📝 SummarXiv

Abstract

Many constructs that characterize language, like its complexity or emotionality, have a naturally continuous semantic structure; a public speech is not just "simple" or "complex," but exists on a continuum between extremes. Although large language models (LLMs) are an attractive tool for measuring scalar constructs, their idiosyncratic treatment of numerical outputs raises questions of how to best apply them. We address these questions with a comprehensive evaluation of LLM-based approaches to scalar construct measurement in social science. Using multiple datasets sourced from the political science literature, we evaluate four approaches: unweighted direct pointwise scoring, aggregation of pairwise comparisons, token-probability-weighted pointwise scoring, and finetuning. Our study yields actionable findings for applied researchers. First, LLMs prompted to generate pointwise scores directly from texts produce discontinuous distributions with bunching at arbitrary numbers. The quality of the measurements improves with pairwise comparisons made by LLMs, but it improves even more by taking pointwise scores and weighting them by token probability. Finally, finetuning smaller models with as few as 1,000 training pairs can match or exceed the performance of prompted LLMs.