Allures of Power Data

The experiment code, data collected from subjects, and analysis files posted here were used in a paper with the following abstract: This paper investigates intrinsic preferences for power over others' outcomes using a laboratory experiment. I decompose the intrinsic value of power into three components: self-recognition (the power-holder's awareness of her own power), other-recognition (others' awareness of the power-holder's power), and having options over others' outcomes. Using a combination of experimental games, including a novel "Points Power Game" in which subjects can pay to control an outcome that has no material consequences for their partner, I find that a substantial majority (86%) of subjects exhibit behavior consistent with intrinsic preferences for power by at least one criterion. While intrinsic preferences for power are widespread, I find little evidence that self- or other-recognition increase willingness to pay for power. The estimates are precise enough to rule out moderate effects, although this may reflect limitations of the recognition manipulation rather than a true absence of effects. In contrast, I find suggestive evidence that subjects intrinsically value having options over others' outcomes: 49% of subjects purchased more power than needed to achieve their chosen outcome. Subjects appear to place greater intrinsic value on options to increase others' outcomes than on options to decrease them. These findings offer preliminary evidence that the mere possession of options over others carries value beyond what distributional preferences alone can explain, although alternative explanations cannot be fully ruled out. The experiment files is the z-tree file "experiment_code.ztt." The data is in the Excel spreadsheet "data.xlsx." The main analysis code is the Stata file "stata_analysis.do," which requires "common_round.dta" and "stata_setup.do" to run. The Bayesian analysis, kindly provided by an anonymous reviewer, is in the Python file "Bayesian.py."