Relative merits and optimal granularity of field normalisation components

Determine the comparative effectiveness of the component choices used in field-and-year normalised citation indicators, including identifying the optimal granularity of subject classification schemes (e.g., OpenAlex domains, fields, subfields, topics, or Scopus ASJC fields) for constructing Normalised Citation Scores (NCS) and Normalised Log-transformed Citation Scores (NLCS) that best align with independent research quality judgements.

Background

Field normalisation aims to make citation-based comparisons fair across disciplines and years, typically via indicators like NCS and NLCS. A critical design choice in these indicators is the subject classification scheme used to define the field-level denominators. The granularity and construction of these schemes (journal-based vs article-based; broad vs narrow categories) can substantially affect indicator behavior and validity.

The paper notes that despite many descriptive studies and limited small-scale validations, systematic evidence on which normalisation components and classification granularities yield the best agreement with expert or gold-standard quality judgements is lacking. This motivates explicit investigation of optimal component configurations.