Comparing Architectures for Supervised Political Scaling

2026-07-01

AI Summary

Existing approaches to scaling political texts are constrained by conventional classification or regression paradigms, leading to performance bottlenecks. This work proposes a unified multi-scale joint prediction framework and systematically evaluates supervised learning strategies spanning pure classification, pure regression, and intermediate hybrid modeling paradigms. Experimental results demonstrate that joint modeling significantly outperforms independent task-specific approaches, with intermediate paradigms exhibiting superior representational capacity and effectiveness for predicting positions along political ideology scales. These findings establish a cohesive and optimized pathway for the automatic scaling of political texts, advancing beyond the limitations of traditional methodological boundaries.

Abstract

Text scaling, the task of positioning political actors on an ideological scale, is a fundamental task in political analysis. To ease the need for manual analysis, various NLP methods have been proposed for this task, including classification- and regression-based approaches, showing successes as well as limitations. The goal of our paper is to consolidate the state of the art in this area. We ask two questions: (a) Can the performance of scaling methods be improved by predicting scales not individually but jointly? (b) Is there a middle ground between classification and regression?

Problem

• political scaling
• text scaling
• NLP
• classification
• regression

Innovation

• political scaling
• joint prediction
• classification-regression hybrid
• ideological scale
• NLP for political analysis