The Global Landscape of Research Misconduct in Computer Science: Mapping Research Integrity Challenges Through Network Analysis of Structural Holes and Centrality in Scopus Retracted Articles

Purpose: Retraction refers to the withdrawal of a published article due to human error or intentional scientific misconduct. Retracted articles reduce the credibility and trust of scientific communities and have harmful consequences in society. In other words, intentional and unintentional errors made by some individuals in research can damage the scientific reputation of an institution, country, and even a subject area. By analyzing the current status of research misconduct, a systematic review of retracted articles provides the opportunity for better planning and policymaking to reduce the number of these articles at the individual, organizational, national, and international levels. This scientometric research aims to study the co-authorship network of countries in retracted articles of computer science based on network centrality measures and structural holes theory.
Methodology: The present applied research adopts a scientometric approach. The statistical population of the research comprises all 8489 retracted articles of computer science indexed in the Scopus database. In addition, VOSviewer software was used to construct and analyze the co-authorship network, Pajek was used to extract aggregate constraint values and also to draw the structural holes in the co-authorship network of countries, SPSS was used for descriptive and inferential statistics, and Excel was used to generate graphs and co-authorship patterns.
Findings: Based on findings, the first retracted articles in computer sciences were published in 2003; the peak year for retraction was 2010, accounting for 33.7% cases. The retracted articles had received zero to 441 citations, and of the 8489 articles examined, 4262 (50.2%) had not received any citations. About 17.4% of the articles were single-authored. The majority of the articles (82.6%) had more than one author; the most common pattern was two-authored articles (36%) and three-authored articles (23%). In addition, the Collaborative Coefficient (CC) among the authors of the retracted articles of computer science was 0.52, indicating moderate co-authorship tendencies. The top three countries in retracted articles of computer sciences are as follows: the highest number appeared in China, India and the United States, respectively; based on the degree and closeness centrality indices, India, China, and the United States, are the top countries respectively, and taking the betweenness centrality index in account, China, India, and the United States, respectively, played a central role in the co-authorship network of studied retracted articles; considering the aggregate constraints values, China, India, and Malaysia played the largest role in covering structural gaps and country connections, respectively. Regarding Spearman correlation test, the relationship between degree and closeness centralities (r=0.949), degree and betweenness centralities (r=0.889), and betweenness and closeness centralities (r=0.889) for 87 countries (connected network) with a significance level of 0.01 was positive and significant. On the other hand, the relationship between degree centrality and aggregate constraints (r=-0.875), betweenness centrality and aggregate constraints (r=-0.766), and closeness centrality and aggregate constraints (r=-0.782) for 87 countries with a significance level of 0.01 was inverse and significant.
Conclusion: The prevalence of retracted articles from different countries indicates the challenges in research ethics in these countries, although different in scale. Network centrality indices and the structural holes theory reveal that China and India played central and broker roles in the co-authorship network of retracted articles in computer science. China and India are new scientific hubs, but the high prevalence of retracted articles in computer science could be a warning sign about the quality and ethics of research in these countries. The increase in the number of retractions in the United States, as a reputable scientific hub, will have a wider impact on the credibility of science in the world. The high number of retracted articles in countries with considerable scientific production indicates the need for greater attention to research ethics and scientific standards in these countries. Due to their central role in scientific and co-authorship networks, these countries have a greater responsibility for maintaining research credibility and trust through sustaining scientific standards and promoting research ethics. Also, it seems that European countries also need to review their research policies due to their broker roles in retracted articles network; in addition, the participation of a greater number of developing countries (Saudi Arabia, Iran, Iraq, Pakistan, Malaysia and …) in the authorship of retracted articles could be due to weak monitoring systems, lack of information about research ethics standards, and pressure to publish more articles from institutions. Taking into account the damage that fraudulent articles cause to the body of knowledge, all countries should help to maintain and expand scientific standards. They are expected to strengthen monitoring systems, train, and raise awareness in the field of research ethics. Strengthening international cooperation in exchanging experiences and improving research ethics standards would also be beneficial.