Today I want to say something that amounts to little more than a good bitch about an issue of relevance only to academics (and even then, often only marginally so). You've been warned.
I hate rational choice-based social science research. I really do. Rational choice models, like 99% of what social scientists do, are simply an outdated technique stolen from economics and ham-fistedly rearranged to accomodate non-economic behavior. Constantly dealing with rational choice-based political science is probably the most irritating part of what I do in a professional setting.
Rational choice has its place. It does. It really does. But that place is not "everywhere" and at all times. Unfortunately that is the place it appears to have assumed. I despise it because rational choice social science is a rigged system that plays to the basest Awe of Science and Big Numbers in uninformed people and usually amounts to no more, in my opinion, than unfalsifiable pseudoscience.
The inherent flaw is the idea (implicitly encouraged by some practitioners but explicitly stated by others) that everything can be quantified. We all understand where the Prisoners' Dilemma falls apart – if the crooks are Mafiosos, they're not going to talk. They value Not Dying more than Utility Maximization. Rational choice research too often picks up at this point and, like a snake-oil salesman, soothingly tells you not to worry because Fear of Dying can be incorporated into your model as a latent variable.
The second I read something and see a "utility" variable (or, even more disingenuously, a quantitative variable used as a "proxy" for utility) I realize that it has become safe to stop reading. It is usually a cheap effort to continue applying a rational choice framework to a situation to which it does not apply. Far, far too often these arguments quickly reduce to tautologies.
If Frank is a Union plumber and he has a choice between the pro-Union Democratic candidate and anti-Union Republican, rational choice theory suggests quite obviously that he will choose the candidate who maximizes his personal utility. Fine. If Frank does choose the Democrat, then everything works out swell. Unfortunately sometimes Frank chooses the Republican, and that's when the bullshit starts to fly. Start making up "qualitative" variables and "measures of utility" and, voila, you've got a published paper.
You see, if we add in a variable for "social conservatism" (either a fabricated measure made from sunshine and farts or a "stand-in" like self-reported church attendance) and suddenly Frank's Republicanism is Rational After All. Amazing!
Any 18 year-old who has passed an Intro to Logic class knows about Moving Goalposts as an inherently flawed rhetorical technique – i.e., Our justification for invading Iraq was WMD, but if we don't find any then our justification was Liberating the Iraqi People, but if they resent our presence then our justification was Ties to Al-Qaeda, but if there were none then our justification was X or Y or Z. This is essentially what the "turn it into a variable" game does in rational choice. We arrogantly assume that we know Frank's decision must be rational, so if he makes a decision which appears irrational based on our theory we are simply looking at the wrong variable. So change the variable until we find one that works in our model to "prove" just how rationally Frank is acting.
This is the Webster's definition of a tautology. Such logic assumes its own conclusions – that individuals act rationally is both the conclusion and the basic premise of rational choice models. Frank will choose the candidate who maximizes his utility. We thought, from his Union membership, that the Democrat would be the choice. But we were wrong: apparently the Republican is the candidate who maximizes his utility. How do we know? Because that's the one he chose. I cannot emphasize strongly enough how often this logic presents itself or is implied in research I read.
It gives me the feeling that researchers create a hypothesis, do some hesitation wound-style "pre-testing" research, and then go back to create their model and choose their variables once they see the results. The end result is a bunch of cocky assholes who gloat about the glories or rational choice, the perfect theory that essentially Explains Everything. Whenever they are wrong, they simply add a new variable that will give them the conclusion they want and start over. Thereby can this simple theory, originally intended to explain a narrow range of economic decisions, be made to explain the entirety of individual and collective human behavior.
And don't even get me started about the New Coat of Paint technique of calling it "bounded rationality" to milk 10 additional years of publications out of a stale idea.