The Tversky Condition Applied to Rating Scales

Abstract

This study is the first empirical investigation of the 1964 Tversky condition applied to rating scales. The Tversky condition posits that the 3-response format will be optimum if testing time is proportional to the length of the test. To this end, 2-, 3-, 4-, and 5-response category forms of a 10-item measure of attitudes in science were randomly administered to 241 third grade students. Reliability and validity were computed for each form. No significant differences were found among the reliability coefficients or among the validity coefficients. The Tversky condition was not confirmed for rating scales. These findings are consistent with results from other studies regarding the lack of substantial differences among reliability and validity coefficients as the number of response categories in a rating scale are varied.