Omega hierarchical (reliability)

[#1]

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.

[#2]

Make sure you submit the request to the JMP Wish List?

[#3]

Thank you @profjmb!  It's a great suggestion. I just added it to our to-do list. We'll eventually get to it! 

Best,

~Laura

[#4]

Hello @profjmb,

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!

~Laura

[#5]

Hi Laura,

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?

Thanks

[#6]

Dear @Fernando,

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."

Best,

~Laura