We have been able to identify in this study a significant source of instrument error in one of the traditionally used question formats of prejudice research. Moreover, we have shown that such errors are specific to certain item contents and to certain answer options. Further, we have demonstrated that qualified answer options can evoke meaningful and circumspect usage; and that respondents are more dissatisfied with the forced response format than the qualified response format. While this study cannot provide a precise estimate of the magnitude of instrument errors, it does appear that the forced response format overstates the degree of acceptance of prejudiced statements. Given the only moderate correlations that have been obtained between most measures of prejudice and other variables, it is clear that the magnitude of the differences observed in this study is sufficient to radically alter such correlations.