JOURNAL OF EDUCATIONAL MEASUREMENT VOLUME 5, NO. 3 FALL, 196B CLOZE TEST READABILITY: CRITERION REFERENCE SCORES JOHN R. BORMUTH University of Chicago The principle purpose of this study was to determine for cloze readability tests, a set of criterion scores comparable to the criterion scores used with oral lead!~g tests employed to measure the readabilities of passages. A secondary purpose was to add to the already accumulated evidence that cloze tests measure the compre- hension difficulties of passages. In practice. Unfortunately, present readability formulas are too inaccurate to be of much help in selecting appropriate instructional materials (Chall, 1958). Trying out materials directly on students by using oral reading tests, when done carefully, yields more accurate results but requires a level of technical competence possessed by few teachers, and the individual testing involved makes this method prohibitively expensive. In effect, educators have no practical means of solving this problem. In Research. Researchers attempting to improve readability formulas share a similar problem. Whether they use multiple-choice or orally administered comple- tion questions as their criterion measure of passage difficulty, they are plagued by the question of whether measured difficulty is really a measure of passage difficulty or of test difficulty. Their research is further limited by the expense of preparing such tests. These factors have combined to stifle research in the critical areas of readability prediction and control. CLOZE PROCEDURE Advantages o[ Cloze Tests. Tests made by the cloze procedure have done much to solve these problems. A cloze test can be made over any passage by replacing every fifth word with an underlined blank space of a standard length. Subjects are instructed to write in each blank the word they think was deleted. Responses are scored correct when they match the deleted words. Tests made by this procedure are simple and economical to prepare and do not confound the passage difficulty measurements with the difficulty of the language and other characteristics peculiar to the test questions, themselves. The research demonstrating the validity of the cloze procedure is now substantial. An excellent review has been written by Rankin (1965). Criterion Reference Scores. The usefulness of cloze tests is limited by the fact that we have no frame of reference by which we can place a value judgment on a cloze test score. There is a rule of thumb used with oral reading tests. Material is said to be sufficiently understandable to be used in a student's supervised instruc- tion if he can answer correcly 75 per cent of the questions on a test over it, and can correctly pronounce 95 per cent of the words as he reads it. Passages at this and easier levels are said to be at the student's instructional level. Passages are