Judicial reliance on CBCA criteria and their predictive relevance in verdict outcomes: A quantitative analysis of South Korean court rulings

Abstract

Purpose

This study aimed to examine how Criteria-Based Content Analysis (CBCA) criteria are referenced in South Korean court decisions involving sexual abuse cases with victims under the age of 13 or victims with cognitive disabilities and whether specific criteria predict verdict outcomes.

Methods

A total of 264 rulings (273 coded cases) issued between 2012 and 2024 were analysed. 19 CBCA criteria and verdict outcomes (guilty vs. acquittal) were coded. Associations were tested using chi-square or Fisher’s exact tests, and predictive validity was examined using binary logistic regression and random forest models.

Results

Quantity of details (CBCA03) and subjective mental state (CBCA12) were the most consistent predictors across analytical methods, remaining significant in logistic regression and ranking highest importance in the random forest model. Logical consistency (CBCA01), although one of the most frequently cited criteria in judicial reasoning, did not predict verdicts in logistic regression but contributed substantially to classification accuracy in the random forest model, therefore suggesting potential nonlinear or interactive effects.

Conclusions

These findings highlight that courts rely most heavily on the quantity of details and subjective mental state when evaluating credibility and that these criteria, which are most predictive of verdicts, align with those demonstrating strong empirical validity in prior research. Logical consistency, although frequently referenced, functioned as a baseline cue—showing limited linear discriminative value while still contributing meaningful predictive information in nonlinear models. Future research should examine CBCA’s application across vulnerable populations and refine interview practices to elicit detailed, uncontaminated testimony.