Dear Statalist users,
I am trying to do power analysis with the following, and noted what I think seems to be an anomaly in the calculation of the effect size.
Here are 2 separate calculations (one to determine the power, and the other to determine the estimated effect size given a power of 0.80). Note that the effect size (delta) is simply the difference between m1 and m2. However, my understanding is that effect size should be a standardized difference, i.e., in terms of how many standard deviations from the mean, and not simply the difference. Is there something wrong with the way the delta has been calculated below? Note that I have also specified the standard deviations of each group (and they are both different from each other and from 1).
Thank you.
Group: Variable | Obs Mean Std. Dev. Min Max
-------------------------------------------+----------------------------------------------------------------------------
Control Group: EffortMin0 | 52 4.215385 .812738 .1 5.7
Treatment Group: EffortMin0 | 50 4.486 .5613431 3.4 5.9
Estimated power for a two-sample means test
Satterthwaite's t test assuming unequal variances
Ho: m2 = m1 versus Ha: m2 != m1
Study parameters:
alpha = 0.5000
N = 102
N1 = 52
N2 = 50
N2/N1 = 0.9615
delta = 0.2706
m1 = 4.2154
m2 = 4.4860
sd1 = 0.8127
sd2 = 0.5613
Estimated power:
power = 0.9050
Estimated experimental-group mean for a two-sample means test
Satterthwaite's t test assuming unequal variances
Ho: m2 = m1 versus Ha: m2 != m1; m2 > m1
Study parameters:
alpha = 0.5000
power = 0.8000
N = 102
N1 = 52
N2 = 50
N2/N1 = 0.9615
m1 = 4.2154
sd1 = 0.8127
sd2 = 0.5613
Estimated effect size and experimental-group mean:
delta = 0.2013
m2 = 4.4167
I am trying to do power analysis with the following, and noted what I think seems to be an anomaly in the calculation of the effect size.
Here are 2 separate calculations (one to determine the power, and the other to determine the estimated effect size given a power of 0.80). Note that the effect size (delta) is simply the difference between m1 and m2. However, my understanding is that effect size should be a standardized difference, i.e., in terms of how many standard deviations from the mean, and not simply the difference. Is there something wrong with the way the delta has been calculated below? Note that I have also specified the standard deviations of each group (and they are both different from each other and from 1).
Thank you.
Group: Variable | Obs Mean Std. Dev. Min Max
-------------------------------------------+----------------------------------------------------------------------------
Control Group: EffortMin0 | 52 4.215385 .812738 .1 5.7
Treatment Group: EffortMin0 | 50 4.486 .5613431 3.4 5.9
Estimated power for a two-sample means test
Satterthwaite's t test assuming unequal variances
Ho: m2 = m1 versus Ha: m2 != m1
Study parameters:
alpha = 0.5000
N = 102
N1 = 52
N2 = 50
N2/N1 = 0.9615
delta = 0.2706
m1 = 4.2154
m2 = 4.4860
sd1 = 0.8127
sd2 = 0.5613
Estimated power:
power = 0.9050
Estimated experimental-group mean for a two-sample means test
Satterthwaite's t test assuming unequal variances
Ho: m2 = m1 versus Ha: m2 != m1; m2 > m1
Study parameters:
alpha = 0.5000
power = 0.8000
N = 102
N1 = 52
N2 = 50
N2/N1 = 0.9615
m1 = 4.2154
sd1 = 0.8127
sd2 = 0.5613
Estimated effect size and experimental-group mean:
delta = 0.2013
m2 = 4.4167