Are sciences more of a continuum from hard to soft than those two as discrete categories? How? Why? I think the former. My book here somewhat touches on that question:
Book Respectfully and Unfortunately, The Improbability of, and Da...
This continuum would certainly apply to psychology. There is some use of mathematical models and a great deal of methodology applied in some cases but there are many non-quantitative hypotheses only tested statistically. Sometimes the concepts employed in these studies are not clearly defined. Worse still the sampling of subjects is opportunistic and stimuli are often selected on an ad hoc basis. Mostly it is not tight (hard) science.
I am not sure whether sociology should be included in the discussion of science, in which case you need to go beyond soft to hairy. I have met quality material but innumeracy and and lack of clarity are all too common features of the thinking found here.