Addressing issues that have plagued researchers throughout the last decade, this book provides new insights into the many existing problems in statistical modeling and offers several alternative strategies to approach these problems. Emphasizing the importance of statistical thinking behind all analyses, the authors use specific examples in epidemiology to illustrate different model specifications that can imply different sets of causal relationships between variables. Each model is interpreted with regard to the context of implicit or explicit causal relationships. The authors also use vector geometry where applicable to provide an intuitive understanding of important statistical concepts.