Research Interests: Understanding The Mechanisms Of Government Secrecy And Their Relationships To Theories In International Relations, Comparative Politics And Intelligence Studies.
Keita Omi is an Assistant Professor in the Social Sciences division of the School of Arts and Sciences. He specialises in developing the study of foreign policy, intelligence, and government leadership. Keita did his PhD in Political Science at the University of Utah in the United States. He received a Masters in Political Science from San Francisco State University, and a Bachelors in Computer Science from Reitaku University in Chiba, Japan. At the core of his project is an interest in understanding the mechanisms of government secrecy and their relationships to theories in international relations, comparative politics and intelligence studies. His research work investigates American presidential/executive politics with a comparative method and examines why and how secret decisions are made at the top of the government hierarchy by national leaders, and whether their decisions vary from one American presidency to another. A developing research agenda attempts to theorise the concept of secrecy that plays a significant role in the formation of government statecraft. Additionally, Keita is currently working toward publication of a translation work of Year of the Wild Boar: An American Woman in Japan, written by Helen Mears.
My research investigates presidential/executive politics in the United States with a comparative method. The current dissertation research that I am working on examines why secret decisions are made at the top of the government hierarchy by leaders who are relatively free from bureaucratic constraints and whether they vary from one American presidency to another. A developing research agenda analyzes how American covert statecraft is formed and what role secrecy or deception plays in the formation of American statecraft.
Research on American Statecraft:
In my dissertation, tentatively titled, “The Logic of Covert Statecraft: A Cross Presidential Study of American Presidential Decision Directives Documents,” I theorize the concept of secrecy that plays a significant role of government statecraft. My study relies on a qualitative content analysis of presidential documents for the identification of operational codes of American statecraft. The data sources for foreign policy decision makers include 1,357 presidential documents, especially decision directive documents (272 NSAMs during Kennedy administration, 100 NSAMs Johnson administration, 264 NSDMs during Nixon administration, 84 NSDMs during Ford administration, 63 PDs during Carter administration, 325 NSDDs during Reagan administration, 79 NSDs during George H.W. Bush administration, 75 PDDs during Clinton administration, 66 NSPDs during GW Bush administration, 30 PPDs during Obama administration) from 1961 through 2015. I coded these documents by hand and currently developing and testing my arguments and hypotheses about presidential unilateral power, collective action issues, coordination issues, public opinion, and government elite behaviors. I plan on expanding this project by leveraging cutting-edge methods in computer science, coding text-as-data with supervised and unsupervised machine learning.
Does a manual coding by human beings differ from textual analysis with supervised machine learning techniques? If so, to what extent and how? These are the questions that occurred to me when I completed the manual coding of over 1,000 presidential documents for my dissertation project with a decent coding manual and by trusting all my six senses (taste, sight, touch, smell, hearing, and the sixth sense). I had a concern that a sole reliance on a manual coding approach may increase the risks of building theories based on my poor understanding and unfamiliarity of individual executive cases; that is coding one document wrong could lead to a completely different picture of the puzzle I am trying to elucidate. In the field of computer science, there are many research conducted about Natural Language Processing (NLP). This research scholarship is developing text analysis methods with both unsupervised and supervised machine learning techniques. Research findings from the subfield of computer science would provide an better method for answering my future research questions above. I hope my future research will bring the two fields, political science and computer science, closer, generating more interdisciplinary studies.
Work in Progress:
Refereed Journal Publications: