In one week, Michael Borenstein sent me the necessary supplementary materials to prepare me for including CMA in the course. Given the variety of options available in CMA, I thought that CMA might be a better fit for my class.
I planned to use another meta‐analysis software package, but learned about CMA one week before the first day of class. I recently taught an introduction to meta‐analysis course to graduate students from diverse disciplines including ecology, kinesiology, economics, forestry, veterinary medicine, family studies, and plant pathology. For these more sophisticated methods, the programme also reports the information required to compute additional statistics, such as tau squared within and between studies (enabling me to compute the R squared statistic), which are not provided by some other programmes but are commonly reported in published meta‐analyses. This meant I could delve deeper into my data to see what was really going on. Furthermore, even though I had over 50 moderators to assess, CMA made it simple to test each moderator, whilst offering the option to test moderators according to other specific study characteristics. Upon running the analysis, the programme would compute standardised effect sizes for each study (even though I might have used around 10 different types of data entry), as well as an overall effect size.
I found one useful and time‐saving aspect of Comprehensive Meta‐Analysis is that it allowed me to enter effect size data from articles in a number of formats.
This can make the process of a metaanalysis more time consuming that it necessarily has to be. Given that publications report a wide range of values from analyses (e.g., means and standard deviations, r, F, t values, eta squared, partial eta squared, etc.), it can be extremely difficult to compute effect sizes that take each of these factors into consideration.
0 kommentar(er)
