Welo Data Unveils New Framework for Evaluating Causal Reasoning in Large Language Models