Measuring Group Performance Fairly: The h-Group, Homogeneity, and the α-Index

Abstract

Ranking research groups plays a crucial role in various contexts, such as ensuring the fair allocation of research grants, assigning projects, and evaluating journal editorial boards. In this paper, we analyze the distribution of h-indexes within research groups and propose a single metric to quantify their overall performance, termed the α-index. This index integrates two complementary aspects: the homogeneity of members’ h-indexes, captured by the Gini coefficient (g), and the h-group, an extension of the individual h-index to groups. By combining both uniformity and collective research output, the α-index provides a consistent and equitable metric for comparative evaluation, essentially calculated as the average relative h-group multiplied by (Formula presented.) and normalized by the maximum value of this quantity across all analyzed groups. We describe the full procedure for computing the index and its components and illustrate its application to computer science conferences, where program committees are compared through a resampling procedure that ensures fair comparisons across groups of different sizes. Additional results are presented for postgraduate programs, further demonstrating the method’s applicability. Correlation analyses are used to establish rankings; however, our primary goal is to recommend a fairer index that reduces deviations from those currently used by governmental agencies to evaluate conferences and graduate programs. The proposed approach offers a more nuanced assessment than simply averaging members’ h-indexes and can be applied broadly–for example, to university departments and research councils–contributing to a more equitable distribution of research funding, an issue of increasing importance.