A Direct Comparison of the Predictive Capabilities of Two Cloze-Derived Readability Formulas
Two Cloze-derived readability equations were cross-validated on three Cloze reading comprehension tests to determine if they differed significantly in their predictive capabilities. The equations were Coleman's Formula 4 and Bormuth's Machine Computation Formula; the tests were the 1969 Bormuth Cloze Reading Comprehension Test and the 1967 Miller-Coleman Readability Scale (600 response and 2400 response conditions). Results indicated that: each formula's scores correlated significantly with reader scores on each test ( p < .001); each formula generated statistically equivalent coefficients on the Miller-Coleman 600 response and 2400 response tests ( p > .05); the Bormuth formula generated a significantly higher validity coefficient on the Bormuth test than the Coleman formula did ( p < .001).