Author Topic: Different case numbers in output files (and data file)  (Read 562 times)

philipp

  • Newbie
  • *
  • Posts: 2
    • View Profile
Different case numbers in output files (and data file)
« on: August 25, 2016, 09:08:12 AM »
Hi all!

I've merged two data sets with anchor items in one data file including about 920 cases alltogether. In the shw-output only 481 cases in file and in estimation are reported. However, at the end of item analysis output is stated N=924. WLE estimates are calcuated for all 924 participants. BTW: When I am calculating the two data sets separartly the correct number of cases (about 450 each) is reportet in the shw-output.

This deviation between N and cases in estimation seems a little bit strange to me. What do you think might be the reason? Just a bug? Or are about 50% of my data skipped during calculation?

Thanks in advance! Phil.

Eveline Gebhardt

  • Administrator
  • Full Member
  • *****
  • Posts: 103
    • View Profile
    • Email
Re: Different case numbers in output files (and data file)
« Reply #1 on: October 05, 2016, 08:49:26 AM »
Hi Phil

That sounds strange indeed. Could you please send me the command and data file?

Cheers
Eveline

Eveline Gebhardt

  • Administrator
  • Full Member
  • *****
  • Posts: 103
    • View Profile
    • Email
Re: Different case numbers in output files (and data file)
« Reply #2 on: October 05, 2016, 08:50:40 AM »
P.S. Please send both command and data files to gebhardt@acer.edu.au.

Alvin Vista

  • Administrator
  • Newbie
  • *****
  • Posts: 18
    • View Profile
Re: Different case numbers in output files (and data file)
« Reply #3 on: November 28, 2016, 04:41:08 AM »
Hi Philipp,
ConQuest treats the pid variable as the identifier that links data from a single case. In your dataset, your pid variable is repeated in study 1 and 2. I believe this is by design, as you might want to look at the effect of 'study'. In other words, you might later want to include 'study' as a facet in your model.
Rest assured, all information was used in calculating the parameter estimates, and ConQuest has correctly indicated that the information came from 481 cases (although each case may have appeared in more than one instance). ConQuest also computes ability estimates (e.g., WLEs) for each set of responses, regardless of whether they are from the same person. This is not an issue in your case because the responses occurred at different timepoints and was according to your design.

Cheers,
Alvin