<p>Retraction is a key mechanism for correcting the scientific record, yet its effectiveness in curbing scholarly influence remains contested. Existing methods for estimating the retraction effect often rely on restrictive assumptions, large sample sizes, and complex modeling techniques that limit their robustness, interpretability, and case-level applicability. This study proposes a novel metric, the Retraction Impact Index (RII), to quantify the retraction effect with minimal assumptions and high flexibility. RII is defined by comparing the slope of a paper's citation trajectory before and after retraction, requiring only a linear trend assumption, and is normalized to allow intuitive interpretation across cases. Using a large-scale dataset of 20,443 retracted papers collected from Web of Science, we validated RII's reliability, internal consistency, and face validity through case analysis and distributional assessments. Further, we conducted matched-sample comparisons and subgroup meta-analyses across 19 journals to evaluate RII's effectiveness in distinguishing retraction impacts at the journal level. Results show significant variation in retraction effects, and access models have minimal influence on it. RII proves to be a scalable, interpretable, and statistically robust tool for retraction studies, suitable for both single-paper evaluations and large-sample analyses. This new metric offers practical value for researchers, journal editors, and policymakers seeking to monitor, evaluate, and improve post-retraction scholarly communication.</p>