The purpose of this paper is to fill the gap of no discussion about alternatives to the raising analysis of modal auxiliaries like must, may, ... etc. When we analyse the behaviors of modal auxiliaries, we have used a traditional method of the raising analysis of modals. But we did so without explicit motivation and discussion of alternatives. We need some discussion of alternatives. During the discussion, we compare the empirical predictions of the raising analysis (type) with the predictions of several different kinds of non-raising analyses. We find that the behavior of the modal auxiliaries is indeed accounted for more successfully by the raising analysis than by any of those alternatives. In addition, we have seen that there is a way to account for both de re and de dicto readings within an analysis of modals that does not involve raising of the subject from an underlying position below the modal nor any covert lowering-operation like reconstruction, but treats the subject as a genuine argument of the modal. The price we had to pay was a rather complicated lexical type for the modal, and some type-shifting operation affecting either the modal or the DP.