Comparing means and medians (with different dispersion indeces) from different studies I'm doing a systematic search and I have a pool of articles that for each outcome report different central values (mean and median) and dispersion indices (SD, 95% CI, IQR). Is there a statistical test that I can get to a value that describes each variable? Thank you very much
 A: While there are statistical tests that can be run with meta-level data such as this, it seems that you want to summarize the measured effect for something over a number of research articles and reports.  Thus, you are asking to do a meta-analysis.
There is a large body of research on meta-analysis that you can draw from in order to accomplish this.  In brief, there often is a challenge exactly as you describe in attempting this.  Here, you want to try to describe the central tendency (say the mean of the population), but you have different measures of center from different reports.  Likewise for the variability of the population.
Without going into too much detail, one strategy that you can use is to record the mean and/or median as the dependent variable in your meta-analysis data collection.  But, you will also want to have another variable in the data set to record if it was reported as the mean or median (or something else).  Same will need to be done for the measures of spread you record.
Once you have this, you can then transform the estimates into a common metric.  For example, you can divide the IQR by about 4/3 to get the SD.
One final comment is that you want to make sure you are collecting as much additional information from each of the published reports as is reasonably possible, as this will allow a more detailed analysis of the final meta-analysis model.  The key here is to be as transparent in everything so future researchers can assess potential concerns form the estimates your research reports.
