How to determine sample size

#1
Hi!
I'm trying to figure out how to estimate sample size for a retrospective study evaluating differences in volume of specific heart chambers during systole and diastole. I want to complete an a priori sample size estimate for paired t-test. But I don't have any preliminary data to estimate standard deviation. A few averages with confidence intervals where published previously. How do I estimate sample size without prior data?
 

Karabiner

TS Contributor
#2
You can calculate the standard deviations for the previous studies from the confidence intervals,
as long as you have the respective sample sizes (use the formula for the calculatíon of standard
error of the mean).

With kind regards

Karabiner
 

hlsmith

Not a robit
#3
Turn average with CIs into standard error and then into std dev. This is a good lesson in back-calculations. Post back if you have problems.