This study attempted to control differences in achievement when examining omitting tendencies of examinees. Test data of randomly sampled examinees (7 samples of 2,000 examinees each) from one national administration of the ACT Assessment were used. The number of responses omitted per examinee was examined over all examinees and over only those who omitted responses. The relationship between test scores and number of omits was negative and weak. More examinees at lower levels omitted responses (and more of them), but some at higher levels also omitted a surprising number of responses. No significant differences in tendencies to omit were found between males and females. Statistically significant differences in the tendency to omit were found among examinees of different ethnic groups, especially between African origin and Caucasian origin samples and between Hispanic origin and Caucasian origin examples, but the differences were all less than one omit on average on the mathematics test. Controlling for potential achievement level differences between ethnic groups by using covariates resulted in only a minimal reduction in omitting tendencies between the groups, suggesting that this is not the explanation for differences between ethnic groups. Ten tables summarize study findings. (Contains 22 references.) (SLD)