Consider an intervention that is ethical : we will mail every single resident of the state of New York a piece of paper that contains on it the gender of the two main candidates for NY governor. This is action A.
Consider a non-intervention that is also ethical: we will not mail ever single resident of the state of New York a piece of paper that contains on it the gender of the two main candidates for NY governor. This is action not-A.
Action A is ethical. Action not-A is ethical. On what principle would we argue that it is unethical to choose randomly which one to do?
Let’s take it further. Now we’re not mailing all New Yorkers, we’re mailing some New Yorkers. It’s ethical to mail all. One what basis would we argue that it is unethical to mail a randomly chosen subset of residents? Recall that it is also ethical not to mail them anything.
And finally, it is ethical to mail a randomly chosen subset of residents with the intent of comparing receivers and non-receivers?
The back story behind these thought experiments: I have read a lot of objections in the past day to the idea of field experiments at all because they involve assigning people to being treated without their knowledge. I want to isolate exactly what about that is problematic, where in the chain of choices the ethical problem emerges. Perhaps this can help to clarify. And I’m honestly interested if there’s a deep ethical principle that might be at work here.
Between writing this post and actually posting it, I’ve come to believe that action A above is not self-evidently ethical even if it is minimally invasive or harmful. The ethicality depends on the intent. If, for example, that mailer were sent out with the intent that it would discourage women from voting, then it would not be ethical.