Who Punishes Leaders for Lying About the Use of Force? Evaluating The Microfoundations of Domestic Deception Costs

Abstract

It is a common stereotype that leaders lie, but for all our beliefs about how normal it is for a president to lie to the public we know next to nothing about how voters might actually view this conduct. Drawing from literature in behavioral economics, we theorize that voters apply their attitudes towards interpersonal lies when judging leaders, with people who see lying as more socially acceptable being less willing to punish leaders for exhibiting this behavior. Using a novel vignette-style survey experiment, we find strong support for our theory: despite widespread willingness to punish leaders who lie, individuals high in Machiavellianism and Self-Monitoring are far less willing to punish presidents for lying than their low-type peers, and in some cases do not punish at all. Using robustness checks and causal mediation analysis, we confirm that views regarding the moral acceptability of lying drive responses to presidential deception.

Public Reactions to Secret Negotiations in International Politics

“…Recent experimental work provides micro-level evidence that the public is less favorable towards leaders who lie, mislead, deceive, or conceal information from their publics in foreign affairs. Yarhi-Milo and Ribar (2023) show that the American public punishes leaders for lying in international politics, although there is variation in the extent to which individuals perceive lying as acceptable. Likewise, Maxey (2021) finds that U.S. presidents tend to incur political costs for misleading the public about military action.…”

“…There is limited political science research concerning how the public perceives transparency and secrecy in foreign affairs. A small literature explores public attitudes towards secrecy and deception in the context of military action (e.g., Carnegie, Kertzer and Yarhi-Milo, 2023; Maxey, 2021; Myrick, 2020; Yarhi-Milo and Ribar, 2023) rather than international negotiations. 4 It is not clear, however, whether we can extrapolate findings about military force to perceptions of secrecy in other areas of foreign policy, such as diplomatic negotiations.…”

“…To address this gap in the literature, this paper complements a handful of recent studies that use survey experiments to evaluate public attitudes about topics related to secrecy, such as covert action, lying, and deception in foreign affairs (Carnegie, Kertzer and Yarhi-Milo, 2023; Maxey, 2021; Myrick, 2020; Yarhi-Milo and Ribar, 2023). Most of these studies focus on public opinion about secrecy in the context of using military force.…”

Rethinking the Tradeoffs of Within-Subjects Designs in International Relations Experiments

“…2Some within-subjects designs do not randomize the order of the conditions because they include control conditions as the baseline and initial condition to which the following treatment conditions are compared to (e.g. Bush and Prather, 2021;Demarest, Jost and Schub, 2024;Naoi, Shi and Zhu, 2022;Spilker, Nguyen and Bernauer, 2020;Yarhi-Milo, Kertzer and Renshon, 2018;Yarhi-Milo and Ribar, 2022). been particularly popular among the two strands of research often reliant on small samples: works on individuals' strategic interactions that use lab experiments (e.g.…”

“…This is particularly true for the studies with within-subjects designs for which one of the conditions is the "baseline" or control condition (e.g. Bush and Prather, 2021;Demarest, Jost and Schub, 2024;Naoi, Shi and Zhu, 2022;Spilker, Nguyen and Bernauer, 2020;Yarhi-Milo and Ribar, 2022).9 Because the controlled conditions establish the baseline, these studies assign subjects to the controlled conditions first and then other treatment conditions afterwards. In particular, priming studies are subject to this constraint because it is di cult to "undo" priming; once a researcher introduces subjects to a stimulus, asking them to not think about it afterwards is very di cult.10 This implies that priming studies' estimates of treatment e ects may include the design-induced order e ects from having the controlled condition first and may di er from estimates from between-subjects designs.…”