The objective of the selection process was to enable the identification of early to mid career researchers from Australia across all fields of research in science, social sciences and the humanities.
As a first step, we selected all the papers indexed in the Web of Science core collection between 2000 and 2015 where at least one address included Australia or New Zealand. For each publication in this set, we identified the different authors using their unique researcher identifiers (ResearchID and ORCID) as well as an algorithm based name disambiguation method that attributed an internal author identifier. Out of a total of over 977,500 papers were retrieved, a total of 1,359,050 authors were identified in the set.
For each of the identified authors, we only selected those who had published at least one paper in Australia between 2009 and 2014 and who had published at least one paper indexed in the Web of Science core collection in the last 10 years (2005-2014) but did not publish any before 2004. This subset of authors defined the total number of Australian early to mid career researchers. We used the date of first publication as an indicator of a research career start so we did not use the absolute age of the researchers. A final set of 343,550 author identifiers were selected to cover candidate Australian early-mid career researchers.
We retrieved all the papers published by the candidate Australian early-mid career researchers but we removed from the set any identifier with five or less papers and any paper with more than 100 co-authors. For each of the remaining identifiers, we calculated the number of papers published, the number of papers classified in the top 10% for their citations (of their Web of Science category, publication year and document type) and the percentage of such papers to the number of papers they have published.
Each author identifier was then attributed to a broad domain of research based on the Web of Science categories of the papers they co-authored. These domains were built to represent research output of a similar size in Australia. An author would be included in one of these 10 broad domains if 40% or more of their papers are classified in one. Remaining authors are classified in a “multidisciplinary” domain.
We selected the author who had the best overall performance as well as in each of the 11 domains. When performance could not be clearly differentiated between two authors, including for authors in more than one domain, both researchers were selected. Performance of authors was measured by the highest percentage of papers in the top 10% when the number of papers was higher than the median of their domain. When the percentage of top 10% papers was close (within 10%), we manually selected authors using criteria such as the highest number of Highly Cited papers as defined in the Essential Science Indicators (ESI) in July 2016, or the highest number of papers in which the author had an important role (i.e. first or reprint author).
The gender and the career stage of the selected authors were then checked manually and only researchers who met the criteria were selected for the award.
Authors already selected in previous citation awards in Australia were excluded from this selection. If after manual checks, we identified authors who had published before 2000, papers not indexed in the Web of Science or whose papers were not attributed their identifiers, we have maintained them in the selection only if they would meet the other selection criteria.