Analyzed social judgment theory of attitude change and studied placement of a communication and opinion shift in an experiment on visual perception. In a 3 × 3 design plus a control group, 110 undergraduate college students (50 men and 60 women) received a communication after estimating the number of dots on 14 slides. The communication varied at three levels of ambiguity and three levels of discrepancy. Ss gave estimates of the communication from memory (measuring assimilation) and gave a second set of their own estimates (measuring opinion shift). Assimilation varies directly with ambiguity (p < .00l) but is unrelated to discrepancy. Opinion shift varies significantly as a function of discrepancy (p < .00l) but is unrelated to ambiguity. Authors hypothesize assimilation and opinion change may be negatively correlated at small discrepancies but directly correlated at large discrepancies.