Part IV: Specify the Methods

You have a question and a design. Now you have to decide who will be in your study, how many of them you need, whether your procedures will work in the setting, and how you’ll collect the data. These decisions sound like logistics. They aren’t. A well-designed trial can still fail because it couldn’t enroll enough participants, and a validated measure can still produce bad data if the field team doesn’t administer it the way it was intended.

Most studies don’t fail dramatically. They fail quietly: a sample that doesn’t represent the population, a trial too small to detect the effect it was built to find, a protocol that falls apart in the first month of fieldwork.

The gap between the study you design and the study you run is where many of these problems start. Some of that gap is unavoidable. The goal of this part is to help you anticipate it, keep it small, and be honest about what it means for your results.

WHAT THIS PART COVERS

Sampling and Participant Recruitment starts with the chain of inference from the population you care about to the people you actually enroll. The Demographic and Health Surveys show what rigorous probability sampling can achieve at national scale. Most teams won’t have those resources, so the chapter also covers what to do when your sample isn’t representative, when probability sampling matters and when it doesn’t, and how sampling goals differ for prevalence studies, case-control studies, qualitative research, and trials. It ends with methods for reaching hard-to-reach populations, including venue-based sampling and respondent-driven sampling, and with what those methods can and can’t promise.

Power and Sample Size asks how many participants you need. The answer depends on the kind of question you’re asking, so the chapter sorts sample size into three buckets: saturation for qualitative studies, precision for descriptive estimates, and power for hypothesis tests. You’ll see what drives power, how to run standard calculations with free tools, and how to use simulation when a formula won’t fit your design.

Preliminary Studies and Pilot Trials covers the work you do before committing to a definitive study: formative qualitative research, feasibility and acceptability studies, and pilot trials. The chapter is clear about what these studies can tell you (whether recruitment, delivery, and measurement work in your setting) and what they can’t (how large the effect will be). It ends with how to turn preliminary findings into a go/no-go decision.

Data Collection and Fieldwork turns to the work of collecting data in the field: building relationships and approvals before you start, hiring and training a field team, collecting quantitative and qualitative data, using routine health information systems, and building a data pipeline that keeps raw data intact and analysis reproducible.

Each chapter takes up one part of the same problem: turning a research question into procedures that produce evidence you can trust. Who is in the study, how many, whether the procedures work, and how the data get collected. None of these decisions is purely technical, and getting any one of them wrong can undo the careful thinking that went into your question and design.