Question 203:

Asked on May 22, 2008

Tags: Sample Size , ANOVA , Power , t-test

1

Answer:

No answer provided yet.

For each of those parameters you are wishing to estimate, you need to first how they are measured (as a continuous measurement or discrete--such as presence or absence). It looks like they might be continuous measurements.   The next thing you need is an estimate of the standard deviation, which is a required piece of information to calculate sample size for continuous data. 

You then need to know how you will be comparing the populations (will you be using a t-test, an ANOVA or just confidence intervals?).

Once you have that data, you'll want to know what amount of difference you'd hope to detect (e.g. is a 20 point cholesterol difference too large or small to be meaningful) and then use common levels of Power, such as .70, .75 and .80 to find out how large your sample needs to be.