Cohen

1
Cohens d and Hedges g For two Independent Means
Assuming
Power
Then use the Power Table for deltas
Example Con1 Con2 Con3 Mean 10 15 25 SD 18 20 22
n 20 20 20
Power for contrasting Condition 1 and condition 2
Power from table is approximately .15
2
How many subjects do we need for power .8?
We need the delta that corresponds to power of .8
( delta 2.8 from the table)
Our example from the previous page.
Thus, we would require approximately 250 subjects
per condition to have a power of .8 Remember,
there is only an issue of power if the null
hypothesis is false.
3
For Repeated Measures
The denominator is the SD in the difference
scores.
Variance Sum Law / Assuming Homogeneity of
Variance