Strong words or moderate words: A comparison of the reliability and validity of responses on attitude scales

Abstract

A common assumption in attitude measurement is that items should be composed of strongly worded statements. The presumed benefit of strongly worded statements is that they produce more reliable and valid scores than statements with moderate or weak wording. This study tested this assumption using commonly accepted criteria for reliability and validity. Two forms of attitude scales were created--a strongly worded form and a moderately worded form--measuring two attitude objects--attitude towards animal experimentation and attitude towards going to the movies. Different formats were randomly administered to samples of graduate students. There was no superiority found for strongly worded statements over moderately worded statements. The only statistically significant difference was found between one pair of validity coefficients (r = 0.69; r = 0.15; Z = 2.60, p [less than or equal to] 0.01) and that was in the direction opposite from expected, favoring moderately worded items over strongly worded items (total scores correlated with a general behavioral item).