Estimating sample size for reliability study

BG2

New Member
#1
Hello, I'm having problems calculating the required sample size for my PhD study.

I have designed my first study, which is comparing moral reasoning abilities in people with a brain injury to a healthy comparison group. I have used previous similar research and G Power to estimate the required sample size needed to see a significant difference between groups (based on means and SD found in previous studies) and I think I have done this part correctly (I hope!).

However, a main aim of my study is also to assess the psychometric properties of 2 instruments designed to measure moral reasoning abilities, to see if they are valid to use with a brain injury sample. I'm not sure how to calculate the required sample size for this part of the study. I will be assessing internal consistency of the 2 instruments (Cronbach's alpha), test-retest reliability (ICC) and convergent validity (correlation between scores on the 2 instruments). Is there a way to calculate the required sample size for such studies in G Power, or is there another way to do this? I've found a few articles on estimating sample size for reliability studies but they're a bit confusing and I don't know if I'm just overcomplicating it!

Any help you can offer would be greatly appreciated!

Thank you :)
 
#2
Hi BG2,

As far as I'm aware, G*Power has no options for reliability studies, but I'm pretty sure that n-Query does.
Also, Norman and Streiner (Health measurement scales) provide a fairly simple way of calculating the sample needed for a reliability study using an ICC. It involves a number of steps and typing out all the equations is a bit difficult on a forum like this one, but you could probably get hold of the book, and use a spreadsheet to make the calculations for you. That way, you can see what the impact of changing parameters is.
Also , you could check out the package pwr (available in the free statistical program R) to determine sample size for correlations.

Hope this helps, good luck with your study!
Cheers,

Martijn