Impossible experiments

The fantastic BPS Research Digest posed an interesting question to bunch of psychologists a while back: "What's the most important psychology experiment that's Never been done?" Now (the considerably less fantastic) Psychology Today has done something similar: they asked their resident bloggers to dream up the experiment they would most like to do if ethics or logistics didn't stand in the way. A couple of the answers were a bit, well, unimaginative (especially from the philosophers - tellingly, they proposed Baconian, not Galilean experiments). I'll echo Vaughn and say I most liked Bella DePaulo's proposal:

I'd like to take couples who are living together and randomly assign half of them to marry and the others to stay unmarried. Then we could really know something about the implications of co-habitation vs. marriage. More outrageously, take people who are not in a serious romantic relationship, and assign half of them, at random, to marry. Single people are randomly assigned to a spouse who is chosen at random, or to a spouse who fits their description of their perfect partner, or to stay single. Who do you think would end up the happiest a decade later? Same for divorce. If married parents are already at each other's throats, is it better for the children if they divorce, or stay together? Randomly assign half of them to divorce, and half to stay together; then we'll see. Now take married couples who say they are happy and are not considering divorce. Randomly assign half of them to divorce! Now who will be happier ten years hence?

From my recent post about the Stockholm Syndrome it should be clear what my answer to this question would be...

(Hat tip: Mind Hacks).