ABSTRACT
This chapter introduces the reader to the important role that item analysis plays when exploring language test data. This analysis is usually the first statistical procedure to take place once the test data have been entered into a program such as IBM SPSS or Winsteps. It provides three types of information: first of all, it indicates the difficulty level of each item (also referred to as the facility value); in other words, it tells the researcher how many test takers have answered the item correctly. Second, the analysis provides insights into how well the items discriminate between the test takers and thus whether the stronger ones are answering the items correctly and the weaker answering them incorrectly. Third, it demonstrates the degree to which each individual item is contributing to the task’s internal consistency reliability, that is, whether the items are measuring the same construct. A sample study, based on a task consisting of 11 multiple-choice items, which was given to 102 test takers, is used to illustrate these three procedures of item analysis.
