How data can be analyzed is very much a function of study design and conduct. Some parameters cannot be estimated using certain study designs, for example, incidence of disease using a cross-sectional design. But if a variable has not been collected, it cannot be analyzed at all. Therefore it is prudent to plan the data analysis at the same time the study is designed. Whatever the intended ... [Show full abstract] analysis, it is useful to have a conceptual model of the relationships of interest, and the specific hypotheses or research questions (including any secondary hypotheses or questions) under study. If the study is designed to address the research question and conducted appropriately, the analysis is generally straightforward: there should be sufficient statistical power to address the study hypotheses or specific aims, and the analyses will fall directly from the hypotheses or aims. How the data are structured in the computer file makes a difference in the ease of data analysis. Although data can be manipulated, it can take considerable programming to transform data from one structure to another.