Tools for Those Who Summarize the Evidence Base
Resources and networking for those who conduct or interpret meta-analyses related to any phenomenon that is gauged in multiple studies.
I'm using Mix 2.0 to do a meta-analysis and have Cohen's d computed for each study, but the program appears to require that I also compute standard error's for each effect size. Short of computing it by hand, is there a web site calculator, etc., that I could use instead (i.e., simply entering the sample sizes and Cohen's d)?
Thanks.
Comment
Thanks!
The link should work now. Sorry for the trouble!
Thank you for your helpful response. Unfortunately, the link at the bottom appears to lead to a missing page; is there another way of accessing the page?
We can start with a basic definition of the standardized mean difference effect size, ES-sub-sm:
where X-sub-G implies the mean for each compared group (and each group is independent), and s-sub-p is the pooled standard deviation. ES-sub-m would be Cohen's version, what I have often called the "raw" effect size. The version with the superscripted prime is the corrected version, per Hedges (1981), which neutralizes the bias the Cohen version has with smaller samples, using the observed sample size N. (These equations are borrowed from Lipsey & Wilson, 2001, p. 72.)
Then, the SE may be estimated as follows:
Note that the corrected version of the ES appears as part of the SE itself (by convention, do not use the uncorrected version in the calculation).
There are different versions to use for repeated measures effect sizes, so this solution only follows the between-groups formulation. On this website, you'll find a spreadsheet that already includes versions of these calculations. Look here.
Hope that helps!
© 2017 Created by Blair T. Johnson. Powered by
You need to be a member of Meta-Analysis Resources to add comments!
Join Meta-Analysis Resources