Despite demands for rigorous evaluations of educational interventions and improved knowledge about what makes interventions work there are almost no peer-reviewed published evaluations of UK-accredited courses in developing countries [21, 22]. The UK Quality Assurance Agency report only cites two unpublished reviews of courses in China and South Africa, in the last six years . We evaluated the effectiveness of a course to teach research skills to health professionals in Ghana, identified elements that were perceived to be critical in making the course successful and described the evaluation model and its usefulness in a resource-poor setting.
Using the model we have developed to evaluate a novel educational initiative in a resource-poor setting, we have demonstrated that a learner-designed course to teach research skills to health professionals in Ghana can be effective and meet international standards for quality education. The course met the needs of individual learners, who considered themselves to have become confident and competent in research, and of the institution, which increased its research capacity. A secondary but important outcome was the enhancement of learners' professional skills and better research awareness and advocacy within KATH. Critical to the success of the course was ownership by the learners, support from peers and learning by doing research and reflecting on experiences.
Strengths and limitations of the study
In developing countries demand for high quality education is increasing. When resources are scarce it is essential to demonstrate that educational interventions are effective. Our study provides a scheme for evaluating an innovative educational course, based on the theory of social learning, which was feasible to implement and interpret in a developing country context. Our study incorporates several features that identify it as research rather than a straightforward evaluation [7, 23]. The study used valid, reproducible and appropriate methods leading to neutral conclusions rather than decisions; it was not influenced by the funder and it advanced knowledge, contributed to theory and explored opportunities for transferability. A major strength of this study was the use of diverse methods to examine process, content and outcomes  and two different methods  to assess the same outputs.
Because of the nature of social learning our results need to be interpreted cautiously because measured benefits may have been influenced by other educational experiences and variations in the students' work or personal environment . In the absence of subject-relevant benchmark statements, assessment of learners' research competence was judged against UK higher education quality standards which included alignment of learning outcomes with curriculum content and assessment, use of assessments to support learning, and opportunities to reflect on learning . Self-reported ratings tend to overestimate student confidence and competence especially when the evaluation is conducted by the tutors as in this study. As skills are learned, students' self-rated ability more closely reflects actual ability levels  This 'research-naivety' among the students in our study may partly explain why they had slightly higher post-course scores in the RSES (7.4–9.7) than students in a previous study (55.3 – 82.4, scored out of 100 rather than out of 10: see table II p194 in reference 17). Despite these limitations, the RSES and SOC have been found to be useful for demonstrating the process of developing research competence and confidence and for ranking improvements in research-related attitudes, intentions and actions in previous studies [16–18]. The evaluation methods we used were simple to adapt and use in the Ghanaian setting but their usefulness needs to be assessed in a variety of different developing country contexts. Such tools can benefit from complementary qualitative data from learners' reflections and the nominal group technique to obtain richer and deeper understanding of how learners acquired research skills.
Transferability of evaluation model to other settings
Each educational setting is unique. Our learners and their learning environment had some characteristics that may not be reproduced elsewhere and which may impact on potential sustainability and transferability of our methods and findings. Our learners were highly motivated to learn research skills in order to pass professional examinations and to obtain an internationally-recognised Diploma qualification. The learners had designed the course themselves and therefore understood what they needed to do to succeed. The experience of future cohorts may be less intense, as they will not be 'pioneers' and this may influence their motivation and commitment to the success of the course. The learners were a unique mix of middle-grade, health professionals who were prepared to share their wealth of pooled expertise to compensate for the lack of local tutors. Learning outcomes can be affected by the social mix, culture and lifestyle of students  and our approach may not work so well with more junior or less motivated or cohesive learners. The KATH managers had a clear vision and strategy for developing their institution's research capacity  and provided resources rapidly and flexibly to ensure the success of the course. Although by the end of the course only a few students considered themselves to be capable of managing a research project independently, our findings show that through the course the students have developed an appreciation of the process of research and the role that research can play in improving evidence-based clinical care. This will contribute to improving the research culture within the institution .
Study outcomes in the context of theories about social learning
Our research highlights how learners set up their own 'community' of research practitioners at KATH and utilised the learning opportunities provided by the social aspects of their work environment [27, 28] to underpin changes at institutional level . Such 'communities of practice' and peer support groups facilitate learning , promote reflective interactions [30, 31] and generate 'creative turmoil' and innovation , and may explain why the DPDM course played such a key role in strengthening KATH's institutional, as well as individuals', research capacity.
To promote knowledge around coherent topics, educational research should be located within a theoretical framework . The course that we evaluated was developed using theories of social learning and workplace-learning synergies. The course evaluation, which demonstrated that students had achieved the learning outcomes, and our research into effective strategies to promote student learning in a resource-poor environment, demonstrated the importance students placed on learning by sharing their own varied knowledge and experiences. Our research therefore confirmed previous findings that learning outcomes are influenced by the social and professional diversity of students  as some learners felt disadvantaged within the cohort because of perceived differences in professional status or lack of previous research experience .
The findings from our study support theories from developed countries concerning the process of social learning at work [4, 8, 34]. Key characteristics of work-based learning are that it is managed by the learners, it is team based, innovating and empowering, and it can be enhanced by group activities that promote reflective practice and higher order thinking . Our research has demonstrated that these concepts can be successfully applied to a work-based course for health professionals in Ghana. This study therefore contributes evidence that the process of social learning at work, and theories about the role of social interactions and institutional culture on improving effectiveness of learning which have emanated from developed countries, also apply in developing countries.
Implications for educators and future research
International development policies are urging UK universities to expand access to high quality education particularly in Africa. There is an urgent need for practical tools, such as rigorous assessment processes , RSES, SOC models and qualitative analysis of student reflections, to guide delivery and evaluation of accredited courses in resource-poor countries. These tools should be sensitive to the need to understand and build on social and institutional interactions to promote effective learning. The combination of an innovative learner-designed course underpinned by peer-supported learning, originating in a developing country, and an educational quality framework, generated by developed countries, offered unique opportunities for bilateral exchange of best practice between South and North.