Item Analysis

Item analysis is the set of qualitative and quantitative techniques and procedures used to evaluate the characteristics of items of the test before and after the test development and construction. An item is a basic building block of a test, and its analysis provides information about its performance. Item analysis allows selecting or omitting items from the test, but more important, item analysis is a tool to help the item writer improve an item. Anthony Nitko suggests that some classroom uses of an item analysis might be estimating whether an item functions as intended, providing feedback to students about their performance, providing feedback to the teacher about student difficulties and ideas of curriculum improvement, revising assessment tasks, and improving item-writing skills.

Item analysis can be used both for dichotomously scored (correct or incorrect) items and polytomously scored (with more than two score categories) items. The main purpose of item analysis is to improve internal consistency or internal structure validity, focused on confirming a single-factor or one-trait test. If the trait is not one factor, then the use of item analysis might tend to lower validity. If a test has two factors (or content divisions) or is multifactored (more than two content divisions), the calculation of item statistics for each item (or option) should be focused on the subtotal for the relevant set of items rather than on the total test score. Item analysis in this case is used to improve the internal consistency of each subset of items with no intention to change the dimensionality of the entire set. In these cases, an overall reliability index would be stratified alpha (or battery alpha) rather than the regular coefficient alpha as used for a one-dimensional test.

A test that is composed of items selected based on item analysis statistics tends to be more reliable than one composed of an equal number of unanalyzed items. Even though the process of item analysis looks sophisticated, it is not as challenging as it seems. Item analysis software programs are user friendly and make the task much simpler. The most time-consuming part of item analysis might be the process of tabulating the data to be analyzed by the software program. Statistical programs such as SPSS, an IBM product, and SAS can be used for item analysis as well. The information needed to start an item analysis procedure after the administration is the work of the examinees. In the subsequent sections, several aspects of the item analysis process are discussed.