Quantifying scholars’ marginal contributions to average-based journal metrics

Abstract

Attributing journal metrics to individual scholars is essential for modeling research influence, yet current methods rely on a restrictive assumption of monotonicity. Traditional credit allocation frameworks, such as Full and Fractional counting, assume that adding a research article to a portfolio never decreases a scholar’s attributed standing. While valid for cumulative metrics, this logic is mathematically inconsistent with average-based metrics like the Journal Impact Factor (JIF). Such indicators are inherently non-monotonic functions where performance below the established mean exerts a reductive impact on the aggregate score. This study challenges the monotonic baseline by introducing a marginal attribution framework that captures how individual marginal contributions can either increase or decrease an average-based journal metric. We propose three novel methods: Marginal Full Credit , Marginal Fractional Credit , and Marginal Shapley . The latter adapts the game-theoretic Shapley value using an average-based characteristic function to calculate marginal contributions that may be negative. To ensure scalability for large datasets where exact computation is prohibitively complex, we employ a statistically rigorous and commonly used bounded Monte Carlo approximation. Empirically, we test these methods on a dataset of articles from twelve leading Information Science journals. Our findings demonstrate that, because citation distributions are highly right-skewed, over 50% of scholars are associated with negative signed marginal effects on the journal-period average. These values should not be interpreted as penalties or judgments of scholarly quality; rather, they are a structural inevitability of average-based indicators and reflect the mechanical behavior when publication-level citation rates fall below the relevant journal mean. By relaxing the monotonicity constraint, this study aligns attribution models with the mathematical reality of scholarly metrics, providing a nuanced understanding of how individual output shapes collective performance indicators.