MBESS

MBESS is an R package that was developed primarily to implement important but nonstandard methods for the behavioral, educational, and social sciences. The generality and applicability of many of the functions contained in MBESS have allowed the package to be used in a variety of other disciplines. Both MBESS and R are open source and freely available from The R Project's Comprehensive R Archive Network. The MBESS Web page contains the reference manual, source code files, and binaries files. MBESS (and R) is available for Apple Macintosh, Microsoft Windows, and Unix/Linux operating systems.

The major categories of functions contained in MBESS are (a) estimation of effect sizes (standardized and unstandardized), (b) confidence interval formation based on central and noncentral distributions (t, F, and χ2), (c) sample size planning from the accuracy in parameter estimation and power analytic perspectives, and (d) miscellaneous functions that allow the user to easily interact with R for analyzing and graphing data. Most MBESS functions require only summary statistics. MBESS thus allows researchers to compute effect sizes and confidence intervals based on summary statistics, which facilitates using previously reported information (e.g., for calculating effect sizes to be included in meta-analyses) or if one is primarily using a program other than R to analyze data but still would like to use the functionality of MBESS.

MBESS, like R, is based on a programming environment instead of a point-and-click interface for the analysis of data. Because of the necessity to write code in order for R to implement functions (such as the functions contained within the MBESS package), a resulting benefit is “reproducible research,” in the sense that a record exists of the exact analyses performed, with all options and subsamples denoted. Having a record of the exact analyses, by way of a script file, that were performed is beneficial so that the data analyst can (a) respond to inquiries regarding the exact analyses, algorithms, and options; (b) modify code for similar analyses on the same or future data; and (c) provide code and data so that others can replicate the published results. Many novel statistical techniques are implemented in R, and in many ways R has become necessary for cutting-edge developments in statistics and measurement. In fact, R has even been referred to as the lingua franca of statistics.

MBESS, developed by Ken Kelley, was first released publicly in May 2006 and has since incorporated functions contributed by others. MBESS will continue to be developed for the foreseeable future and will remain open source and freely available. Although only minimum experience with R is required in order to use many of the functions contained within the MBESS package, in order to use MBESS to its maximum potential, experience with R is desirable.