Transdisciplinary Transfer: Adapting Psychological "Inference Loops" to Political Concept Validation in Text Analysis

Political scientists often struggle with the validity of computationally derived constructs—what does "populism" or "polarization" really mean in a dataset? Goddard & Gillespie (2025) introduce "inference loops" in psychology: iterative cycles of manual coding, model development, and concept refinement. This project adapts their RAMP methodology to political science, creating a workflow where computational models and human coders iteratively critique and refine operationalizations of key political concepts. This would go far beyond static dictionary or supervised approaches (Yanovets & Smal, 2020), making computational political text analysis not just more accurate, but better aligned with evolving theoretical debates. The impact? More robust, generalizable, and interpretable findings in computational political science.

References:

1. POLITICAL DISCOURSE CONTENT ANALYSIS: A CRITICAL OVERVIEW OF A COMPUTERIZED TEXT ANALYSIS PROGRAM LINGUISTIC INQUIRY AND WORD COUNT (LIWC). A. Yanovets, O. Smal (2020). Naukovì zapiski Nacìonalʹnogo unìversitetu «Ostrozʹka akademìâ». Serìâ «Fìlologìâ».
2. The repeated adjustment of measurement protocols method for developing high-validity text classifiers.. Alex Goddard, A. Gillespie (2025). Psychological methods.