Reproducible Statistical Analysis and Reporting Activities and Goals in the Department of Biostatistics

In the Vanderbilt Department of Biostatistics, reproducible statistical analysis and reporting is of the utmost importance. The goals of reproducible analysis and reporting are
  1. To combine programming code with report text in a single self-documenting file
  2. To document the code and its results, including interpretation of the results
  3. To allow an analysis to be re-executed and the resulting report be re-typesetted by running a single command
  4. To avoid all transcription of calculations and tables into text

Besides improving accuracy of reports, reproducible analysis saves time and allows a single command to be run whenever the data or the analysis changes. It also solves the problem of updating analyses after a long pause in the project, in which case the analyst frequently forgets the needed sequence of steps.

Biomedical researchers have praised the comprehensive and beautiful pdf reports we provide them. Our pdf reports are generated with Rís knitr tools, using the R language for computation. The reports combine text, plots, and code. This provides maximum information for the researchers and greatly facilitates their manuscript writing process and other presentations.

All new MS biostatisticians at Vanderbilt receive training in R's knitr and other reproducible reporting tools. We routinely use the following tools of reproducible research.
  1. knitr - an extremely flexible tool for producing the typical statistical report using R and LaTeX
  2. R tools for creating LaTeX markup code for advanced statistical tables optionally containing results of statistical tests
  3. greport - a package of extremely high-level R functions for clinical trial reporting using R and LaTeX. The reports produced by greport contained advanced tables and are very graphical. greport has been used in production mode in a large international NIH clinical trial. greport is used for recurring reports where more control is needed by the analyst than that provided by knitr.
  4. Makefiles

Detailed information
Topic revision: r7 - 22 Apr 2015, FrankHarrell
 

This site is powered by FoswikiCopyright © 2013-2020 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding Vanderbilt Biostatistics Wiki? Send feedback