I want to determine the sample size necessary in a study of 3 different treatment groups using a one-way ANOVA. The clinically significant difference here is +/- 10.

When entering means into a program (I have both PASS and G*Power), should I be entering something like 120, 110, 130? I'm getting very small sample sizes as a result (n=6 per group) and want to make sure that I am doing this correctly.

  • 1
    $\begingroup$ What value of the error variance are you providing to these programs? $\endgroup$ – whuber Sep 5 '18 at 17:56

With an estimated standard deviation of $\sigma = 5,$ for a three-group, one-way ANOVA with normal data, $n = 6$ replications in each group would give reasonable power (at significance level 5%) detecting inter-group differences of about 10. (That's for differences about double the standard deviation.)

Here are power curves from Minitab 17.

enter image description here


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.