We introduce a new instrument to evaluate the impact of behavior on outcomes when the behavior may be a function of unobserved variables that also affect outcomes. The instrument is introduced through a test of patient sensitivity to increases in the quality of care provided by doctors. We utilize the Hawthorne effect, in which the very presence of a research team causes doctors to provide measurably superior quality care for any type of patient to show that patients respond to this increased quality and are more likely to be very satisfied. Using the Hawthorne effect as an instrument allows us to examine the responsiveness of satisfaction to improvements in quality despite the fact that patient satisfaction is subjective and jointly produced with quality during the course of a consultation.