Meta-analysis of correlation

This revision is from 2013/12/22 14:18. You can Restore it.

This module can conduct Bayesian meta-analysis with power prior. The URL for this module is http://webbugs.psychstat.org/modules/metacorr/.

The module has an interface as shown in the figure below.

To use the program, one needs either to upload a new data file or select an existing file. Note names of the existing files are shown in the drop down menu. The existing file has to be a text file in which the data values are separated by one or more white spaces. The first line of the data file will be the variable names, which will be used in the model.

Next, a user chooses the model to use. For example, the user can choose to use either the random-effects model (default option) or the fixed-effects model. Then, information on the model can be provided. Both the Correlation and Sample size are required for all analysis, which can be specified using the variable names in the data set. For example, if we use “fi” to represent the correlation between financial performance and another variable in the data set, then “fi” should be input in the field of Correlation in the interface. Similarly, “n” is used in the Sample size field because in the data set, “n” is also the variable name for sample size. In addition, a user can also specify the variables for power and covariates used in the model.

Finally, one can elect to control Markov chain Monte Carlo (MCMC) method and output of the meta-analysis. For example, the total number of Monte Carlo iteration and the burn-in period can be specified. In the output, one can require the output of the estimates for random effects \zeta_{i}

, DIC, and diagnostic plots for all model parameters including the random effects. If one checks the option Email notification, an email will be sent to the user once the analysis is completed.