Author Topic: Indicating the direction of the latent trait in Bock's nominal response model  (Read 665 times)

Koichi

  • Newbie
  • *
  • Posts: 1
    • View Profile
    • Email
Hello.

In general terms, we know that the direction (+/-) of the latent trait doesn't make sense in Bock's nominal response model (NRM).
But, when we want to do item analysis of a test based on NRM, there exists "high (+)" or "low (-)" level in the ability scale.

I found that the direction of the latent trait in NRM depends on the runs in ConQuest estimation.
I attached 2 files of IRCCC of a single item. What I expect to get is "expected_IRCCC", which was gained through a trial estimation of 30 items.
When I run another calibration of 257 items, I get "not_expected_IRCCC", where "high" ability locates in the minus (-) direction.
I appreciate if you could help me to control the direction of the latent trait in the command file or in the drawing of IRCCC.

Best Regards,
Koichi

dan_c

  • Administrator
  • Jr. Member
  • *****
  • Posts: 84
    • View Profile
    • Email
Can you post your datafile, and syntax or create a reproducible example using the online examples so we can reproduce this?

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