Experimental prudence in political science (Part II) 

 As I posted the other day, experiments in political science have great potential, but they have some unique risks as well, particularly when the manipulation may change the output of some political process.  What happens if your experiment is so successful (in the sense of having a large causal effect) that it changes the outcome of some election?   How would you explain such an outcome when you report your results?  "The manipulation produced an estimated increase in party A's support of 5000 votes, with a standard error of 250.  (Party A's margin of victory was 2000 votes.  Sorry about that.)"   This seems like a good way to alienate the public once word got out, not to mention your colleagues working with observational data who now have another variable that they have to account for in their studies. 


 Having said that, I am just an observer in this field, and I'm sure that many people reading this blog have thought a lot more about these issues than I have.  So, to continue the conversation, I'd like to propose the following questions:  

 At what point does an experimental manipulation become so significant that researchers have an obligation to inform subjects that they are, in fact, subjects? 

 Do researchers have an obligation to design experiments such that the net effect of any particular experimental manipulation on political outcomes is expected to be zero? 

 Would it be appropriate for a researcher to work consistently with one party on a series of experiments designed to determine what manipulations increase the probability that the party will win elections?  Do the researcher's personal preferences matter in this regard?  

 To what extent are concerns mitigated by the fact that, in general, political actors could conduct these experiments on their own initiative?  What if those actors agree to fund the research themselves, as was the case in the 2002 Michigan experiments? 

 If a university were to fund experimental research that was likely to promote one political outcome over another, would it risk losing its tax-exempt status?  This one is for our resident lawyer....