Béhague and Storeng Respond

Thank you for the opportunity to respond to Yeh and Brandes’s letter regarding our article.1 We are encouraged by their interest and will address their three points below. First, Yeh and Brandes are concerned that our use of opportunistic or purposive sampling, being nonsystematic and nonprobabilistic, has biased our results. However, for certain types of research questions, purposive sampling generates data of higher quality, validity, and pertinence than does systematic sampling.2 This is particularly the case when studying socially complex phenomena and when seeking to explore the views of specific types of informants. Additionally, unlike sampling from a known population base, systematic approaches are inappropriate when sampling from an unknown sampling pool, as is the case with internationally based expert communities. In such cases, purposive sampling may in fact ensure better variability of the sample than would systematic sampling (see, for example, sampling methods used by Costello et al.3). Second, the assertion that our article concludes “advocacy efforts for maternal health would benefit from additional evidence” misinterprets our results. Our informants highlight, rather, that the pressure to garner support and funds for maternal health is forcing experts to pursue an exclusive focus on vertically oriented interventions and, thus, to prioritize experimental research designs. Although these forms of evidence are often more effective in advocating for maternal health, they are also detrimental to promoting a much-needed emphasis on integrated health systems initiatives. The final comment regarding the relative merits of the randomized controlled trial study design is more an affirmation of our observations than a request for clarification. However, our article showed that alternative models for generating evidence should not be restricted to quasi-experimental or “rigorous” evaluation designs, as Yeh and Brandes propose. Such designs attempt to reproduce, rather than critically assess, the experimental model upon which the randomized controlled trial is based.4 Ensuring improved population health depends on diverse phases of programmatic activity, including the development of interventions, evaluation, monitoring, and in-depth problem solving. Research inputs for each of these phases emerge from distinct epistemological approaches.5,6 For example, developing new interventions requires not “rigorous” evaluation, but “formative research,” composed usually of observational epidemiology, sociology, and anthropology. If one accepts these epistemological distinctions, it would be incorrect to conclude, as Yeh and Brandes do, that the “experimental approach represents a standard to attain.” At certain moments in the public health process, recourse to the experimental approach would be an erroneous choice altogether

Assertion · Environmental health · Nonprobability sampling · Pathology · Population · Psychiatry · Psychological intervention · Research design · Sample (material) · Sampling (signal processing) · Social science · Sociology · Systematic sampling · Child and Adolescent Health · Computer Science · Food Security and Health in Diverse Populations · Health Policy Implementation Science · Medicine · Psychology · Social Psychology

• Exploring Potential Pathways Between Parity and Tooth Loss Among American Women

  Stefanie L Russell, Jeannette R Ickovics et al.•American Journal of Public Health•2008

  Add to list

• Collapsing the Vertical–Horizontal Divide

  Dominque P Behague, Dominique P Behague et al.•American Journal of Public Health•2008

  Add to list