I'm a journalist on deadline, and I'm looking for someone well-versed in statistics who can help me interpret some intraclass correlation coefficient estimates.
Specifically, how high does a test-retest reliability coefficient need to be for a test to be considered reliable? I've got a couple of published studies examining a particular psychological test's test-retest reliability, and one researcher I interviewed told me that the intraclass correlation coefficient estimates for this test are too heterogeneous and low to make this a reliable test. The other says they're fine and the test is valid.
I'm looking for a disinterested party to help me sort this out.
If you can help, please drop me an email at [log in to unmask]
thanks,
Christie
Christie Aschwanden
Email: [log in to unmask]
Website: www.christieaschwanden.com
Blog: www.lastwordonnothing.com/category/christie/
Twitter: cragcrest
|