The first thing that comes to mind for me, which is only touched upon in the paper, is that the subject matter of the social and natural sciences are different in quality. The subject matter of natural sciences react. Atoms, chemicals, cells etc. more or less do the same thing each time you measure them, intervene in them, control and manipulate them. But social science subject matter 'act', that is they have an influence back on the world that is not predictable, arising from their consciousness. Social scientists are also part of the subject matter they are studying (i.e. social beings). This has all sorts of implications. You can bracket off a very specific part of the natural world and have a specific idea about how it works, and test it or not. You can produce a hard and fast yes or not answer to this specific idea. But with social science, you are testing a wider, more loose and fuzzy idea which is informed by the fact that you are part of the subject matter itself, thereby having an idea how it works beforehand. This also links into what the article says about subject matter changing behaviour to the research's expectations, especially apparent in survey research which constitutes a sizeable amount of social science research. Finally, it would have been nice to have some commentary on the implications of interpretivist ontology (inc. interviews, ethnography etc). link into these findings and hypothesis-testing research. Dan.