Svend Juul and Morten Frydenberg’s An Introduction to Stata for Health Researchers, Fourth Edition is distinguished in its careful attention to detail. The reader will learn not only the skills for statistical analysis but also the skills to make the analysis reproducible. The authors use a friendly, down-to-earth tone and include tips gained from a lifetime of collaboration and consulting.
The book is based on the assumption that the reader has some basic knowledge of statistics but no knowledge of Stata. The authors build the reader's abilities as a builder would build a house: laying a firm foundation in Stata, framing a general structure in which good work can be accomplished, adding the details that are particular to various types of statistical analyses, and, finally, trimming with a thorough treatment of graphics and special topics such as power and sample-size computations.
Juul and Frydenberg start not only by teaching the reader how to communicate with Stata through its unified syntax but also by demonstrating how Stata thinks about its basic building blocks. The authors show how Stata views data, thus allowing the reader to see the variety of possible data structures. They also show how to manipulate data to create a dataset that is well documented. When demonstrating analysis techniques, the authors show how to think of analysis in terms of estimation and postestimation. They make the book easy to use as a learning tool and easy to refer back to for useful techniques.
Once they introduce Stata to new users, Juul and Frydenberg fill in the details for performing analysis in Stata. As would be expected from a book addressing health researchers, the authors mostly demonstrate the statistical techniques that are common in biostatistics and epidemiology: case–control, matched case–control, and incidence-rate data analysis; linear and generalized linear models, including logistic, Poisson, and binomial regression; survival analysis with proportional hazards; and classification using receiver operating characteristic curves. While presenting general estimation techniques, the authors also spend time with interactions and techniques for checking model assumptions.
While teaching Stata implementation, Juul and Frydenberg reinforce habits that allow reproducible research and graceful backtracking in case of errors. Early in the book, they introduce how to use do-files for creating sequences and log files for tracking work. At the end of the book, they introduce some useful programming techniques, such as loops and branching, that simplify repetitive tasks.
The fourth edition has been substantially revised based on new features in Stata 13. The updated material has been streamlined while including new features in Stata.