Causal Evaluation of Planning Strategies in Large Language Models Through Interpretable Quality Prediction and Counterfactual Reinforcement Learning. JAAIR 2026, 5 (1).