nested-data

This paper evaluates confidence interval methods for classifier performance metrics under conditions typical of social science text classification, providing guidance on accurate interval estimation for metrics like recall and precision when using large language models.