The Effect of Observation and Deception in Field Experiments: Evidence from a Two-sided Audit Study

One of the strengths of natural field experiments is that subjects’ actions are observed while they are unaware they are in an experiment. This might be particularly important for actions that are affected by implicit biases. One common natural field experimental design is the use of resume audit studies. In a traditional audit study the experimenter sends fictitious resumes with randomly varied traits (e.g. race, gender) to a subject pool made up of employers who are unaware they are in an experiment. Audit studies are sometimes criticized because they deceive subjects and do not allow subjects to consent to participate. An important question is whether the subject being aware they are in an experiment and dealing with fictitious applicants actually changes their behavior. We leverage a unique setting where we elicited responses from real recruiters about fictitious job applicants both with the deception usual of audit studies and with informed consent where the recruiters knew this was part of an experiment. We show that results differ when recruiters are aware they are taking part in an experiment and the subjects are fictitious.