Implausible Denials: How Leaders Use Revealed Covert Actions to Achieve Foreign Policy Goals
Restricted (Penn State Only)
- Author:
- Bloch, Chase
- Graduate Program:
- Political Science
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- June 07, 2023
- Committee Members:
- Lee Banaszak, Program Head/Chair
Glenn Palmer, Major Field Member
Roseanne McManus, Co-Chair & Dissertation Advisor
Shomir Wilson, Outside Unit & Field Member
Burt Monroe, Co-Chair of Committee - Keywords:
- International Relations
Conflict Studies
Covert Actions
Text Classification - Abstract:
- Recent literature on covert actions has identified many reasons that they are useful even if they are visible and attributable. The most important of these reasons is that covert actions can reduce escalation risks. In my dissertation, I examine how an official denial or non-acknowledgement of a covert action can be more important than the secrecy of the covert action itself. In the first chapter, we use a survey experiment to compare three causal pathways via which denials of revealed covert actions can lower preferences for escalation: reputation, emotional response, and uncertainty of the perpetrator. We find that all three mechanisms explain how denials moderate escalation preferences, with uncertainty being the strongest mechanism and emotional response being the weakest. In the second chapter, I address the question of why leaders ever acknowledge covert actions if denials are so effective and acknowledgements risk escalation. Using newly collected data on responses to covert actions, I find that media attention, presidential popularity, a unified congress, the morality of the action, and the risk for escalation all contribute to a leader’s decision to acknowledge or deny covert actions. Finally, I use a novel machine-human hybrid classification method to build a dataset on covert actions from news articles. With the new coding method, I classified 38,000 news articles with 95% accuracy after hand labelling only 2,500 training cases. The dataset itself demonstrates the potential to collect data on covert actions much sooner than the 25 years allowed by the FOIA act.