The Stanford Achievement Test, Form J, was administered by classroom teachers, consisting of a battery of six sub tests: Paragraph Meaning, Word Meaning, Spelling, Language, Arithmetic Computation, and Arithmetic Reasoning. All of these sub-tests involve reading except Arithmetic Computation. Scores are stated in grade equivalents on a national norm. The battery median grade equivalent was used in data analysis in this study.

The Wechsler Intelligence Scale for Children was administered to each sample third grade child by a clinical worker. The relationship of intelligence test scores to school achievement is a well established fact (in this case, ** f); therefore, in the investigation of the present hypothesis, it was necessary to control this factor.

The criterion score used in the statistical analysis is an index of over -- or under-achievement. It is the discrepancy between the actual attained achievement test score and the score that would be predicted by the I. Q.. For example, on the basis of the regression equation, a child with an I. Q. of 120 in this sample would be expected to earn an achievement test score of 4.8 (grade equivalent). If a child with an I. Q. of 120 scored 5.5 in achievement, his discrepancy score would be +.7, representing.7 of one year of over-achievement. A child with an I. Q. of 98 would be expected to earn an achievement test score of 3.5. If such a child scored 3.0, his discrepancy score would be --.5, representing.5 of one year of under-achievement. In this manner, the factors measured by the intelligence test were controlled, allowing discovered differences in achievement to be interpreted as resulting from other variables.