Determining Sample Size

#1
Hi all:

I'm having a bit of trouble with the following exercise:

You are asked to design a prospective cohort study comparing the risk that non-smoking mothers will have a low birth weight baby depending on the father's smoking status.

Suppose:

The risk of having a low birth weight baby is 10% for families in which neither parent smoke

We would like to be able to detect a risk difference of pi1 - pi2 = .15

We plan to use a two-tailed test with alpha = 0.05


How many families need to be enrolled in this study? (Assuming equal numbers of families in which neither parent smoke and families in which only the father smoke)


I recognize that this is pretty much asking to find n for a two sample proportion (binomial) samples. I'm not sure how to proceed. Please advise!
 

Dason

Ambassador to the humans
#2
I believe we need to know the power you would like to have. Say the true risk difference is .15. How often do you want to be able to detect that difference? A typical value that is used is a power of 80%.