This study examines the effects of (1) generality equivalence between attitudinal and behavioral entities and (2) the order and proximity of the attitude and behavior questions on the observed relationship between attitudes and behavior. The behaviors examined are sentence recommendations for six hypothetical defendants. The sample is composed of 185 line probation officers from New York State. The variables chosen conform to Martin Fishbein's model of behavioral intentions. As predicted, the observed relationship is affected by both level of specificity and proximity of the questions. Order of the questions has little effect.