Module 13. Classical Test Theory Item Analysis
In Module 12, we discussed how best to construct maximal performance (i.e., knowledge) tests. After you put in hours and hours (if not days and days or weeks and weeks) constructing such a test, the day will finally come when you actually have to give the test to someone. Once you administer the test to a designated group of test takers, you will want to evaluate it. That is, you will want to know if the test worked the way you hoped it would and if it is accomplishing what you set out to accomplish. If the test is not up to your high standards, are you going to simply throw out the entire test? We hope not. Instead, you will want to determine which specific items may be causing problems by performing an item analysis . You can then eliminate and/or replace the lackluster items or, better yet, revise and reuse the problematic items. Think about it, you just spent a lot of time and painstaking effort writing items to create your test, so you do not want to be throwing out items needlessly or, worse yet, indiscriminately. Thus, the questions become “Which items do I keep unchanged?” “Which do I throw out?” “Which do I try to salvage with some well-placed revisions?” The answers can be found in your favorite classical test theory item analysis statistics.