American Indian Constitutions: An Automated Content Analysis

• In Event: The Politics of Customary Institutions: Comparative Data and Analysis

Abstract

We develop new data on the constitutions and political institutions relevant to policy outcomes across American Indian and Alaska Native (AIAN) nations. There is an extensive literature on how institutional characteristics can affect policy outputs, such as violent crime rates, and many researchers have tried to apply similar approaches to the study of AIAN nations. However, existing data on AIAN institutions have largely been based on either expert coding and surveys. These approaches are costly and time consuming to carry out. They require pre-specified codebooks or questionnaires. Finally, the end product will become obsolete if institutions change. We present a new and original digital database of the full text of constitutions for American Indian polities. This allows for content analysis of the documents and helps facilitate more flexible and customizable measures of complexity and the quality of institutions. We implement a text-as-data approach using machine learning techniques to analyze the effect of relevant AIAN institutional design on criminal violence. We compare our machine coded measures against existing alternatives including human coded constitution data by Cornell and Kalt (2000). We show that machine coding replicates and validates expert coded data; applying these to the larger corpus allows us to provide more extensive data on institutions and governance outcomes in AIAN communities.

Authors

• Rebecca Elizabeth Cordell, University of Texas at Dallas

• Kristian Skrede Gleditsch, University of Essex

• Florian G. Kern, University of Essex