What effect varying the phrasing of attitude statements would have across sex and grade was determined. Three content parallel forms were developed to assess student attitudes toward integration. Each form contained 15 items and differed only in terms of item phrasing either positive, negative or a combination of both. Six hundred and twenty-two, fourth, fifth, and sixth grade students responded anonymously to one of the three forms. The data indicated that positive, negative, and mixed item phrasing resulted in differential response patterns across grade level. In addition, differences were noted in the internal consistency estimates across forms. Implications of the study were discussed from a test constructor-test user perspective. (Author/GK)