# sample size calculation

#### dcmarvel

##### New Member
Hi. I am trying to calculate a sample size for a study and need help!!
We are looking at pre - post values, minimil important difference = 2 (8.5 baseline, 10.5 post from other study), sd in pop = 5.9, alpha = .05, power = 80%.
I keep getting a ss of 13 per group which seems incredibly low. Any suggestions?
Thanks

#### jamesmartinn

##### Member
I can answer your question after my lab meeting later on in the evening. But just a question of interest, do you goto York University? Oddly enough, we just covered power in an intermediate stats class and your question looks strangely fimilar to the ones assigned in my hw. That or maybe we use the same book lol?

but yeah, ill give you a hand a bit later on for sure.

In the mean time, check out this program called G*Power. It's a tool for Power analysis, VERY useful.

#### jamesmartinn

##### Member
okay, i'm assuming this is a power analysis for an independent t - test for two samples and not repeated measures t-test since you didn't specify a correlation between your two variables..

n = (delta / d) ^2 x 2

d = u1 - u2 / sd
= 2 / 5.9
= .34

For Power = .80, delta = 2.80 (From the Power Table)

n = (2.80/.34)^2 * 2
n = 136

So you need 136 people in your sample.

#### dcmarvel

##### New Member
Thanks so much for that JamesMartinn,
it is actually a pre-post. So we are looking at change within an individual.
Will this change the calculation at all?

Thanks again.

#### jamesmartinn

##### Member
If it's a repeated measure study, you need to have an estimate of the population correlation between the two variables to compute the required sample size.