Prompt Stability Scoring for Text Annotation with Large Language Models
This addresses the replicability issue in classification routines for applied researchers, though it is incremental as it adapts existing reliability scoring methods.
The researchers tackled the problem of reproducibility in text annotation using large language models by proposing a general framework for diagnosing prompt stability, resulting in the Prompt Stability Score (PSS) and a Python package, tested on six datasets and 3.1 million rows of data.
Researchers are increasingly using language models (LMs) for text annotation. These approaches rely only on a prompt telling the model to return a given output according to a set of instructions. The reproducibility of LM outputs may nonetheless be vulnerable to small changes in the prompt design. This calls into question the replicability of classification routines. To tackle this problem, researchers have typically tested a variety of semantically similar prompts to determine what we call ``prompt stability." These approaches remain ad-hoc and task specific. In this article, we propose a general framework for diagnosing prompt stability by adapting traditional approaches to intra- and inter-coder reliability scoring. We call the resulting metric the Prompt Stability Score (PSS) and provide a Python package \texttt{promptstability} for its estimation. Using six different datasets and twelve outcomes, we classify $\sim$3.1m rows of data and $\sim$300m input tokens to: a) diagnose when prompt stability is low; and b) demonstrate the functionality of the package. We conclude by providing best practice recommendations for applied researchers.