Okay, I'm faced with a problem that I'm hoping someone on the list
can provide a better solution to than the one I've come up with so far.
I am aware of the W statistic (Feldt, 1969) for testing the
difference between 2 coefficient alpha values. Currently, I'm working
on some data from a 3x3 design trying to identify which condition(s)
produced significantly larger reliability estimates. I could (and
actually have) conducted 36 paired comparisons and then adjust the
Type I error rate for significance, but I am wondering if there is an
extension or alternative that would allow me to do a more omnibus
test of the coefficient alpha values prior to the paired comparisons.
Additionally, if you know of any measure of effect size you think
would be appropriate when discussing the difference between two
coefficient alpha values, I'd love to know so I can include it.
Any help is greatly appreciated
- marc
=============================================
G. Marc Turner, PhD, MEd, Network+, MCP
Senior Lecturer & Technology Coordinator
Department of Psychology
Texas State University-San Marcos
San Marcos, TX 78666
phone: (512)245-2526
email: [EMAIL PROTECTED]
---
