Distinct patterns in the accuracy of child testimonies: Evidence from a cluster‐based comparison with case records

Abstract Background Child victim testimonies are often used as critical evidence in forensic contexts, yet their accuracy can vary considerably depending on age and crime type. Understanding distinct patterns in testimonial accuracy is therefore essential for improving legal assessments. Objective This study aimed to identify distinct accuracy profiles in children's testimonies and examine how demographic and detail‐specific variables influence memory performance. Methods Transcripts from 50 forensic interviews conducted at a Sunflower Centre were analyzed. K‐means clustering was used to classify testimonies based on accuracy patterns. Group differences were assessed, and logistic regression was attempted to predict cluster membership. Results Two clusters emerged: Cluster 1 (older children, higher accuracy) and Cluster 2 (younger children, lower accuracy). Cluster 1 recalled a significantly greater proportion of correct details ( M = 92.6% vs. 82.8%, p < 0.001) and was older on average ( M = 14.5 vs. 9.6 years, p < 0.001). At the item level, both clusters showed high accuracy for identity‐related details (e.g., perpetrator gender and category). However, significant group differences were found for relationship to perpetrator, incident date and time, and perpetrator age, all more accurately recalled by Cluster 1. Cluster 2 demonstrated lower accuracy for contextual and temporally sensitive details, though they retained some concrete items (e.g., occurrences, school age) comparably. Logistic regression did not identify significant predictors of cluster membership, largely due to sample size and distributional limitations. Conclusions Findings highlight heterogeneity in children's testimonies and support the utility of cluster‐based approaches in forensic evaluations. Accuracy‐based profiles provide a more nuanced understanding of testimonial variability, beyond single demographic variables or abuse type, and emphasize the importance of content‐specific assessment of memory performance in child victims.