Since an att-object must always be encountered within some situation, about which we also have an att, soc behavior must be a function of at least 2 att's: att-toward-object (Ao) & att-towardsituation (As). Opinion expression & changes in opinion expression are both forms of verbal behavior &, thus, also functions of Ao & As. When a change of opinion is empirically found it may arise from a change in Ao, or As, or both, or neither. The classical paradigm of attitude change studies-pre-test, treatment, posttest-cannot ascertain which of these 4 possibilities produced the opinion change & must therefore be discarded. 3 new methods are then proposed for determining whether opinion change does or does not represent an attitude change: test for opinion change in at least 2 diff post-test situations; test for diff'ial change in several opinions in one post-test situation; test for behavioral changes accompanying opinion change. Illustrative res findings are cited. AA.