Author Topic: person fit for unidimensional and multidimensional models  (Read 811 times)

castleca

  • Newbie
  • *
  • Posts: 8
    • View Profile
    • Email
person fit for unidimensional and multidimensional models
« on: February 07, 2018, 07:28:22 PM »
Hi,
I am trying to compare the person fit statistics between two models with the same data, but a different dimensional structure (a 1D model compared to a 3D model). When I ask for person fit statistics along with the WLE estimates (by specifying show cases !estimates=wle,pfit=yes), ConQuest produces what looks like a MNSQ statistic accompanying the unidimensional estimates, and a t-statistic for the multidimensional estimates. Is this correct? And if so, is there any way to get a common fit statistic, short of performing the complex calculation (Wang & Chen, 2005) myself? I have attached the WLE estimates from each model. Any help is appreciated.

Thanks,
Courtney

dan_c

  • Administrator
  • Jr. Member
  • *****
  • Posts: 84
    • View Profile
    • Email
Re: person fit for unidimensional and multidimensional models
« Reply #1 on: February 08, 2018, 10:57:14 PM »
Hi Courtney,

Can you provide your data and command file, or produce a minimum working example that reproduces this for us?

https://conquest-forums.acer.edu.au/index.php?topic=1239.0

castleca

  • Newbie
  • *
  • Posts: 8
    • View Profile
    • Email
Re: person fit for unidimensional and multidimensional models
« Reply #2 on: February 09, 2018, 01:48:35 AM »
Hi Dan,

When I reran the command file trying to reproduce the error, the multidimensional wle estimates now produce a mean square. No idea what the problem was, but it seems to be fixed. Thanks for your attention anyway.

Courtney