Hello,
How many respondents are necessary for test-retest reliability?
In my opinion, with test-retest reliability men analyze the correlation between two different scores therefore respondent number above 30 is enough.
Could you also please give reference?
Meryem
You should do a power analysis to determine the right sample size. The effect size varies on whether you measure a stable construct and the time between test and re-test.
Assuming a test-retest correlation of .50, you would need a sample size of 46, if you would want a power of .95. https://www.anzmtg.org/stats/PowerCalculator/PowerCorrelation
Dear Paul,
Thanx for the useful link.. For a long time, I was looking for the answer of this question. Again thank you 👍🏿☺️
The number always matter. If you are testing, be sometimes conservative and look for small 'n'.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics