Validation of the Coleman Readability Formulas

Abstract

The following four readability formulas developed by Coleman were cross-validated: (a) X′ = 38.45 + 1.29 X1, (b) X′ = −37.95 + 1.16 X1 + 1.48 X2, (c) X′ = −34.02 + 1.07 X1 + 1.18 X2 + .76 X3, (d) X′ = −26.01 + 1.04 X1 + 1.06 X2 + .56 X3 −.36 X4, where X′ is predicted cloze score in percentage of correct insertions for a 100-word passage, X1 is its number of 1-syllable words, X2 is its number of sentences, X3 is its pronouns, and X4 its prepositions. The formulas were cross-validated by obtaining cloze scores for 7 new passages from 20 Ss matched with Coleman's original Ss. The rs between obtained cloze scores and those predicted from Formulas a, b, c, and d were .83, .88, .87, and .89.

A computer readability formula designed for machine scoring.

“…From the original data for Miller and Coleman's (1967) 36 passages, a new prediction equation was computed with the predictors being letters per 100 words and sentences per 100 A multiple R of .92 may seem rather high, but it stould be noted that these passages cover an unusually wide range, stretching from first-grade mati rial to extremely difficult technical prose, A ci oss-validation study by Szalay (1965) for a simi ar readability formula derived from this scali; gave support for an R of this magnitude. The multiple R for that formula was .897, sligl tly less than ours since it used a less accurate indi ;ator of word difficulty (number of onesylli ble words in the passage).…”

Modification of the Coleman Readability Formulas

“…Coleman reported respective multiple R's for them of .85, .89, .90, and .91. The R's can be accepted as accurate since they were verified by a cross-validation study by Szalay (1965). Although Szalay used a new sample of passages as well as a new sample of subjects, the correlations between his obtained cloze scores for a passage and its cloze scores predicted by the formulas were respectively .…”