My snobby psychometrics colleagues don't like alpha (even though it's used 98% of the time), because one can get a high alpha even if one has no general factor. Omega (hierarchical; not general) is the solution. I am writing to plea for consideration for future JMP (or if anyone is sufficiently challenged, an add-in). Thanks.
I just wanted to let you know that JMP 16.0 will have an option to obtain coefficient omega from a CFA. Whether users obtain omega hierarchical or general will depend of the type of CFA that was fit. For example, a standard CFA with 3 factors would result in omega_general for subscales. However, if such CFA is extended to a bifactor model (by adding a general factor that loads on all manifest variables), the result will be omega_hierarchical for the general factor and omega_hierarchical for subscales for the group factors. This follows the review of omega coefficients in Rodriguez, Reise, & Haviland (2015). These options are available in the Early Adopter program so we'll keep improving them as we receive feedback.
Thank you again for your suggestion!
Thanks for the info.
I'm trying to obtain the coefficient omega using the early adopter but I cannot find it.
Please could you provide indications about where to find it in the program?
We included coefficient omega as part of a "mini dashboard" of related statistics that quantify the adequacy of the measurement model at hand, thus, you'll find it in under the red triangle menu for a fitted model as, "Assess Measurement Model."
There are no labels assigned to this post.