Post

Region: Forest

Nattily dressed anarchists on bicycles wrote:~snip~

I mean, I would probably suggest the opposite to what you've said about the natural sciences, but we do agree in a general sense. Much of our current understanding of physics is based on supposition; it's a "scientific" subject to the extent that it's applied mathematics. Chemistry, and to a greater extent biology, I would consider to be more "scientific" in the sense that hypotheses in those subjects are deduced from real and reproducible observations which (if done properly) do not change based on the particular equation you choose. Biology is messy in the sense that the real world is often surprising and unexpected - but then, since the real world is unexpected, surely the best scientific method is able to deal with the unexpected instead of assuming that the unexpected will not occur.

Of course, such distinction is meaningless because none of those three subjects could exist without the others. From my personal experience, collaboration usually provides the best results possible (different methods, unique perspectives and all that) - and this extends to the social sciences too. The best research of modern psychology incorporates neuroscience, the best studies of history incorporate genetics, the best work in the field of anthropology incorporates zoology, the best work in archaeology incorporates organic chemistry etc.

That's real sh*t about your adviser - I know first hand that the person who supervises your research project can make or break the experience, and thankfully I've been more or less lucky so far. I'm sure you know that p-values are used outside of the social sciences and, while somewhat arbitrary, the line for statistical significance has to be drawn somewhere. However, any scientist worth their salt will tell you that a p-value of 0.0501 could fall either way - the answer is really to investigate further - and (although it would upset your advisor) p=0.0499 is exactly the same. But as you've already mentioned, f*ck your supervisor.

ContextReport