How were the data analyzed?
Katherine W. Reeves · 2025
Epidemiologists and health professionals need to be able to evaluate whether data were analyzed appropriately and whether appropriate conclusions were drawn from the results. This chapter discusses what statistical approaches are used in various situations and reviews the most common regression approaches used in epidemiologic studies. Statistical power and how it impacts our interpretation of study results are discussed, as are problems associated with multiple comparisons and missing data. Confounding, where a third factor affects the apparent association between the exposure and outcome being studied, can be controlled through a variety of approaches. Questions are presented that can help evaluate whether confounders were identified and appropriately controlled in the study. Effect modification is explored, especially with regard to interpreting results showing the exposure-outcome association is modified by a third factor. Finally, all this is put together to identify what students should focus on as they read the statistical methods section of journal articles, with an eye to how the statistical analysis impacts the study’s internal validity.