Step 2: Estimate the SD

On step 2, StatMate asks us to enter the standard deviation (SD) we expect to see.

If you expect to see a lot of scatter among values (high SD), then you'll need a large sample size. On the other hand, if you expect to see very little scatter among the values (low SD), you won't need so many subjects. There simply is no way to estimate the sample size you need unless you can estimate the amount of scatter you expect to see.

For this example, we use data from other studies of plaelet alpha2-adrenergic receptors done for different reasons. These studies show that the average number of receptors per platelet is about 250 and the standard deviation is about 65. Why so high? It probably is a combination of biological variation, experimental error in counting the receptors, and experimental error in counting the platelets. Using a prior study to obtain an SD value is usually better than using a pilot study.

Choose a significance level

StatMate next asks you to choose a significance level, a (alpha). This is the P value below which you deem results "statistically significant" and thus reject the null hypothesis.

Ideally, you should set a value for a based on the consequence of making a Type I error (concluding that a difference is statistically significant when in fact the treatment is ineffective and the difference was due to random variation). If the consequences of making a Type I error are serious, set a to a smaller value, and you'll need more subjects. If the consequences of a Type I error are minor, set a to a higher value so you can get by with fewer subjects.

Most investigators always set alpha to 0.05, two-tailed, and we'll do the same.

Note: The button "Edit powers and Ns..." gives you a chance to revise the list of powers and sample sizes used on the next screen. In almost all cases, the defaults will be just fine and we'll keep the defaults for this example.

>> NEXT