Calculating Sample Size - Statistics (May/22/2008 )

Hi all,
How can I calculate the sample size?
I found an online sample size calculator, but it needs that I provide the confidence interval, I tried to get what it means, but everybody is explaining without providing any examples. If I have a key-article for the same work, what is the confidence interval in an article? I am working on surface receptors on NK cells in a deisease pre and post treatment.
Thanks in advance.

Check out some results here:

http://search.vadlo.com/b/q?keys=confidenc...mp;sn=158621799

Powerpoints: http://search.vadlo.com/b/q?sn=158621799&a...erval&rel=2
..

Ahhh, statistics, the bane of the biologist.

The sample size is the number of independent samples you have. This could be the number of patients with disease X studied or any set of treatments. For example if I am doing a study, I might have patients all with imaginary disease of whom 30 were treated with hugs , 40 treated with a blow to the head and 30 with a sugar pill as a control (this is a really badly designed expt, can you tell me why?) The sample size is 100, but if you were supplied with figures on how patients did with each treatment you could work out whether the treatments did better or worse than the control and how confident you were in the numbers.
Confidence intervals are error values based on how reliable the data is. This means that for a 95% CI you would expect your calculated mean, regression etc to be wrong 5% (1/20th) of the time. I take it you are trying to do meta-analysis on published work. In the case of published work you need to look at the errors on any numbers they give (e.g. 70% +/- 3%), these are directly related to the confidence interval, the bigger the error and the smaller the sample size, the wider your confidence interval will be.

I suggest you get a good basic stats book and start from there, this isn't something that is easy to explain without knowing exactly what you are looking for.

Note that before and after treatments are not independent samples as one is dependent on the other (i.e. 100 cases taken before and after treatment does not give you a sample size of 200).