An empirical review of the different variants of the Probabilistic Affinity Index as applied to scientific collaboration

TL;DR

This paper reviews various versions of the Probabilistic Affinity Index (PAI), analyzing their conceptual and empirical differences, and demonstrates their application in identifying scientific collaboration partners across countries.

Contribution

It provides a comprehensive review of PAI variants, compares their empirical results, and offers guidelines for selecting appropriate variants based on research context.

Findings

PAI variants show minimal differences in results across countries.

The choice of PAI variant depends on specific research questions.

The paper offers a streamlined procedure for applying PAI.

Abstract

Responsible indicators are crucial for research assessment and monitoring. Transparency and accuracy of indicators are required to make research assessment fair and ensure reproducibility. However, sometimes it is difficult to conduct or replicate studies based on indicators due to the lack of transparency in conceptualization and operationalization. In this paper, we review the different variants of the Probabilistic Affinity Index (PAI), considering both the conceptual and empirical underpinnings. We begin with a review of the historical development of the indicator and the different alternatives proposed. To demonstrate the utility of the indicator, we demonstrate the application of PAI to identifying preferred partners in scientific collaboration. A streamlined procedure is provided, to demonstrate the variations and appropriate calculations. We then compare the results of implementation for five specific countries involved in international scientific collaboration. Despite the different proposals on its calculation, we do not observe large differences between the PAI variants, particularly with respect to country size. As with any indicator, the selection of a particular variant is dependent on the research question. To facilitate appropriate use, we provide recommendations for the use of the indicator given specific contexts.