I have been prompted in the last few days to think about the dreaded question of "What's it all about anyway, when you get right down to it? What's the point?". I'm not talking about the meaning of life (which I have figured out already) but social science, and especially political science. The issues that I have been focusing on are the idea of 'cumulation' and the implications of a constructionist position for the aim of social scientific research. If that sounds ambitious, I am thinking about it in quite basic and unsystematic terms.
Constructionism includes the idea that social phenomena have no existence apart from what they mean to people. So, if everybody stops thinking about and talking about and acting as if a particular university exists, then it does not. Also, if the idea of a university is not part of the total social cluster of ideas, then it is not possible (or rather, highly unlikely) that anyone would start to think, talk, and act as if there was a university. Finally, what it is that a university means, the entirety of the actions and ideas that are associated with a university, is not unitary. By this I mean that you could have a society in which university is something slightly different from another society. This is in fact the case in the world - being at a liberal arts college in Massachusetts is a very different proposition from being at a massive regional normal university in China. So, does it make sense to aim at making general propositions about, say, the effect of having a university degree? Economists do this all the time (I imagine). If by general proposition, I mean a proposition that applies just as much in 11th century Bologna as it does in 21st century Washington DC, then it is clearly preposterous to say that it is a worthwhile enterprise. There is a clear analogy to international political phenomena, as war today involving European powers is different enough from war in mediaeval Africa for us to say that propositions about the causes, consequences, and practice of war must be bounded at least partly by the spatio-temporal context.
If this is so obvious, why would anyone think otherwise? The model of the natural or physical sciences is a powerful lure. One of the things that it is said that the physical sciences have is 'cumulative' knowledge. I am not sure what this means. If it means that later scholars use previous scholars' work in their own work, then anyone has that. If it means that scholars do not challenge earlier scholars' work, then this is patently false as scientific breakthroughs can take the form of discovering that the previous ideas were all wrong. A very convincing account of why there is this high-consensus, rapid-discovery science in the natural sciences is Randall Collins' 1994 article (“Why the social sciences won’t become high-consensus, rapid-discovery science.” Sociological Forum 9, no. 2: 155-177) where he attributes these features of the natural sciences to the appropriation of research technologies. This makes natural science look useful to outsiders (because of spin-off technology, like the internet) and directs attention away from continually revisiting and challenging the work of previous scholars. All of this is unrelated to epistemological validity, i.e. is physics more right than sociology, which I would say is a different question. So, my contention is that cumulation is not a goal that I would sacrifice very much to attain.
But cumulation is not the only thing that 'naturalists' (those wanting social science to be like natural science) want. They also want prediction. Much of the prestige and the justification for believing the natural sciences to have it right comes from being able to say stuff like, "I swing this ball on a rope here at 0.02 millifrutors and the magnesium hydrosulfate will turn green in 3.4 seconds" and then it does, again and again. Social scientists could do this sort of thing as well, like the prediction that 200 or whatever million Americans will get up tomorrow morning and go to work, many of them driving on the right-hand side of the road to get there. For some reason (and I think this reason is actually much more important than other people think it is) this is not impressive to anyone. Polisci also does other types of prediction; see Abramowitz, Alan I. 2008. It’s About Time: Forecasting the 2008 Presidential Election with the Time-for-Change Model. International Journal of Forecasting 24 209-217. This is very limited compared to what the natscis are able to demonstrate in laboratory or experimental settings, but it is not impossible in principle to use social scientific theory to predict human behavior. Again, I'm not sure that this kind of prediction is all that, or even largely what, we as social scientists should be doing.
So, why is constructionism such a threat to naturalism? Partly because many of the phenomena that political scientists study change out of all recognition. The state at all did not exist prior to about the 14th-15th centuries and the 18th century european state is so massively different now that any propositions about the causes or effects of political units of that type are unlikely to apply to anything now or in the future. It is not just acceptable that we can say that there are a set of starting conditions and whenever these starting conditions obtain, their effects also obtain. This happens in natural science, even in cases like geology or evolutionary biology they are working with similar types of evidence as social scientists, but the difference is that there are recurring starting conditions. In social science, if constructionism has any force, sets of starting conditions are only similar within tightly bounded social contexts (which are themselves bounded in space and time). This is why much of social science cannot have the same properties as natural science.
What does this mean for how we use social scientific research? I don't know, but I'm thinking about it.