Message/Author |
|
|
Hi, I´ve run a CFA for threee questionnaires. I´ve treated the items as ordered categorical; they are not normal distributed. I would like to know if there is any indice (like Cronbachs-alpha) to indicate the reliability of the questionnaires. I don´t want to use Cronbachs-alpha because my data do not comply with the requirements (like metric data and normal distribution). Thank you very much for your help! |
|
|
You can look at the precision of the factor by examining the information function or the standard errors of the factor scores. |
|
|
Thank you for your answer. If I got you right, there is not one coeffizient for the whole three-factor-solution of the whole questionnaire (like cronbachs alpha,..)? Is it possible to compute the discriminability for each item with Mplus? |
|
|
You are correct. There is not one number. Factor score discrimination is available only for continuous outcomes. |
|
Back to top |