Author Topic: Reliability indices for within-multidimensional model  (Read 502 times)

OG

  • Newbie
  • *
  • Posts: 8
    • View Profile
Reliability indices for within-multidimensional model
« on: April 10, 2016, 11:32:29 AM »
Dear Eveline Gebhardt,

Using a unidimensional and a two-dimensional model ConQuest runs nicely. Using a within-multidimensional model, treating the second dimension as a bifactor, ConQuest returns very small correlation and covariance between the two dimensions (might be okay) but the reliabilities are reported strangely (see below). It might be caused by identification problems, but I'm not sure how to fix it. Any suggestions?

RELIABILITY COEFFICIENTS
Dimension: (Dimension_1)                                                       
 MLE Person separation RELIABILITY:   1.#QO                                     
 WLE Person separation RELIABILITY:   1.#QO                                     
 EAP/PV RELIABILITY:                  1.#QO                                     

Dimension: (Dimension_2)                                                       
 MLE Person separation RELIABILITY:   1.#QO                                     
 WLE Person separation RELIABILITY:   1.#QO                                     
 EAP/PV RELIABILITY:                  0.525 

« Last Edit: April 10, 2016, 11:40:24 AM by OG »

Eveline Gebhardt

  • Administrator
  • Full Member
  • *****
  • Posts: 103
    • View Profile
    • Email
Re: Reliability indices for within-multidimensional model
« Reply #1 on: April 11, 2016, 06:13:33 AM »
Hello

I am not sure what causes this problem. Would you be able to email me your command and response data file (gebhardt@acer.edu.au), so I can try to run it and see what is happening?

Best wishes
Eveline