The figure shows two line plots of the power of a test.
The first line plot shows a plot of power against the difference to be detected, measured in standard deviations, in a comparison of means of two groups of 100 observations. The power scale runs from 0 to 1.0, with horizontal lines ruled at those values. The line is an ogive curve, rising slowly at first from power just above zero when the difference is to be detected is zero, then more quickly unitl power is 0.5 at about 0.3 standard deviations, then the rate of increase slows and the power approaches, but does not reach, 1.0 when the difference to be detectd is 0.6
The second line plot shows a plot of power against the sample size per group when the difference to be detected is 0.5 standard deviations, in a comparison of means of two groups of 100 observations. The power scale runs from 0 to 1.0, with horizontal lines ruled at those values. The line is a curve, rising steeply at first from power about 0.15 when the sample size is 10, then more slowly as power gets closer to 1.0. By sample size 150 in each group the power is very close to 1.0.
Back to "Sample size in guidelines trials".
This page maintained by Martin Bland.
Last updated: April 6, 2004.