MetaWin 1.0 Homepage


MetaWin is a program specifically designed to perform meta-analyses. Using MetaWin, you can perform fixed effects and mixed effects models: the traditional parametric models of meta-analysis. Additionally, nonparametric resampling tests can be performed. Input files for MetaWin are simple text files, and complete instructions for designing these files are provided in both the manual and help file that comes with the program. The two versions of the program, one for Windows 3.1 and one for Windows95, come in a single package. Cost for a single user, just $70.

MetaWin, written by Michael S. Rosenberg, Dean C. Adams, and Jessica Gurevitch, is designed to perform meta-analyses. Using MetaWin, you can perform fixed effects and mixed effects models: the traditional parametric models of meta-analysis. In addition, nonparametric resampling tests can be performed. Input files for MetaWin are simple text files, and complete instructions for designing these files are provided in both the manual and the help file that comes with the program.

MetaWin increases the flexibility of your data analysis in several ways. The easy-to-use data file format permits you to specify several different grouping categories (sex, location, etc.), which allows you to change the focus of your analysis without having to create additional data files. MetaWin also allows you to chose the model of meta-analysis (fixed effects or mixed effects) by merely clicking the mouse. Resampling statistics, including randomization tests and bootstrap confidence limits, can be included in the analysis as well, with the number of iterations specified by the user. Either parametric or nonparametric weights can be chosen for the resampling statistics, the implications of which are discussed in the manual. In addition, a refine analysis window allows you to ignore particular studies or classes of studies. This is often necessary after preliminary data analysis reveals outlier studies that adversely affect the results of the analysis. This also allows you to compare studies based on one set of classes (e.g. male, female), while ignoring studies based on another set of classes (e.g. marine).

The output from MetaWin provides all of the necessary meta-analytic statistics, including the sample statistics for each study. These include each study's effect size and parametric and nonparametric weights. In addition, summary meta-analytic statistics are presented as well, including: the cumulative effect sizes for each class, the 95% confidence limits, the bootstrap and bias-corrected bootstrap confidence limits (if specified), as well as the within-class heterogeneity. Tables for both fixed effects and mixed effects models are provided if the mixed effects model was specified. Between-class heterogeneity is tested against a Chi-square distribution as well as a randomized distribution (again if specified). Finally, the grand mean effect size and its bootstrap confidence limits are provided. The output can be saved to a text file for later use.

MetaWin comes with a 65 page manual that includes an introduction to meta-analytic techniques, as well as a brief summary of resampling statistics. The manual gives detailed instructions for installing MetaWin and generating data files, and provides step-by-step instructions on how to use the program. In addition, a help file is provided with MetaWin, which allows the user direct access to most of the information present in the manual. Each purchase of MetaWin comes with both a Windows95 and Windows 3.1 version; their differences are discussed in the manual.


Revised by Dean C. Adams, dcadams@iastate.edu , 7 December, 1999.