Summary
Establishes the classification map for observational research, contrasting descriptive versus analytic and experimental versus observational designs, and works through the Hernan-Rubin unified approach with its thought experiment, design-before-data discipline, and forward projection. Introduces the hierarchy of evidence while noting that its paradigmatic commitments devalue qualitative knowledge, and walks through case reports, case series, and surveys using HIV/AIDS in 1981 and thalidomide as historical examples of hypothesis-generating discoveries. Closes with the cross-sectional design, prevalence versus incidence, three sampling approaches, and reverse-causation problems like the dog-ownership-and-blood-pressure example.