Post-1998 reform, Indonesia started a new phase in various fields, among them the policy for advancing science and technology. Putera and Jannah (2012) analyzed the national science policy and concluded that the 2000–2011 period did not sufficiently support research advancement. In response, the government enacted Law No. 11/2019 on the National Science and Technology System, aimed at accelerating research ecosystem development through the creation of the National Research and Innovation Agency (BRIN) and broader reforms in governance, funding, and cross-sector coordination ( Burhaniet al., 2021; Indonesia 2019).

Over the past decade, Scopus-indexed publications have become the primary marker of scholarly status in Indonesia, shaping promotion decisions and publication-linked incentives (Minister of Research, Technology and Higher Education, 2017). This emphasis is national in scope and formalized through Ministry regulations on lecturer evaluation and promotion (Minister of Research, Technology, and Higher Education, 2017; Ministry of Higher Education, Science, and Technology, 2025). The promotion criteria specify minimum publication thresholds. For example, eligibility for promotion to Associate Professor includes at least one Scopus-indexed article as the first author in a reputable international journal, while Full Professor promotion adds journal-metric thresholds (e.g., SJR or JIF cut-offs). In practice, these rules are often operationalized through annual output targets and incentive schemes, which have attracted criticism for amplifying publication pressure amid uneven resources and oversight.

Scopus remains the most preferred indexing database among Indonesian researchers (Nasution, 2023), prompting targeted training initiatives to improve their publication skills (Febrian et al., 2024; Nanere et al., 2025; Susanti et al., 2025; Syairofi & Irawan, 2025). Academic writing is also shaped by competition for recognition (what Merton (1973) called ‘credit’) as a form of symbolic capital that influences career progression, professional reputation, and access to resources. While Fry et al. (2023) suggest such policies can boost productivity and capacity in low- and middle-income countries, critics highlight potential downsides. For instance, Suyitno (2023) refers to high-output academics as ‘servants and worshippers of Scopus’. In Russia, Lavakov and Teixeira da Silva (2025) observed that performance metrics heightened institutional pressure and undermined research integrity, a concern relevant to Indonesia.

To address these challenges, Al-Fatih (2025) recommends reforming faculty evaluation to be more inclusive, strengthening research integrity, and providing curated lists of reputable publishers. Conversely, Irawan et al. (2023) call for a shift in focus away from metrics toward maintaining research integrity and aligning funding with national priorities.

This study also draws on recent Indonesian scholarship on academic misconduct and ethics. Susilowati et al. (2023) analyse generational and gender perspectives on misconduct, revealing nuanced attitudes toward plagiarism and compliance. Al-Fatih (2025) warns of declining publication standards under current incentive systems. Agusta et al. (2025) review ethical principles and common forms of misconduct, offering a conceptual foundation for understanding integrity challenges. Together, these studies reinforce the need to examine retraction patterns from both gender and policy perspectives.

Prominent retraction cases in Indonesia have drawn public attention to research ethics and editorial responsibilities. One high-profile example was the withdrawal of a manuscript on the Gunung Padang archaeological site in Cianjur, which attracted intense discussion on social media because of its controversial claims about prehistoric chronology ( Retraction Watch, 2024a; Tempo, 2024a). In another case, an economics professor at a private university was found guilty of academic misconduct while attempting to boost Scopus-indexed output, resulting in dismissal from the deanship, revocation of the professorship, and multiple retractions (Retraction Watch, 2024b; Tempo, 2024b). These cases highlight both the growing awareness of academic malfeasance and the fragile oversight mechanisms governing Indonesian research.

Retraction serves as a corrective measure, but it often reflects deeper systemic deficiencies, including institutional pressure, ethical breaches, and inadequate editorial oversight. The Committee on Publication Ethics (COPE) stresses that retractions are essential for preserving the scholarly record when research is deemed to be unreliable (COPE Council, 2019). Prior research has examined various aspects of retraction, including the mitigating effect of international collaboration (Zhang et al., 2019), the higher occurrence of retractions in gold open-access journals (Zheng & Fu, 2024), and disciplinary differences (Grieneisen & Zhang, 2012). Fanelli et al. (2015) further argue that institutional culture, financial incentives, and national regulatory frameworks must be considered in understanding retraction patterns. However, as Shorlec (2024) notes, this process can involve disputes between authors and editors without independent review, underscoring its procedural limitations.

Studies on gender and retraction have shown mixed results. Fanelli et al. (2015) observed that men are more frequently linked to retracted papers in absolute numbers, largely due to their over-representation in senior academic roles, yet the reputational damage from retraction may disproportionately affect women. Zheng et al. (2025) found that male first authors were more often implicated in misconduct-related retractions, such as plagiarism and authorship conflicts. Decullier and Maisonneuve (2021) reported that 59.2% of retractions involved male authors, compared to 28.6% involving female authors, with distinct patterns of error between the genders. Li et al. (2023) showed that cultural and policy environments shape gender disparities in retraction outcomes. Similarly, Pinho-Gomes et al. (2023) reported that 60.9% of retracted articles had male first and last authors, while Sebo et al. (2023) found women underrepresented among retracted medical publications. Ribeiro et al. (2023) highlighted that women accounted for only about 25% of self-retractions, suggesting fewer voluntary withdrawals for honest error among female scientists.

Large-scale analyses have also revealed recurring structural patterns. Maddi et al. (2025) noted that retracted works often appear in high-impact or open-access journals, are produced by medium-sized teams (three to ten authors), and are concentrated in health and life sciences. In contrast, multinational collaborations and project-based funding show lower retraction rates. Across these contexts, gendered differences emerge: male-led teams are more often linked to ethical breaches, whereas female-led teams are more frequently associated with procedural or administrative issues. This asymmetry reflects the broader challenges of recognizing women’s contributions and equitably distributing accountability.

These disparities align with the Matilda Effect (Rossiter, 1993), which describes the systemic undervaluation of women’s scientific work. In collaborative projects, women may be excluded from leadership roles or disproportionately blamed during misconduct investigations, especially in contexts where gender equity policies are weak. Therefore, retraction-related stigma can have long-term career impacts on female researchers, who remain underrepresented in many disciplines. In Indonesia, women constitute 45.45% of lecturers nationally (Directorate General of Higher Education, Research, and Technology, 2023), but only 32.2% at Institut Teknologi Bandung (ITB), a leading STEM institution, illustrating persistent gender imbalances in high-prestige fields.

While research on retraction and gender is growing, most evidence comes from the Global North. Little is known about how gendered authorship structures interact with retraction mechanisms in the Global South , where policies, cultures, and institutional incentives differ markedly. In Indonesia, scholarly discussions on these issues remain limited, often confined to news coverage or informal commentary. This study addresses this gap through the first large-scale, data-driven analysis of gender and retraction by Indonesian academic authors. By examining the gender composition of authors across disciplines, article types, publishers, and stated retraction reasons, we highlight the intersectional factors shaping both the incidence of retraction and its reputational consequences in a performance-driven academic environment.

The analysis revealed that the most frequent retraction cause fell under Label 1: Investigation and computer-generated content, which includes journal-led, institutional-led, or third party-led investigations (n = 96), use of paper mills, and manipulation of AI-generated content. This category’s dominance signals both the persistent challenge of systemic misconduct and the emerging threat of automated content misuse in scholarly publishing.

Two other categories also featured prominently: Label 2: Notice – limited or no information (n = 90), and Label 3: Problems with results, referencing, and attribution (n = 76). Although these reflect a broad range of editorial and authorial issues, their frequency suggests that ambiguous retraction notices and technical or citation errors are common triggers for withdrawal.

For consistency with prior studies, the numerical codes in Table 1 (1–10) correspond directly to the Retraction Watch Database taxonomy rather than ranking them by frequency in our dataset. This alignment facilitates cross-study comparisons and avoids confusion when interpreting the results. Notably, Label 9: Integrity and ethical compliance problems was among the least reported causes (n = 10). While infractions in this category are serious, their low frequency may indicate underreporting or limited institutional scrutiny in cases involving Indonesian researchers.

Overall, the variation across categories illustrates that retractions stem from a complex interplay of deliberate misconduct, procedural lapses, and inadequate editorial transparency. These findings reinforce the need for robust institutional ethics training, clear editorial guidelines, and regular oversight to reduce both opaque and preventable retractions in the future.

Table 2 summarizes the distribution of author gender composition, subject area, article type, and publisher of the 196 retracted publications analysed. The most common authorship pattern was mixed-gender teams led by men (39.8%), followed by men-only (22.4%) and mixed-gender teams led by women (22.4%). Women-only teams were uncommon (5.6%), and sole women accounted for only 2.6% of the cases. This underrepresentation mirrors broader gender imbalances in Indonesian academic publishing and suggests that women-led outputs, while fewer, may be subject to distinct retraction trends.

Retracted publications were most frequently found in Business (17.3%), other less represented or interdisciplinary areas (16.3%), and Computer science (13.8%). Disciplines such as Government and Chemistry had the fewest retractions (≤3.1%). These differences likely reflect both variations in publication volume and differences in editorial review processes across disciplines.

Most retracted items were research articles (52.0%), followed by conference papers or abstracts (36.7%), with review articles and meta-analyses comprising less than 12%. This predominance of primary research is consistent with global retraction trends, in which methodological flaws, data issues, or ethical breaches in original studies often trigger withdrawals.

Publisher data show that over half the retractions were concentrated in either IOP Publishing (27.6%) or the Other publisher category (31.1%), which includes multiple smaller or less internationally prominent outlets. Major publishers such as Elsevier (12.2%) and Springer (8.2%) also appeared, underscoring that retraction risk spans both large, established publishers and smaller entities, albeit potentially through different mechanisms (e.g., procedural oversight versus volume-driven editorial pressure).

To better understand these patterns, we compared retraction proportions with the estimated overall publication output for each gender composition using Scopus metadata for Indonesian-affiliated works during the same period. This normalization allowed for the assessment of whether certain gender groups were over- or under-represented relative to their publishing volume. Where precise base-rate data could not be obtained, we refrained from making causal claims about susceptibility, in line with the reviewer’s feedback to avoid unsupported interpretations.

To address research question 1, we examined whether the disciplinary subject of retracted publications was associated with the gender composition of the author teams. A Pearson chi-square test using six gender categories and eleven subject areas revealed a statistically significant association (χ² (50) = 72.911, p = 0.019). This indicates that retraction patterns differ across disciplines depending on authorship composition.

Mixed-gender teams led by men accounted for the largest share of retractions (n = 78), with high representation in Biochemistry/Biology (n = 11) and Medicine (n = 10). Men-only teams were particularly common in Computer science (n = 10) and Engineering (n = 5). Mixed-gender teams led by women (n = 44) had a more even disciplinary spread, spanning Technology, Engineering, and Education.

Women-led authorship, whether by women-only teams (n = 11) or sole woman authors (n = 5), was rare and appeared mainly in Education and Other subjects, with minimal presence in the STEM fields. This underrepresentation may reflect gender disparities in research leadership, field participation, or visibility in retraction cases.

While these trends point to possible interactions between discipline-specific publishing norms and gendered authorship patterns, caution is warranted in their interpretation. The lower proportion of women-led teams in STEM-related retractions may signal structural inequities or differences in retraction exposure, rather than differences in research integrity itself.

The Linear-by-Linear Association test was non-significant (p = 0.639), which is expected given the nominal nature of the variables and does not undermine the overall chi-square result. Taken together, these findings underscore the importance of viewing retractions not only as indicators of error or misconduct but also as reflections of broader structural and demographic dimensions in Indonesia’s rapidly evolving research environment.

To address research question 2, we assessed whether the type of retracted article was associated with the gender composition of the author teams. A Pearson chi-square test revealed a significant association, χ²(15) = 30.222, p = 0.011, indicating that retraction patterns varied by authorship composition and publication format.

Retracted research articles were most frequently produced by mixed-gender teams led by men (n = 41) and men-only teams (n = 24), followed by mixed-gender teams led by women (n = 22). Women-led groups (whether women-only or sole woman authors) were almost absent from this category, accounting for only three articles.

Conference abstracts and papers showed a similar trend, but with an unexpectedly higher share of retractions among women-only teams (n = 10) than among mixed-gender teams. This may suggest disciplinary preferences for conference-based dissemination among women or particular vulnerabilities associated with this publication type. Meta-analyses, although rare in the dataset (n = 8), were predominantly authored by male-led mixed-gender teams (n = 6), suggesting a concentration of leadership in this category.

The Linear-by-Linear Association was significant (p = 0.005), indicating a gradient in which certain gender compositions are more frequently linked to publication types that carry greater methodological complexity or perceived prestige.

These patterns suggest that gendered authorship dynamics influence not only the disciplinary distribution of retractions (as shown in RQ1) but also the types of academic outputs that are most at risk. Male-dominant or male-led mixed teams may be more engaged in producing research-intensive outputs, increasing their visibility and exposure to retraction risk. Conversely, female-dominant teams appear under-represented in high-stakes formats, potentially due to structural barriers, disciplinary norms, or editorial practices that shape publication opportunities and the likelihood of post-publication scrutiny.

To address research question 3, we examined whether the gender composition of the author teams was associated with the publisher type for retracted articles. A Pearson chi-square test revealed a significant association, χ²(40) = 67.077, p = 0.005, indicating that the patterns of retraction across publishers varied according to authorship composition.

The largest cluster of retractions came from IOP Publishing (n = 54), predominantly authored by mixed-gender teams led by men (n = 21), followed by mixed-gender teams led by women (n = 14) and men-only teams (n = 11) in that order. Although women-only and sole woman authorship appeared in IOP titles, they were rare, suggesting some gender diversity but limited representation in these retracted outputs.

Elsevier (n = 24) and Springer (n = 16) displayed different profiles: Elsevier’s retractions were almost exclusively male-dominated, whereas Springer’s showed a more balanced distribution across gender categories. Notably, women-only authorship was absent from Elsevier, Emerald, Frontiers, and Hindawi, pointing to a restricted presence, or reduced visibility, of female-led work in these journals.

The Other publishers group, comprising outlets with fewer than five retractions each (n = 61), demonstrated the widest diversity in gender composition, although male-led mixed teams and male-only teams remained the most common. Despite the significant chi-square result, the Linear-by-Linear Association was not significant (p = 0.937), suggesting that the relationship is not linear but is shaped by categorical differences between publishers.

These findings imply that publisher-specific dynamics, such as disciplinary focus, editorial policies, and perceived accessibility, may shape the gender profile of authors whose work is later retracted. Differences in retraction frequency could also reflect variations in editorial oversight and ethical enforcement, raising broader questions about how gendered authorship interacts with distinct journal ecosystems, particularly under conditions of high publication pressure and uneven ethical infrastructure.

To address research question 4, we examined whether the principal reason for retraction (i.e., the first-cited rationale in each case) was associated with the gender composition of the author teams. This approach isolates the primary justification provided by journals or publishers and avoids inflation from multiple coded reasons for each article. A Pearson chi-square test revealed a significant association (χ ² (40) = 58.724, p =0.028).

Across the 196 retracted publications, the most common first-cited reasons were Error in the article (n = 41), Notice – limited or no information (n = 40), Authorship and affiliation problems (n = 39), and Duplication and plagiarism (n = 28). Clear gender-related patterns were observed. Male-dominated teams (Men only and mixed-gender teams led by men) accounted for the majority of retractions due to authorship and affiliation issues (27 of 39, 69.2%) and article errors (24 of 41, 58.5%). In contrast, women-only or sole woman authorship teams were more frequently retracted for unclear notices (n = 6) or technical faults (n = 4) rather than for violations of integrity or authorship.

Mixed-gender (male-led) teams appeared across all nine retraction categories, with substantial representation in authorship disputes (n = 18) and peer-review problems (n = 4). Men-only teams dominated the Notice category (n = 12), possibly reflecting negotiated or ambiguous retraction processes. Female-led teams were largely absent from retractions involving duplication, integrity breaches, or investigative actions, which may reflect lower publication volumes, concentrations in specific disciplines, or differential editorial scrutiny.

These results suggest that the grounds for retraction and the way they are framed may be shaped by systemic gender dynamics. Male-led teams are more often linked to professional misconduct or clear procedural breaches, whereas female-led teams are more frequently associated with opaque or procedural reasons. This asymmetry raises important questions: Are retraction standards and transparency consistently applied across gendered authorship structures? Are certain teams more exposed to overt misconduct labelling, while others are shielded by less explicit notices? Addressing these issues will require complementary qualitative research to determine whether these patterns reflect genuine differences in misconduct incidence, disparities in publication practices or editorial bias in the construction of retraction narratives.