ConQuest Community
Help desk => Questions and Answers => Topic started by: oysteg on July 04, 2015, 11:40:23 AM

Hi
How can I (post hoc) identify response dependence between pairs of items in ConQuest? Is there a way to resolve response dependence by e.g. forming subtest structures? (Please direct me to any tutorial or note I might have missed).
Best regards, Nanco

Hi Nanco
Yes, that is possible in ConQuest. You use the FIT procedure for this. This command is decribed in the Command REference Guide. Please let me know if you need further assistance.
Best wishes
Eveline

Hi
Thanks for your response. I am new to ConQuest. I took a look at the Command Reference Guide but I am still not able to 1) run an analysis to check for response dependence or 2) create a subtest structure accounting for possible response violation of local independence. I would be happy if anyone could post examples of command files. Best regards, Nanco.

Hi Nanco
After the estimate command you can add commends like these:
/* local dependence between item 1 and item 5 */
fit 1,5 >> pair.fit;
/* local dependence within two sets of 5 items */
fit 15:610 >> sets.fit;
First check if the weighted infit meansquere is significantly larger than 1. If it is significant, a practical criterion for local depence is a weighted infit MNSQ of 1.3.
Eveline

Is there a command for outputting the residual correlation matrix in ConQuest (to check for dependent item pairs)? Are principal component analysis of residuals and eigenvalues available?

Hi OG, sorry not in ConQuest.

Hi Nanco
After the estimate command you can add commends like these:
/* local dependence between item 1 and item 5 */
fit 1,5 >> pair.fit;
/* local dependence within two sets of 5 items */
fit 15:610 >> sets.fit;
First check if the weighted infit meansquere is significantly larger than 1. If it is significant, a practical criterion for local depence is a weighted infit MNSQ of 1.3.
Eveline
Hi,
If I run the 'fit' command without arguments (e.g., fit >> fit.fit), what exactly does the output show me? I thought that it would show me fit values for item pairs. I ran the command on a 6item scale and got 36 fit values, which makes sense, but it does not seem to be the case that the fit value for item pair 1 and 2 is the same as for 2 and 1, which I thought would be the case. I could not find a good description of the analysis and output in the manual or command reference.
PerErik

fit >> fit.txt
This exports the fit for each generalised item (i.e., the default design matrix). Each row is an item, and this is the same as the fit statistics shown in the "show file".
Fit statistics for item pairs, would show you the fit of these two items, relative to the rest of items in the model. This kind of analysis is typically used to interrogate whether pairs of items belong together (e.g., to assess dimensionality or dependence). This would look like:
fit 12:34 ... >> fit.txt
(that is, the fit of the sum of items 1 and 2 in the default design matrix, relative to the rest of the items)
More complex fit analysis can be done by importing a custom design matrix:
https://conquestmanual.acer.org/s400.html#fit (https://conquestmanual.acer.org/s400.html#fit)
more complex