Page 14 - The lraternational Journal of the Royal Society of Thailand.indd

The International Journal of the Royal Society of Thailand
Volume XV-2023

standard established by the United Nations Statistics Division (available at: https://
unstats.un.org/unsd/methodology/m49/). Furthermore, as this article exclusively
focuses on researchers with Thai affiliations, we excluded Thailand from the South-

eastern subregions when calculating the proportions.

Finally, the characteristics of co-authorship networks were described using
the following variables: 1) Number of co-authors (network size); 2) Edge density,
calculated as the ratio of the number of edges to the total number of all possible edges;

3) Number of communities, classified using the greedy optimization of modularity
algorithm; and 4) Clustering coefficient (transitivity), representing the probability that

the adjacent researchers of a researcher have co-authored an article.

Statistical analyses

Monotonic relationships among the characteristics of researchers, their
co-authors, and networks was assessed by computing pairwise Spearman’s correlation

coefficient . Subsequently, the principal component analysis (PCA) was performed on
the centered and normalized values of each variable to reduce the dimensions of

highly correlated characteristics among all researchers and identify variables that
explain most of the variations. In the PCA, we included only the proportions of
geographical subregions for most co-authors of Index Researchers IR_01 and IR_02,

which encompassed Thailand, Non-Thailand South-eastern Asia, Southern Asia,
and Western Asia. Additionally, given the suggestion that paper mill services are

primarily based in Eastern Europe (Abalkina A, 2021), we also incorporated the
proportion of co-authors from Eastern Europe into the analysis. The PCA was conducted
using the FactoMineR (Lê et al, 2008) and factoextra (Kassambara & Mundt, 2020)

packages.

As the final part of this case study, we further examined the detailed characteristics
of the co-author networks of two groups of researcher, each comprising three researchers:
1) High-performance junior researchers (HJRs), defined as researchers with the lowest

Euclidean distance from the coordinates of Index Researcher IR_01 on the first two
dimensions of the individual PCA plane; 2) Senior researchers (SRs), defined as

researchers who showed the most alignment on the same dimension and direction
with the variable “time since the first publication”.

A Potential Use of Ego-centric Co-authorship Network Analysis for
6 Inspecting Research Misconduct: A Case Study of Thai Researchers