Reliability; Test Validity IDENTIFIERS *ikilAcrimination Index

A study examined the nature of the questions contained in two major standardized reading comprehension tests in terms of their centrality to the text. It was hypothesized that the use of a discrimination index for item selection would tend to favor relatiVely trivial questions. Half the reading selections from the Stanford Diagnostic Reading Test (SDRT), which gives more weight to item discrimination indexes, and the Metropolitan Reading Test (MPT), which does not, were randomly selected. The multiple choice questions' that followed each reading selection were transformed into statements by adding the correct answer to the question stem. Thirty faculty and graduate students from a school of education read each of the 10 passages and were asked to rate\each statement on a four-point scale as to whether it was central or peripheral to cOmprehension of the passage. Thirty-seven percent of the items from the SDPT were rate