I have a set of studies that I’m trying to meta-analyze with OpenMeta[Analyst]. I have raw means and SD but would like to analyze standardized mean differences (Cohen’s d) for ease of interpretation (especially because the DV is in squared units).
Here’s the data I inputed into OpenMeta[Analyst]…the last 3 columns are the standard mean differences calculated, with upper and lower bounds for CI:
When I try analyzing the same data with a different tool (Meta-Essentials), however, I get:
Study name
Cohen's d
CI Lower limit
CI Upper limit
Online
0.34
0.14
0.53
College of DuPage
0.49
0.23
0.76
Dominican University
0.30
0.00
0.60
Concordia University
0.26
-0.06
0.57
Notice that this tool gives considerably higher effect size estmates, especially for the first two studies (0.278 vs 0.34 for Online study; 0.38 vs 0.49 for College of DuPage study).
Why is this? I’ve checked using a third online calculator using both the raw means and the t test results and my results always agree with the higher effect size estimates from Meta-Essentials. What’s strang, though, is that in other data-sets I’ve analyzed wth OpenMeta[Analyst] I get no discrepancy… not sure if there is something special happening here in OpenMeta[Analyst] in terms of pooling variance across all studies to estimate effect size? Any hints?
I have a set of studies that I’m trying to meta-analyze with OpenMeta[Analyst]. I have raw means and SD but would like to analyze standardized mean differences (Cohen’s d) for ease of interpretation (especially because the DV is in squared units).
Here’s the data I inputed into OpenMeta[Analyst]…the last 3 columns are the standard mean differences calculated, with upper and lower bounds for CI:
When I try analyzing the same data with a different tool (Meta-Essentials), however, I get:
Study name
Cohen's d
CI Lower limit
CI Upper limit
Online
0.34
0.14
0.53
College of DuPage
0.49
0.23
0.76
Dominican University
0.30
0.00
0.60
Concordia University
0.26
-0.06
0.57
Notice that this tool gives considerably higher effect size estmates, especially for the first two studies (0.278 vs 0.34 for Online study; 0.38 vs 0.49 for College of DuPage study).
Why is this? I’ve checked using a third online calculator using both the raw means and the t test results and my results always agree with the higher effect size estimates from Meta-Essentials. What’s strang, though, is that in other data-sets I’ve analyzed wth OpenMeta[Analyst] I get no discrepancy… not sure if there is something special happening here in OpenMeta[Analyst] in terms of pooling variance across all studies to estimate effect size? Any hints?
Thanks,
Bob