/*
This program can be used to compute the sample size needed
to achieve a specified power for the overall F test in
single-factor ANOVA. The user must specify
alpha=significance level of test
t=number of treatments
variance=error variance
sumtausq=sum of squared treatment effects
The program will compute the power of the overvall F test for number of
reps (r) ranging from 2 to any user specified value. By default, the power
calculations will stop at r=30, but that is easily changed in the second
line of the code below.
The original values for alpha, t, variance, and sumtausq correspond
to Example 2.3 on page 64 of Kuehl.
*/
data one;
do r=2 to 30;
alpha=0.05;
t=3;
variance=0.22;
sumtausq=0.38;
dfnum=t-1;
dfden=r*t-t;
lambda=r*sumtausq/variance;
power=1-probf(finv(1-alpha,dfnum,dfden),dfnum,dfden,lambda);
output;
end;
proc print noobs;
var r power;
run;