Paper (in progress)
Gannon, J Andrés. (2025) “A Text-as-Data Approach to Measuring Interstate Relations Using National Security Strategies”. Working paper.
Abstract: This study develops and validates a novel text-as-data method for measuring interstate relations from political texts. Building on advances in natural language processing, I construct a directed dyad-year index of a state’s expressed salience (prominence) and affinity (evaluative stance) toward other states in its national security strategy. The measure is derived through a pipeline that integrates sentence-level segmentation, entity recognition, sentiment analysis, stance detection, and best–worst scaling calibration. Applied to a new corpus of 438 national security strategy documents issued by 92 countries (1962–2024), the approach produces reliable, continuous measures of interstate posture that closely track but also meaningfully diverge from traditional benchmarks such as alliance portfolios and UN voting patterns. This methodological contribution provides scholars with a replicable tool for extracting conveyed relationships from a wide range of political texts, enabling new insights into the articulation of political relationships.
