Data, statistics, and theory: A comment on Bates, McDonald, MacWhinney, and Applebaum’s “A maximum likelihood procedure for the analysis of group and individual data in aphasia research

Although there certainly is much room for disagreement on specific cases, all scientists will undoubtedly agree that not all putatively theo-retically relevant observations are in fact so, and, more importantly, that not all forms of inferences from observation to theory (or vice versa) are equally acceptable. Indeed, a significant part of scientific practice in all sciences is concerned with distinguishing between theoretically useful and nonuseful observations, and between valid and invalid arguments in rea-soning from theory to data, and vice versa. From time to time, when. someone attempts to distinguish explicitly between valid and invalid meth-ods in a scientific domain, these concerns become the focus of heated, public controversy. This seems to have been the case with my colleagues’ and my effort to define the range of valid inferences that may be drawn from the performance of brain-damaged subjects for hypotheses abou