Cognitive resilience and sustained engagement: Response time dynamics in PISA 2022 mathematics

OBJECTIVE: To investigate whether cognitive resilience is associated with a buffering pattern against effort withdrawal as subjective task difficulty rises, using response time (RT) dynamics from PISA 2022 Mathematics process data. METHODS: Drawing on 51,190 item responses from the South Korean sample, cross-classified multilevel models with plausible-value pooling were fitted to examine the interaction between cognitive resilience (CR) and relative difficulty (RD) on log-transformed response time. Within-between decomposition of RD and multiple sensitivity analyses were conducted to disentangle within-person situational regulation from between-person differences and to assess robustness across data-cleaning specifications. RESULTS: Higher relative difficulty was consistently associated with shorter time-on-task (β ≈ -0.19, p < .001), reflecting a general tendency toward effort withdrawal as task demands increase. Students with greater cognitive resilience showed a significantly smaller reduction in time-on-task (CR × RD: β ≈ 0.02, p < .001). This moderating effect operated primarily at the within-person level and remained significant when analyses were restricted to correct responses only, consistent with productive persistence rather than unproductive wheel-spinning. CONCLUSION: Cognitive resilience is associated with a buffering pattern against effort withdrawal under challenging conditions in a large-scale mathematics assessment. The findings offer a process-oriented framework for understanding academic persistence beyond final scores and highlight the importance of fostering self-regulatory capacities in demanding learning environments. However, causal interpretations are precluded by the observational design, and the self-reported resilience measure captures perceived regulatory capacity rather than direct assessment of executive or metacognitive processes.