And then there's this article, which I suspect it's possible that both of us will agree with: http://blogs.scientificamerican.com/cross-check/is-social-science-an-oxymoron-will-that-ever-change/
Even as biological sciences start folding in lessons from data analysis that is more typical of social sciences, social sciences seem to be scrambling to adopt the language and methods of hard sciences, which I think is a mistake. Social science is statistics to hard science's calculus; both employ methodological approaches that are uniquely suited to different types of data, and both are advancing apace. You can even have the most complex of both worlds in fields like epidemiology and epigenetics. Copping the language of hard science by bastardizing its language in an effort to gain legitimacy is embarrassing and reflective of an inferiority complex that is, IMO, unnecessary.
I also think that the hangup on the idea of experimental data is a throwback to a more primitive era of science when, lacking sophisticated technology or data-collecting methods, it was the best we could do. Many gains in our knowledge of molecular biology and brain function are dependent on "experiments" that are simply advanced methods of observation. Does that render them not science? I would beg to differ, and as our observational technology improves, we will increasingly have methods for looking inside a living cell and observing mechanisms that could previously only be inferred from experimental machinations that attempted to approximate true observation of processes too small to observe directly.