Ooops...self-inflicted perils in data collection

Evidence on how the very act of conducting a survey can affect our estimates of the impact of a policy. This adds to the list of effects we have to be mindful of when using primary or secondary data on agent behaviour: the Hawthorne Effect (when participants in a study alter their behaviour in response to being part of the study) and the John Henry Effect (when members of the control group perceive to be in competition with members from the treatment group and change their behaviour accordingly)...