The likely effect size for a particular comparison (e.g., drug and placebo), with a particular measurement instrument. To illustrate, if the treatment effect is expected to be only 0.2, about 400 subjects will be required per group to show the contrast as significant, setting alpha at 0.05 and power at 0.8. To detect a larger effect of 0.5, one would need about 64 patients per group(137, ls7). Effect sizes also help in adjusting com arsons between studies that used different measurements: an instrument with a large effect sine will show a higher mean improvement following treatment than an instrument with a smaller effect size.