Hello all,

I have a few scales (4 - 6 items each) that should be reliable but are not. Cronbach alphas range from .55 to .68, which are not improved by deleting individual items. My question is: if I compute factor scores in place of a simple mean - essentially, use ad hoc weighted means - would reliability no longer be an issue? Or would I just have an unreliable scale with weights?

Thanks -warren

Similar questions and discussions