Introduction

There is no question that health practitioners and researchers value ‘evidence’ about which interventions improve health outcomes. As health disparities have grown internationally between North and South countries and within industrialised nations, it becomes even more important to focus our energies on what matters: what policies, institutional practices, social and community interventions, and behavioural lifestyle actions make a difference to population health. The key questions, however, are not related to the ‘value’ of evidence, but instead

focus on the ‘what’ and ‘how’: what constitutes sufficient robustness or ‘weight of evidence’ to be able to attribute causes to the specific interventions, nomatter what level(s) of the socioecologic framework they address; and how canwe best collect this evidence, or what are the best research and evaluation designs? The dominant ideology to answer these questions has emerged from the

‘evidence-based medicine’ literature that privileges the randomised control trial (RCT) as the best mechanism to prove internal validity in efficacy trials, and then testing intervention in effectiveness trials in less controlled community settings (Walker and Jacobs 2002). Inherent in this translational process is the assumption that external validity increases as the settings of the effectiveness trials become more diverse. The Cochrane and Campbell Collaborations and new tools like CONSORT are used in the conduct of systematic literature reviews to identify the best evidence of outcomes and those biases that may compromise study findings (Wormald and Oldfield 1998; White 2001; Bellg et al. 2004). While there is no question that RCTs are best for identifying internal validity and

for controlling bias, there are serious limitations to accepting the RCT as the only or even the best evidence for public health or population group interventions which take place in complex social environments, especially in determining external validity (Green and Glasgow 2006). Systematic literature reviews themselves may be limited in their focus on outcome effectiveness, which regards as less important other ‘evidence’ critical for generalisability: for example, process evaluation, qualitative descriptions of contextual factors, and the quality of the intervention itself (White 2001). Community researchers interested in promoting health improvements as a result of our research face two major challenges: the role of context and the role of the researcher-researched relationship. This introductory chapter will discuss several issues in an attempt to address these challenges:

• The differences between medical and public health/health promotion interventions in dynamic social and community environments.