (1988)	US	To investigate the convergent and discriminate validity of psychological empowerment measures	Two study samples: 392 undergraduate students 49% males 88% white Mean age 18.95 years 205 community residents mean age 41.9 years 92% white	Not given	Self-administered	11 domains a) Internal political efficacy b) External political efficacy c) Mastery d) Self-efficacy e) Perceived competence f) Desire for control g) Civic duty h) Control ideology i) Chance control j) Internal control k) Powerful others	Neither	Not given
Zimmerman and Zahniser (1991)	US	To investigate the internal reliability and criterion validity of the Socio-Political Control scale	390 undergraduate students (study 1) 88% white Even gender distribution 205 community residents(study 2) 55% female mean age 41.9 years 92% white 143 church goers(study 3) 52% female 96% white	17	Self-administered	2 domains a) Leadership competence b) Policy control	EFA	Subscales 0.75–0.78
Zimmerman et al. (1992)	US	To investigate the internal reliability and construct validity of an intrapersonal empowerment measure	N = 916 61% female Mean age 44 yrs 48% white 47% African American	18	Interviewer administered	3 domains a) Personal and community control b) Perceived effectiveness c) Perceived difficulty	Neither	Subscales 0.68–0.79

Inclusion and exclusion criteria

Studies included were those that focused on the development of an empowerment scale or the adaptation of an existing scale, and reported on the results of reliability or validity testing. Excluded were studies published in a language other than English, books, reports, dissertations and non-peer-reviewed materials as well as those focusing on empowerment for disease management or the performance of narrowly defined roles (e.g. employment responsibilities).

Data extraction and quality assessment

Data were extracted according to the PRISMA guidelines for reporting systematic reviews. All searches were stored using an EndNote library. All potentially relevant studies were screened by one of the reviewers (S.C.). The initial screening of articles was done by reading the titles and abstracts. In the final inclusion/exclusion phase, papers retained for inclusion in the preliminary phase were independently reviewed the by three authors (S.C., A.R. and B.S.). The assessment of measurement properties and the methods used in the development of each tool were independently performed by the three authors (S.C., A.R. and B.S.). Following extraction, the characteristics of studies were recorded, including country of origin, sample size, number of items in the scale, method of data collection, factors extracted and reliability scores, as shown in Table 1.

The methodological quality of the scales was evaluated across three dimensions: item development, reliability, and validity (Table 2). Item development was evaluated by determining whether information from a literature review, empirical study or panel of experts, was used in instrument development. Reliability assessment addressed whether internal consistency and test–retest reliability were reported. Validity was assessed by examining the methods used to determine content validity (if items measure the constructs of interest), structural validity (degree to which the scores on the scales reflect the dimensionality of the construct), internal construct validity (if relationships between scales are consistent with the hypothesis) and external construct validity (whether scales converge with and discriminate scores on other measures in the hypothesized way) (Mokkink et al., 2010) (Table 2).

Table 2:

Methods adopted in the development of the scales included in the review (marked as ✓or x)

Study	Item development			Reliability		Validity
Author, year	Panel of experts	Literature review	Empirical study	Internal consistency	Test–retest reliability	Content validity	Structural	Internal construct validity	External construct validity
Akey et al. (2000)	x	✓	✓	✓	x	x	✓	✓	✓
Bann et al. (2010)	✓	✓	✓	✓	x	✓	✓	✓	✓
Brookings and Bolton (2000)	x	✓	✓	✓	x	x	✓	✓	✓
Frans (1993)	✓	✓	✓	✓	x	✓	✓	✓	✓
Gagnon et al. (2006)	✓	✓	✓	✓	✓	✓	✓	✓	x
Haswell et al. (2010)	✓	✓	✓	✓	x	✓	✓	✓	✓
Holden et al. (2005)	✓	✓	✓	✓	x	✓	✓	✓	x
Johnson et al. (2005)	x	✓	✓	✓	x	✓	x	✓	✓
Kasmel and Tanggaard (2011)	✓	✓	✓	✓	x	✓	x	✓	x
Koren et al. (1992)	✓	✓	✓	✓	✓	✓	x	✓	✓
Peterson et al. (2006)	x	✓	x	✓	x	x	✓	✓	✓
Peterson et al. (2011)	x	✓	x	✓	x	x	✓	✓	✓
Rissel et al. (1996)	✓	✓	x	✓	x	x	✓	x	✓
Rogers et al. (1997)	✓	✓	✓	✓	x	x	✓	x	✓
Rogers et al. (2010)	x	✓	x	✓	x	x	✓	✓	✓
Segal et al. (1995)	x	✓	x	✓	✓	x	✓	x	✓
Speer and Peterson (2000)	x	✓	x	✓	x	✓	✓	x	✓
Zimmerman and Rappaport (1988)	x	✓	x	x	x	x	✓	x	✓
Zimmerman and Zahniser (1991)	x	✓	✓	✓	x	x	✓	x	✓
Zimmerman et al. (1992)	x	✓	x	✓	x	✓	✓	x	✓

Quality assessment of the measurement properties of the scales used in the review was done using a ratings scale (Table 3) previously used in a systematic review of urbanicity scales (Cyril et al., 2013). Scales were rated using the following six criteria; using an a priori explicit theoretical framework, assessment of content validity, internal reliability scores (α >0.7), exploratory factor analysis to test structural validity, confirmatory factor analyses to determine internal construct validity, and assessment of external construct validity. Scores ranged from 0 to 6: 0 if none of the above 6 criteria were fulfilled and 6 if all 6 criteria were fulfilled. The interpretation of scores regarding the quality of scales used in the studies was ≤2 = poor quality; 3–4 = medium quality; ≥5 = high quality.

Table 3:

Ratings for each of the scales included in the review (1 if done and 0 if not done)

Author, year	Followed an a priori explicit theoretical framework	Reported efforts towards content validation	Exploratory factor analysis	Confirmatory factor analysis	Relationships with theoretically related construct (external construct validity)	Reliability scores above 0.7	Total score	Interpretation, ≤2 = poor quality; 3–4 = medium quality; 5–6 = high quality
Akey et al. (2000)	1	1	1	1	1	1	6	High quality
Bann et al. (2010)	1	1	1	1	1	1	6	High quality
Brookings and Bolton (2000)	1	1	1	1	1	1	6	High quality
Frans (1993)	1	1	1	0	0	1	4	Medium quality
Gagnon et al. (2006)	1	1	1	1	0	1	5	High quality
Haswell et al. (2010)	1	1	1	0	1	1	5	High quality
Holden et al. (2005)	1	1	1	1	0	0	4	Medium quality
Johnson et al. (2005)	1	1	1	0	1	0	4	Medium quality
Kasmel and Tanggaard (2011)	1	1	1	0	0	0	3	Medium quality
Koren et al. (1992)	1	1	1	0	1	1	5	High quality
Peterson et al. (2006)	1	0	1	1	1	1	5	High quality
Peterson et al. (2011)	1	0	1	1	1	1	5	High quality
Rissel et al. (1996)	1	1	1	0	1	0	4	Medium quality
Rogers et al. (1997)	0	1	1	0	1	1	4	Medium quality
Rogers et al. (2010)	0	0	1	1	1	0	3	Medium quality
Segal et al. (1995)	1	1	0	0	1	1	4	Medium quality
Speer and Peterson (2000)	0	1	1	0	1	0	3	Medium quality
Zimmerman and Rappaport (1988)	0	0	0	0	1	0	1	Poor quality
Zimmerman and Zahniser (1991)	1	1	1	0	1	1	5	High quality
Zimmerman et al. (1992)	1	1	0	0	1	0	3	Medium quality

RESULTS

The search yielded 1438 articles. One hundred and eighty-one articles were excluded, because they were either duplicates (n = 122) or non-peer-reviewed (n = 59). Screening of titles and abstracts resulted in 1212 articles being excluded. The full texts of the remaining 45 articles were read and checked for eligibility leading to the exclusion of 28 articles. Seventeen articles met our inclusion criteria. An examination of the bibliographic references of the retained articles identified a further 3 articles meeting our inclusion criteria, giving a final number of 20 studies (Figure 1).

Fig. 1:

Flow chart of study selection.

Characteristics of the included studies

Of the 20 studies retained for this systematic review, 17 were carried out in the USA, 1 in each of Australia, Canada and Estonia. Sample sizes varied from 120 to 2059 participants. The studies' populations varied, with seven studies focusing on culturally and linguistically diverse populations including African Americans and Hispanics, one study focused on Indigenous Australians and another on elderly people (Table 1).

Development and refinement of scale items

Item development followed a combination of approaches that included a literature review (n = 20), prior qualitative research (n = 12) and consultation with an expert panel (n = 9). However, only eight studies (40%) incorporated all three steps (Koren et al., 1992; Frans, 1993; Rogers et al., 1997; Holden et al., 2005; Gagnon et al., 2006; Bann et al., 2010; Haswell et al., 2010; Kasmel and Tanggaard, 2011) (Table 2).

The number of items in the empowerment scales ranged from 5 to 64. The number of domains covered ranged from 1 to 11, and included control, competence, participation, self-determination, power, self-esteem, self-capacity, identity, advocacy, assertiveness, motivation, political efficacy, leadership and positive relationships (Table 1). The majority of items (n = 17) were generated through exploratory factor analysis, and only 8 (40%) studies included a confirmatory factor analysis (Akey et al., 2000; Brookings and Bolton, 2000; Holden et al., 2005; Gagnon et al., 2006; Peterson et al., 2006; 2011; Bann et al., 2010; Rogers et al., 2010).

In sixteen studies (80%), the items were consistent with an a priori theoretical framework, which provided support for the structure of the instruments (Frans, 1993; Holden et al., 2005; Gagnon et al., 2006). Four studies did not refer to an explicit theoretical framework (Zimmerman and Rappaport, 1988; Rogers et al., 1997, 2010; Speer and Peterson, 2000).

Reliability and validity testing

Internal consistency was assessed in 19 studies (95%) and was carried out for all scales, with Cronbach's Alpha scores ranging from 0.70 to 0.95 (Table 1). However, seven of the studies (35%) had sub-scales with poor internal consistency (α < 0.70) (Zimmerman et al., 1992; Rissel et al., 1996; Speer and Peterson, 2000; Holden et al., 2005; Johnson et al., 2005; Rogers et al., 2010; Kasmel Tanggaard, 2011). In one study, an internal reliability measure was not provided (Zimmerman and Rappaport, 1988).

Test–retest reliability was reported in only three studies (Koren et al., 1992; Segal et al., 1995; Gagnon et al., 2006). Only the study by Gagnon et al. which tested the Health Care Empowerment Questionnaire, stated intra-class coefficients (ICC). Test–retest coefficients over a 2-week interval ranged from 0.60 to 0.70 for each of the factors and was 0.70 for the instrument score overall (Gagnon et al., 2006).

One study, which developed the Growth and Empowerment measure for Aboriginal Australians, tested all four components of validity, namely content, structural, internal and external construct validity. Content validity was assessed in 10 studies, and was determined by diverse methods, that included by a panel of experts to rate scale items, calculation of the Lawshe content validity ratio, factor analysis, following an a priori procedure, independent item ratings (kappa coefficients) and group discriminatory analyses (Koren et al., 1992; Zimmerman et al., 1992; Frans, 1993; Akey et al., 2000; Speer and Peterson, 2000; Gagnon et al., 2006).

The internal structure or dimensionality of empowerment items (i.e. good internal factor structure), structural validity was assessed in 17 studies, with internal construct validity examining the extent to which changes in one item cause changes in the sub-scale structures assessed in 13 studies. Of these, absolute standards of good model fit [Comparative Fit Index (CFI) >0.90 and the Goodness of Fit Index (GFI) >0.90] were assessed in 8 studies (Akey et al., 2000; Brookings and Bolton, 2000; Holden et al., 2005; Gagnon et al., 2006; Peterson et al., 2006, 2011; Bann et al., 2010; Rogers et al., 2010) and cluster and/or correlational analyses in five studies (Zimmerman and Rappaport, 1988; Zimmerman and Zahniser, 1991; Segal et al., 1995; Rissel et al., 1996; Rogers et al., 1997). External construct validity was assessed in 17 studies; of these, discriminant validity was reported in 3 studies that showed correlations ranging from 0.52 to 0.69 and correlation coefficients ranging from r = −0.23 to −0.65 (Zimmerman and Rappaport, 1988; Zimmerman et al., 1992; Johnson et al., 2005) and convergent validity was reported in five studies where statistically significant correlation coefficients ranged between 0.36 and 0.81 (Frans, 1993; Segal et al., 1995; Rogers et al., 1997, 2010; Johnson et al., 2005).

Ratings of instrument qualities

The six criteria used to rate the empowerment tools showed 9 (45%) studies were rated as high quality (15% of studies scored 6 and 30% scored 5), 10 (50%) studies were rated as medium quality (30% scored 4 and 20% scored 3) and 1 study was rated as poor quality (Table 3).

DISCUSSION

Empowerment is a principal, guiding value of contemporary health promotion, but there are few published examples of projects that have evaluated impacts upon empowerment at the individual or community level. This review has identified a wide selection of empowerment scales and reported on their measurement properties, which may assist the development of evaluation indicators and measures, and contribute to evidence and theory building about the role of empowerment in health promotion.

Through the use of exploratory factor analysis, most of the studies identified multiple dimensions of empowerment within the scales. However, the extent of scale evaluation varied widely. Test–retest reliability was evaluated in just 3 of the 20 studies. Instrument responsiveness was not reported upon, and only one study examined the predictive validity of empowerment measures in relation to health behaviour (Rissel et al., 1996). This indicates that there is scope for trialing these measures to determine their suitability for use in program evaluation.

Although empowerment is multi-faceted and embodies changes at individual, community and organizational levels, many studies focused upon psychological empowerment. Even the Socio-political Control Scale that was tested in three studies (Zimmerman and Zahniser, 1991; Peterson et al., 2006; Peterson et al., 2011) was principally concerned with motivation, confidence to lead and trust in government, rather than involvement in decision-making and control over matters of personal importance. Only two studies (10%) incorporated behavioural measures of community empowerment, (Speer and Peterson, 2000; Kasmel and Tanggaard, 2011) and two others measured healthcare users' actions to exercise influence over the services they received (Gagnon et al., 2006; Bann et al., 2010). Although one study measured organizational empowerment, it was specific to the context of mental self-help agencies and concerned the extent to which service users could influence organizational structures and decisions (Segal et al., 1995). If empowerment is addressed only at the individual level and not at community and organizational levels, then achievement of health outcomes may not be possible (Israel et al., 1994; Laverack, 2006; Wallerstein, 2006). The limited attention given to community and organizational dimensions of empowerment in the instruments reviewed here indicates scope for further development of measures to better match the strong focus on participation in health promotion.

Most studies evaluated empowerment scales with middle-aged adults. Only one study was undertaken with elderly participants (Gagnon et al., 2006) and two with young people (Holden et al., 2005; Peterson et al., 2011). Some studies (Rissel et al., 1996; Rogers et al., 1997, 2010; Brookings and Bolton, 2000) reported that their participants appeared to have higher levels of education, independent functioning, and empowerment, which may have implications for the generalizability of their findings to less empowered population groups. Only a small number of the studies reviewed conducted group discriminant analyses that examined the ability of the scales tested to differentiate between more and less empowered individuals (Zimmerman and Rappaport, 1988; Zimmerman et al., 1992).

Although several scales reviewed had an explicit theoretical basis, only a small number were developed following empirical, formative research (Akey et al., 2000; Brookings and Bolton, 2000; Bann et al., 2010) which raises some questions about whether the scales are addressing empowerment domains of relevance to different population groups. The Empowerment Scale, developed by Rogers et al.. in 1997, was informed by consultation with consumers of mental health services. This formative research enabled the development of a framework for better conceptual understanding of attributes of psychological empowerment within the context of mental health care. This tool was further refined by Rogers et al., in 2010 by factor analysis and removal of three items to enhance its construct validity.

The lack of administration of empowerment scales to participants across a range of cultures may have resulted in an inadequate demonstration of external validity (Akey et al., 2000; Bann et al., 2010). The instruments measuring individual empowerment have largely been developed using a western, individualist orientation towards empowerment. This orientation places value on independence, personal autonomy, self-determination and rights-based decision making, in contrast with the orientation found in collectivist cultures where interdependence, promotion of hierarchy and mutual obligations and fulfillment of expectations based on ascribed roles and status is valued (Oyserman et al., 2002). As Wallerstein has stated, ‘universal empowerment instruments will be insufficient and will require indicators based on local culture, language and context’ [(Wallerstein, 2006), p.16].

While research about the measurement of empowerment spans several decades, there remains considerable scope for investigation of the role that empowerment plays as a determinant and mediator of health outcomes. It has been reported that empowerment is related to engagement with health programs and the perceived quality of services received (Rogers et al., 2010), including perceived provider support and patient-centered care (Bann et al., 2010). Several studies also support the hypothesis that psychological empowerment is positively associated with participation in community activities (Rogers et al., 1997, 2010; Speer and Peterson, 2000), which may be regarded as an indicator of social well-being. There are few studies, however, which have shown that actions to improve empowerment lead to improved preventive health behaviours; those that are reported tend to focus on maternal and child health, water and sanitation and communicable diseases (Wallerstein, 2006).

Although empowerment has been a strong focus of health programs funded by the World Bank, WHO, USAID and other development agencies, there is limited information on the qualities of the empowerment measurement instruments used in needs assessment and project evaluation. Most of the studies in this review have adopted a ‘top–down approach’ wherein community empowerment was not adequately measured. Interestingly, Laverack et al., have designed a framework to enable ‘top–down’ program planners to measure community empowerment within their program structure. This framework enables an effective measurement of nine domains of community empowerment (i) participation, (ii) leadership, (iii) organizational structures, (iv) problem assessment, (v) resource mobilization, (vi) links to others, (vii) ‘asking why’, (viii) program management and (ix) the role of the outside agents, which represent the organizational influences on the process of community empowerment (Laverack and Labonte, 2000).

Using validated and psychometrically sound tools to measure empowerment would assist the design, delivery and evaluation of empowerment strategies in health promotion programs. Additionally, failure to take into account the ownership of power and capacity by the community in the design of empowerment measures could lead to misunderstandings and misguided programs that may, ultimately, result in disempowerment. This highlights the critical role of formative research and content validity testing in the measurement development process, which would necessarily entail participatory and qualitative techniques. It is important to acknowledge that qualitative methods also have a vital and complementary role to play in understanding the meaning and experience of empowerment for different groups, and the attributes of health promotion strategies that facilitate individual, community and organizational empowerment (Brandstetter et al., 2014). Furthermore, given that empowerment is a complex multi-level construct, mixed-methods approaches (Teddlie and Tashakkori, 2009) will facilitate a deeper understanding of the social and political dynamics through which this is achieved, for instance where community mobilization or policy advocacy is being undertaken.

Limitations

Because the focus of this systematic review has been quantitative measures of empowerment, with ‘reliability’ and ‘validity’ adopted as key search terms, studies that described domains and measures of empowerment but did not report on instrument development or psychometric analysis were not included (Laverack and Labonte, 2000; Laverack and Wallerstein, 2001; Laverack, 2006; Wallerstein, 2006; Wiggins, 2011; Wiggins et al., 2013). In addition, studies using qualitative or mixed methods were not included, and a review of this literature would be a valuable next step in this research and might be undertaken by means of a meta-ethnography.

CONCLUSIONS

At the First International Conference on Health Promotion, in 1986, the primacy of empowerment was recognized in the definition of health promotion, which was stated to be ‘the process of enabling people to increase control over, and to improve, their health’ (World Health Organization, 1986). This systematic review has described the psychometric properties of quantitative scales used in measuring empowerment in health promotion settings. More importantly, it has highlighted gaps in the measurement of the various domains of empowerment using quantitative scales. Most scales measured the individual domain but failed to adequately measure the community and organizational domains, which are equally important for achievement of health program outcomes. Failure to measure empowerment as a multidimensional construct may impede the process of evaluating empowerment both as a process and outcome of health programs.

Furthermore, there has been limited social and cultural diversity in the study populations with whom empowerment measures have been tested and it would appear only a few instances where measures have been evaluated among persons with varying levels of empowerment. A priority for future empowerment research is to investigate and explore methodologies such as mixed methods that would address the limitations of the tools examined in this review, particularly in the measurement of community and organizational empowerment. Of notable significance is the framework developed by (Laverack and Labonte, 2000) which enables the effective measurement of community empowerment within health programs. Such approaches will provide evidence to strengthen the design of health promotion programs, especially those concerned with addressing social disparities in health.

ACKNOWLEDGEMENTS

Professor Andre Renzaho is supported by an Australian Research Council Future Fellowship (FT110100345).