2
$\begingroup$

We administer a confidence survey (likert scale) comprising of 30 items and have conducted PCA to examine dimensionality, item independence and the like. We also conduct student interviews to assess how students are interpreting the statements, following which we make any revisions to the survey.

In addition to student interviews (where we explicitly ask students whether their response to a certain item is being cued by their responses to earlier, perhaps similar items and whether they are treating each item as independent), are there quantitative methods which can demonstrate that students are treating each statement without being primed or cued by their response to an earlier statement?

We were considering IRT for polytomous data (like the graded response model); is this a correct direction to proceed in? I can't think of another quantitative method besides IRT that would allow us to establish item independence and confirm / refute what we obtain from student interviews.

Thanks

Improve this question
$\endgroup$
1
  • $\begingroup$ I think it might help if you clarified what you meant by "independent". I get the impression you mean it as "considered separately on its own merits" rather than in the statistical sense - you don't seem to require the answers to be uncorrelated for example? $\endgroup$
    – Silverfish
    Commented Jan 10, 2015 at 9:49

0

Reset to default

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.