A topic in the Open Knowledge Graph — a free, open map of 15,290 topics and the order to learn them in.
Synthesizes research design architecture for complex social investigations, covering problem formulation, literature integration, hypothesis development, and strategic selection among quantitative, qualitative, and mixed approaches. Emphasizes alignment between research questions and design choices.
Critique and redesign published studies, design a multi-phase study addressing nested research questions, work through trade-offs between internal and external validity.
By the time you reach advanced research design, you have moved beyond choosing between a survey and an experiment. The real challenge now is architecting a study that honestly answers a hard question in a messy social world — where random assignment is often impossible, where phenomena unfold over years, and where the same outcome can mean different things to different communities.
The first decision in any research design is alignment between the research question and the method. Descriptive questions ("How prevalent is X?") call for different designs than causal questions ("Does X cause Y?") or interpretive questions ("How do people experience X?"). A common mistake is selecting a method before fully clarifying the question, then retrofitting the analysis. Advanced design begins with the question, derives success criteria, and only then selects methods. This alignment also determines what validity claims the study can legitimately make.
Internal validity — the confidence that you measured a true causal effect — and external validity — the confidence that findings generalize — are in fundamental tension. Random assignment maximizes internal validity by eliminating selection bias, but laboratory conditions may be artificial. Ethnographic methods maximize ecological validity by studying phenomena in their natural context, but introduce selection effects and researcher interpretation. No single design solves both problems; the goal is to be explicit about the trade-off and scope your conclusions accordingly. Replication across different settings and populations is the primary strategy for building external validity over time.
Mixed-methods designs are not simply "doing both quantitative and qualitative" — they require deliberate integration at the design stage, not just the write-up. A sequential exploratory design uses qualitative findings to build a survey instrument. A concurrent triangulation design collects both types simultaneously and compares conclusions. The integration point (before data collection, during, or during interpretation) changes what questions the design can answer. Treating mixed methods as an afterthought — adding a few interviews to explain a confusing regression result — is a common and costly error.
Finally, advanced research design requires thinking about threats to validity proactively. What are the most plausible alternative explanations for your predicted result? Can your design rule them out, or merely acknowledge them? A study that anticipates and addresses its own weaknesses is more credible than one that ignores them. This is not a sign of weakness — it is the mark of a researcher who understands that the goal is accurate knowledge, not a convincing narrative.
Topics in reflective domains aren't scored by quiz answers. Read, reflect, and mark when you've thought it through.