Development of PAR Empowerment Measure
Development of PAR Empowerment Measure
Name: Linnea Axman
Rank: CDR, USN
Organization: George Washington University
Performance Site: High Schools in Johannesburg, South Africa
Year Published: 2002
Abstract Status: Final
Refugee and internally displaced populations are the result of war and civil violence. Military and civilian relief nurses often find themselves providing health care in international, informal settlements (refugee and internally displaced persons camps) with only rudimentary infrastructure. Residents of these informal settlements are exposed to disease, pollution, violence, and injury. The concept of empowering these communities to take control of the determinants of their health is key to preventing disease and promoting health. Participatory Action Research (PAR) is one method relief nurses can use to empower individuals in displaced populations.
Effective health behavior change begins with the individual. However, very few quantitative tools exist to evaluate the effect of participation in PAR at the individual level; those that exist are incomplete or fail to address multilevel contextual factors. The Participatory Action Research Empowerment Measure (PAREM) was developed to fill this gap in PAR evaluation and to provide an outcome measure for this humanitarian mission-essential nursing intervention technique. Thus, the research aims of this study were to provide evidence in support of the reliability, validity, and feasibility of the PAREM within the context of the historical and sociopolitical structures of displaced populations in South Africa.
Questionnaire development employed both qualitative and quantitative methodologies. The qualitative component was particularly concerned with the cultural relevance and content validity of the instrument and employed focus group and cognitive interview techniques. The quantitative component of the study evaluated the reliability and construct validity of the PAREM using a cross-sectional research design and structural equation modeling. Multilevel and cross-level analyses were used to explore the multiple contextual factors that can affect the development of individual empowerment.
The townships of Johannesburg, South Africa host some of the largest displaced population settlements in the world. These settlements are the direct result of civil unrest, violence, and refugee immigration. Development of the PAREM in this setting was necessary not only to present evidence of the reliability and validity of this measure but also to contextualize the effects of structural violence (e.g., classism, impoverishment, racism, sexism) inherent in these and other displaced populations.
The PAREM was administered to learners (students) aged 13 through 22 years. These learners attended one of seven schools in Eldorado Park, a township of Johannesburg, South Africa that had been participating in the Safe Schools Project (a PAR project) for various lengths of time. These townships are made up of internally displaced peoples that were forced to resettle during apartheid based on perceived membership in a racial group. Refugees from surrounding war-torn and poverty-stricken countries such as Lesotho, Mozambique, Tanzania, and Zimbabwe were also represented.
Using random sampling procedures that stratified on schools, a sample size of 312 was achieved. After accounting for missing data, the effective sample size for most analyses was 298. This sample size provided for a questionnaire that assumed heterogeneity, alpha .05, and a 5% sampling error. Post hoc analyses of the power of this sample data were made specific to the analytic technique and ranged from .64 (regression analysis for interaction effect) to .81 (confirmatory factor analysis for the one-factor model).
The desired sample size of 700 was not achieved due to data collection challenges that included obtaining informed consent from parents and guardians, disruption of school schedules due to the World Summit of Sustainable Development, terrorist bombings, and labor strikes.
The following instruments were used in this study: The PAREM, the Short Form Marlowe-Crowne Social Desirability Scale (SF-MCSDS), and the Modified Perceived Racism Scale (M-PRS).
The PAREM consists of seven Likert-type subscales. The reliability coefficient (alpha) for the 33-item measure was calculated to be .83. The coefficient of stability (correlation) at one-week was .93. Inter-rater agreement was determined using the Kappa coefficient and was determined to be .98.
The SF-MCSDS is a 10-item version of the original Marlowe-Crowne Social Desirability Scale (MCSDS) (Cowne & Marlowe, 1964). This short version has been used in previous research with ethnic minority adolescents (Stanton et al., 1993). The 33-item MCSDS has demonstrated an internal consistence of .88 using the Kuder-Richardson formula 20 and a test-retest correlation of .88. The MCSDS is in the public domain.
The M-PRS is a 23-item measure consisting of 3 subscales. Like the original 51-item scale, the MPRS was employed to measure perceptions of racism. The original instrument was modified for use in a South African youth population. The reliability coefficient for the original PRS has ranged from .88 to .96. The overall M-PRS demonstrated an alpha of .63 in this study population, with the Feelings (about Racism) subscale demonstrating an alpha of .72.
The major findings of the study supported the hypothesis that the PAREM demonstrates stability and internal consistency as a measure of reliability. The content validity of the measure is logically present and was supported by qualitative analysis of the data. Evidence to support the instrument's construct validity was demonstrated through confirmatory factor analysis across groups and levels. No interactions among covariates were observed in the sample data; however, the analysis was limited by sample size, level of program implementation, and effect size. The Process Implementation Index (PII) for the PAREM was lower than desired; however, it established the "real-world" feasibility of its use in the sampled population.
Further research is recommended to provide continuing support for the reliability, validity, and feasibility of the PAREM. Ideally these studies will be framed by ecological systems theory, use designs that are longitudinal in nature, incorporate multiple measures, and will include those contextual variables that can promote or inhibit the development of individual empowerment over time.
The development of the PAREM and the findings of this research study have implications for nursing practice, policy, and research. These implications are specific to the evaluation of nursing interventions within a participatory framework.
Nursing Practice. Quantitative measurement tools such as the PAREM are necessary to evaluate programs using participatory approaches with the aim of behavior change. Because observable behavior change may not be apparent for months to years, intermediate measures of change should be used, not only to demonstrate the efficacy of a behavior change program but also to prevent the premature termination of potentially effective programs.
Nursing Policy. Funding and policy decisions often require information that may not be easily measured early in an intervention process, such as effect size. Nursing policy-makers must demand that investigators, practitioners, and program planners demonstrate that their interventions will contribute significantly to the patients' interests and that they constitute a cost-effective use of scarce resources. Some instrument that is capable of measuring change, whether it is the intended behavioral outcome (decreased interpersonal violence) or some intermediate phenomenon (level of empowerment) that shows movement or change must be included in requests for program support. The development of the PAREM has added one such instrument to the evaluation tool inventory.
Nursing Research. This study incorporated many of the recommendations suggested by the Institute of Medicine (2001) and related subject matter experts, specifically that PAR processes are evaluated using tools that are reliable and valid in the target population; that process measures demonstrate the extent to which the intervention was delivered to the target audience; and that analysis includes individual, organizational, and community-level variation within each treatment condition. Future evaluations of PAR projects should continue to follow these recommendations in order to contribute to the existing body of knowledge.
Recommendations for Future Nursing Research. This study should not be replicated using the same study design. Instead, the PAREM should be used and evaluated in longitudinal PAR projects that incorporate comparison groups and process measures that demonstrate the extent to which the intervention was delivered to the target audience, thus allowing the natural history of empowerment to unfold within a participatory process over time. Future validation studies of the PAREM should include documentation of the quantifiable dose necessary to achieve intermediate and long-term outcomes or the hypothesized quantifiable dose for each intervention and program element that is being evaluated.