There seem to be three levels at which inquiry relevant to the composing process has gone on. One level consists of personal reflections, often by writers themselves (see, for instance, Writers at Work, 1963; a critique of this tradition is given in Bereiter, 1984). Another, carried out mainly by educational researchers, generally follows the empiricist conventions popular in all the behavioral sciences up until the 1970s, of defining quantifiable variables and testing statistical hypotheses about them. (An influential monograph, setting forth principles and topics for this kind of research on writing was Braddock, Lloyd-Jones, & Schoer, 1963.) And then, of course, there is the long tradition of literary scholarship, usually devoted to the close analysis of individual texts. In this chapter we define three additional levels of inquiry that are important to understanding the composing process—process description, theory-testing experimentation, and simulation. These are commonplace activities in cognitive research, but just beginning to be found in writing research. Our argument is not that these three kinds of inquiry should replace the earlier three. On the contrary, the argument is that all six are needed, with a proper understanding of the role of each.