Operant behavioral economic methods are increasingly used in basic research on the efficacy of reinforcers as well as in large-scale applied research (e.g., evaluation of empirical public policy). Various methods and strategies have been put forward to assist discounting researchers in conducting large-scale research and detecting irregular response patterns. Although rule-based approaches are based on well-established behavioral patterns, these methods for screening discounting data make assumptions about decision-making patterns that may not hold in all cases and across different types of choices. Without methods well-suited to the observed data, valid data could be omitted or invalid data could be included in study analyses, which subsequently affects study power, the precision of estimates, and the generality of effects. This review and demonstration explore existing approaches for characterizing discounting and presents a novel, data-driven approach based on Latent Class Analysis. This approach (Latent Class Mixed Modeling) characterizes longitudinal patterns of choice into classes, the goal of which is to classify groups of responders that differ characteristically from the overall sample of discounters. In the absence of responders whose behavior is characteristically distinct from the greater sample, modern approaches such as mixed-effects models are robust to less-systematic data series. This approach is discussed, demonstrated with a publicly available dataset, and reviewed as a potential supplement to existing methods for inspecting and screening discounting data.