Continuation of Part 1
Social (un-)Science cannot be science
Having contrasted social science and the humanities with science, the next question is of course whether social sciences and the humanities should try to become a science. Can the Scientific Method and basic principles of science just be adopted or applied to social sciences?
I have realized the answer to this in the general case is a resounding No! There may be particular situations and questions within the domain of social sciences with can be addressed using science but as the foundation of the field it would not work because of fundamental differences between science and social science and the humanities. Although the dividing line is fuzzy with many fields of research falling in between science and social science (eg. psychology/neuroscience, and medicine).
The first fundamental difference is the problem of experiment. Many aspects of social science deal with such large systems (eg. a national economy) experimentation is impossible. This issue is not unique to social science: ecology, climatology, and astronomy all deal with very large systems in which experimentation is difficult or impossible and yet they manage to overcome these issues by the extensive use of mathematical models to get the most of the few experiments they are able to do. However, the major problem is that social sciences and the humanities study humans. You'll notice the science-y fields I place in the grey area between science and social science (psychology and medicine) also deal primarily with humans. Studying humans creates massive ethical concerns and restrictions which limit possible experiments.
Beyond the ethical restrictions, there are many more issues with studying people. Firstly, humans tend to know when they are being experimented on and will often want to understand what the experiment is about causing them to change their responses. One of the clearest demonstrations the effects knowing you are part of an experiment is the placebo effect in clinical trials. I can only speculate how knowing you are participating in an experiment affects economic behaviour, speech patterns, or interview responses. However, it is highly unethical to experiment on people without their knowledge.
However, in my opinion the most fundamental issue with applying the scientific method to social sciences is that the academic field is inseparable from the thing it studies. The social sciences study society but at the same time are part of society. This is not the case for the sciences. The reason this is a problem is that it undermines the ability to verify/refute predictions which is the most important aspect of science.
To explain this I will use two examples one from science and one from social science. For science we will consider a model predicting cod stocks will be slow to rebound after the halting of fishing. For social science we will consider the prisoner's dilemma which predicts people should always defect in a given situation.
In the cod stock situation it should be obvious that making the prediction has no effect on the outcome of any experiment designed to test the prediction. The cod don't know about the prediction so cannot adjust their behaviour to either support or refute it. However, in the prisoner's dilemma case many many people have heard about the prisoner's dilemma know what the predicted behaviour is and the theory behind why (regardless of what the other player does the defecting will result in a better outcome for the player). Because any experiment to test the prediction will necessarily use people (and typically use university students who are even more likely to have heard of the prisoner's dilemma before) it is entirely possible, even likely, that the prediction will become self-fullfilling - the subjects will modify their behaviour in response to their knowledge of the prediction.
Indeed, a recent report seems to demonstrate this quite clearly, students exhibit the expected behaviour: defecting in the simultaneous case and being more likely to cooperate in the sequentional case; whereas actual prisoners (less likely to have learned about the game before) do not: cooperating roughly half of the time regardless of the version of the game played.
Even social science experiments not performed on their own undergraduate students are likely to be victims of some kind of placebo or circularity effect since most journalists, media personalities, politicians, and social leaders (the people who shape and influence society) studied social sciences or humanities when they attended university.
