AB Illusion

The A/B illusion is a hypothesized phenomenon in which individuals perceive the decision to run a randomized, controlled experiment (e.g., comparing two interventions, policies, or practices) on human subjects as less appropriate than simply implementing one of those alternatives without testing its effects. We are testing this phenomenon by using crowdsourcing platforms (mechanical turk, pollfish) to conduct large scale surveys with a variety of scenarios where policymakers either run an experiment between two policies or implement a single policy with no prior experiment. Work done with Duncan Watts @ MSRNYC and Chris Chabris and Michelle Meyer @ Geisinger.