A Responsive Note on Measures of the Passage Dependence of Reading Comprehension Test Items

In 2 recent issues of this journal, Pyrczak (1972) and Tuinman (1973-1974) have recommended 2 different indices of the passage dependence of reading comprehension test items. Both indices are designed to help test constructors identify faulty items during the item selection and item revision stage of test construction. Careful inspection of the 2 indices reveals that each index is based upon a different definition of passage dependence. The purpose of this note is to examine the definitions that underlie each index and their implications for the use of each index. It seems especially important to consider this topic at the present time, because professional test constructors are now undertaking systematic studies of passage dependence during the test construction phase of their work (e.g., Jean Garren, Zweig Associates, personal communication, 1975). Pyrczak's I index To obtain the data necessary to compute Pyrczak's I index, a group of examinees must attempt the items in the absence of the associated passages. For a set of items, each of which contains the same number of choices, I is a simple linear transformation of the proportion of the examinees who marked a given item correctly. The transformation is designed to take account of the most likely number of examinees who would mark the item correctly on the basis of guessing alone. Thus, for the purposes of this discussion, I may be thought of as an indicator of the proportion who marked the item correctly in the absence of the passage. In notation, I is defined this way: I = P,- Pm