1.1. Overview of Publishing Models

OA is a scholarly publishing model in which research findings and information are made freely available to readers at no cost. In contrast, the subscription-based model requires readers to pay for access to research content. OA journals have different ways of publishing such as gold, hybrid, green, and bronze. In the golden route, the article publishing charge (APC) is paid by the author or their institution and the documents are available to the readers in full text. In the hybrid route, hybrid journals are subscription-based scholarly platforms where authors can pay an APC for OA to specific works, while others remain behind a paywall, requiring readers to pay for full-text access. In green access, authors self-archive their communication and it will be published publically after the embargo period. Publications in bronze OA are freely accessible on publishers’ platforms, but do not include a Creative Commons Attribution License. As a result, they may be subject to modifications at any time, potentially limiting their accessibility and reusability (Cho, 2021b).

This study explores the following questions:

1.2. Research Questions

RQ1) What are the altmetric scores of highly cited articles on pesticide exposure?

RQ2) Do the highest altmetric attention scores correlate with the citation numbers of pesticide exposure articles?

RQ3) Does OA status affect the altmetric and citation scores of highly cited articles on pesticide exposure?

RQ4) Which sources exhibit significant concentration in pesticide research literature and impact?

2.1. Reviews Based on Highly Cited Articles

Cho (2021b) investigated altmetrics for highly cited social science papers (638) from Scopus, evaluating correlations with citations and OA status. Altmetric scores of views, readers, blogs, Wikis, and Tweets linked with citations and OA were analyzed using Spearman correlation and Mann-Whitney U Tests. The consequences demonstrated a strong relationship between citations and readership, with OA papers showing higher altmetric engagement. The top 1,000 articles of Nature on social media websites and tools are examined by Ouchi et al. (2019) to evaluate and correlate the altmetric score and their bibliometric indicators. In that, 98.9% of articles were incorporated with social media. Mendeley, Wikipedia, and CiteUlike are the most altmetric-scored social media sources, and a positive correlation between altmetric score and citation score was also found. Nisha et al. (2022) conducted a bibliometric analysis of the top 100 most cited studies in information science and library science, revealing that most of the top 100 cited research papers (80%) are in paid subscription journals, emphasizing the importance for libraries to maintain these subscriptions for student access to high-quality research. Highly cited articles were primarily from the USA, Canada, and UK, indicating the growing influence of social media in research dissemination. Saberi and Ekhtiyari (2019) examined highly cited documents in the library and information science (LIS) field as identified by Google Scholar, revealing a positive and significant correlation between capture metrics and Google Scholar citations. Similarly, Li and Thelwall (2012) conducted a comparative analysis of 1,397 genomics and genetics articles using Faculty of 1,000 and Mendeley data alongside traditional bibliometric indicators, identifying a significant correlation between citations and Journal Impact Factors.

Previous research on highly cited studies demonstrates a positive correlation between altmetric scores and traditional citations, with OA papers generally achieving greater altmetric presence and visibility.

2.2. Reviews Based on Open Access and Altmetric Score

The impact of OA on altmetrics for LIS research was determined by Cho (2021a), who found a prevalence of Mendeley bookmark readers (63%) and an unreliable comparison of altmetrics in OA papers. Particularly, views, blogs, and tweets display higher sensitivity in OA papers, while Mendeley readers shows a major correlation with citations. Mounce (2015) suggests that OA may offer an altmetric advantage due to increased visibility, noting that seven out of the top ten papers ranked by Altmetric.com in 2012 were OA. Mounce (2015) also points out that while the OA model is relatively well-established, altmetrics are still new and largely unexplored. Bernal (2013) reviews studies on OA and emerging metrics, highlighting that altmetrics provide added value for OA (Poplašen & Grgić, 2017) and that there is a correlation between page views, citations, and social media impact (Priem et al., 2012). It is obvious from the above literature that OA boosts altmetric impact by increasing visibility and engagement across platforms such as Mendeley and social media, strengthening research reach and influence.

2.3. Reviews Based on Pessimistic Correlation between Citation and Altmetric Score

Heydari et al. (2019) analyzed 1,077 highly cited surgery articles and found 62.74% of documents received altmetric scores. Positive correlations were observed between citations and Wikipedia, policy-making documents, CiteUlike, and citation count. A non-significant positive correlation was found with Mendeley readers. Overall, the weak and negative correlation between citation counts and altmetric attention scores indicates that altmetrics serve as a complementary measure rather than a replacement for traditional citations. Azer and Azer (2019) searched for top-cited articles on medical professionalism from the Web of Science database and identified that no correlation was found between citation and altmetric scores for publications before 2007, but for those published in 2007 and after, there was a significant correlation. The overall summary of the literature confirms that altmetrics generally show weak correlations with traditional citations, indicating that they complement rather than replace citations. Positive correlations were noted with platforms such as Wikipedia and policy documents, but with some variations depending on the publication year.

2.4. Reviews Based on Optimistic Correlation between Citation and Altmetric Score

Statistically significant associations were identified for metrics including Twitter, Facebook, research highlights, blogs, mainstream media, and forums. These findings underscore the influence of time in using altmetrics to rank articles and raise questions about the practical utility of altmetrics due to their inconsistent coverage (Thelwall et al., 2013). A study by Mohammadi and Thelwall (2014) examined the correlation between Mendeley readership counts and citations in social sciences and humanities disciplines, and also identified patterns of information flow between scientific fields using Mendeley data, showing potential for capturing knowledge transfer across disciplines and providing early-stage impact evidence. Thelwall et al. (2023) conducted an evaluation of the UK Research Excellence Framework 2021 and found that altmetrics demonstrated a stronger correlation with research quality and field-normalized Scopus citation counts. Among the various altmetrics, Mendeley readers counts emerged as the most reliable indicator across the majority of assessment units. It was also found that altmetrics are generally stronger indicators of research quality in health and physical sciences and weaker in the arts and humanities.

Former studies show a clear correlation between the number of citations an article receives and its free online availability (Davis & Walters, 2011; Gargouri et al., 2010; Harnad & Brody, 2004; Kurtz et al., 2005; Lawrence, 2001). Very few studies have examined the connection between OA and altmetric scores. It is crucial to avoid applying the same criteria to older and newer articles. As Web 2.0 continues to evolve, the growing number of social media users has increased altmetric scores for earlier publications compared to more recent ones—a trend attributed to the expansion of communication channels rather than differences in publication quality (Adie, 2014). Moreover, scientific publications generally receive higher altmetric scores than those in other fields. Despite existing literature, significant gaps remain in understanding the correlation between altmetrics and citations, particularly in specialized fields such as pesticide research. Most studies focus on broad disciplines like social sciences and medicine, leaving the relationship between OA and altmetric scores within pesticide research underexplored. Inconsistencies across studies regarding the relationship between altmetrics and citations further highlight the need for more focused investigation.

This study addresses these gaps by examining the interplay among altmetrics, citations, and OA in pesticide research, providing new insights into the field. Previous research has shown strong associations between altmetrics and social media activity, with timing playing a crucial role in article rankings. Mendeley readership demonstrates a significant correlation with citation counts, and altmetrics serve as stronger indicators of research quality in health sciences than in the arts. However, few studies have explored the link between OA and altmetrics in specialized areas such as pesticide research. The following sections outline the study’s materials, methods, and analytical approaches used to achieve these objectives.

3.1. Materials

To address the research questions, data were retrieved from the Web of Science database, maintained and published by Clarivate Analytics. Although Scopus and Google Scholar also offer citation tracking capabilities, it was essential to limit the search to the Web of Science due to its consistent updates and amendments. On December 20, 2023, data were collected using the keywords ‘Pesticides,’ ‘Herbicides,’ ‘Fungicides,’ ‘Rodenticides,’ ‘Insecticides,’ and ‘Altmetric Score’ in the basic search query. This initial query yielded 28,000 documents. To refine the results, the Web of Science category ‘Toxicology’ was applied, narrowing the dataset to 6,718 documents. Further filtering was performed by restricting the publication years to 2013-2023 to focus on recent developments in traditional citation scores and social media metrics within the pesticide research literature. Ultimately, 3,291 documents were selected for analysis (Fig. 1).

Fig. 1

Data structure of the study.

3.2. Methods

The citation and statistical analyses were conducted using Bibexcel software (2017 ver; Olle Persson, Leuven, Belgium), while the altmetric scores were obtained from the Journal of Altmetrics ( https://www.journalofaltmetrics.org) by utilizing the DOIs of all 3,291 documents, resulting in altmetric scores for 1,961 documents. To determine the association between citation scores, altmetric scores, and OA status, Spearman’s rank correlation analysis was performed using IBM SPSS Statistics 26 (IBM Corp., Armonk, NY, USA). Furthermore, the Mann-Whitney U Test was utilized to determine whether OA articles received a significantly higher number of citations and altmetric scores compared to CA articles (Fig. 2).

Fig. 2

Overview of collected data: Key information visualization.

4.1. Traditional Citation

A substantial portion of documents (68.03%) is published in subscription or CA journals, contrasting with 31.97% exclusively in OA journals. Within OA, prevalence is observed in green access (603 documents, 13,143 citations), followed by hybrid access (170 documents, 3,123 citations), gold access (158 documents, 2,488 citations), and bronze access (121 documents, 2,150 citations). Research focus is especially concentrated in subscription-based journals with green OA being the preferred choice for its absence of article publication charges, and the highest citation was received in the post-embargo period. The cumulative citation count across 3,291 documents stands at 58,578, with the majority of citations occurring in 2013, accounting for 11,049 citations. Publications in recent years are yet to be cited. Subscription journals led with 37,674 citations, while OA journals received 20,904 citations. Traditional citation scores indicate 2,842 cited documents, where most of the documents are articles and 449 meeting abstract documents are uncited.

4.2. Altmetric Score

Among the 3,291 documents, 1,961 have been integrated into social media, predominantly comprising articles and chapters. Of these, 1,751 documents have received altmetric scores, while 210 are pending evaluation. OA journals have garnered the highest altmetric scores, totaling 12,633 from 782 documents, with green access leading at 5,768 scores from 226 documents. Gold access follows with 3,599 scores (221 documents), hybrid access with 2,834 scores (195 documents), and bronze access with 432 scores (100 documents), reflecting fewer publications engaging with social media. Subscription-based journals received 8,856 altmetric scores from 1,179 documents.

Table 2 presents descriptive statistics for traditional citations received by pesticide research literature, categorizing documents into subscribed and OA. Subscribed documents show a publication range of 24 to 274, with a mean of 203.55 (standard deviation [SD]=66.21) representing a moderate level of variability around the average, while OA documents range from 25 to 155 publications, with a mean of 95.64 (SD=37.22) representing more concentrated distribution around the average. For subscribed documents citation, citations range from 45 to 8,125, with a mean of 3,424.91 (SD=2,331.53) indicating a wider variability in citation counts. OA citation documents have citations ranging from 53 to 3,122, with a mean of 1,900.36 (SD=925.67) designating a more concentrated distribution of citation values around the mean. These statistics reveal differences in citation patterns, indicating varying degrees of attention from researchers and reflecting diverse levels of impact, influence, or recognition within the academic community. In contrast, lower dispersion suggests a more consistent level of citation across the publications.

Table 3 outlines altmetric scores for a dataset of 1,961 documents, which has various dimensions of online attention. Mendeley readers mean scores (51.97, SD=71.448) display the highest number of times a research document has been saved or ‘read’ by users, followed by Dimensions citations (27.38, SD=43.660). The number of citation count of 22.92 was less than Mendeley readers and Dimensions citations, but higher than news, blog, Twitter, and Facebook mentions. While analyzing the SD counts, Facebook (89.324) obtains varying level of attention from researchers, followed by Twitter mention (87.214), Mendeley readers (71.448), and Dimensions citations (43.660). While citation is a significant metric for traditional scholarly impact, the altmetric scores, representing online attention and engagement, divulge that on average, documents may obtain more attention on platforms such as Facebook, Mendeley, Dimensions citations, and Twitter compared to traditional scholarly citations.

Levene’s test was conducted to evaluate the assumption of equal variances between OA and CA journals. A p-value of 0.102, which exceeds the typical significance threshold of 0.05, indicates that there is no significant difference in variance between the two groups. The t-test was subsequently applied to assess whether a significant difference exists in mean citation counts between OA and CA journals. The negative t-test value indicates that the mean citation count for OA journals is lower than that of CA journals. However, despite both the Levene’s test and t-test showing p-values below 0.05—implying statistical significance—the observed difference in mean citations did not translate into a significant variation in citation rates between OA and CA journals. The 95% confidence intervals provide a range within which the true difference in means likely falls, underscoring the importance of cautious interpretation (Table 4).

A Spearman’s rank correlation analysis was conducted to examine the relationship between citations and altmetric scores across various sources, including news, blogs, Twitter, Facebook, Mendeley readers, Dimensions citations, and total citations (Table 5). The results showed that blog mentions had a weak positive correlation with traditional citations, while Dimensions citations, Mendeley readers, and Facebook mentions demonstrated strong correlations above 40%. Twitter mentions had the highest correlation with citations at 99%, suggesting that researchers may bookmark these works for future reference.

New mentions were highly correlated with all social media sources except Facebook (0.03). Blog mentions showed weak correlations with most social media sources, except for strong correlations with news (79%) and Twitter (67%). Twitter had weak associations with Facebook, Mendeley, and Dimensions citations but was more closely linked with blogs (69%) and news (67%). Facebook exhibited minimal connections to news, blogs, and Twitter but had strong correlations with Mendeley and Dimensions citations. Mendeley readers were highly correlated with Dimensions citations (94%) but weakly correlated with other sources. Similarly, Dimensions citations had strong associations with Mendeley and Facebook but weaker correlations with other mentions.

The determination of statistically significant differences between altmetric scores and citations is verified through the Mann-Whitney U Test, as depicted in Table 6. The results reveal significant differences found in altmetric scores and citations, except for Dimensions citations (0.650), which is evidenced by p-values less than the predictable significance level of 0.05. Negative Z-scores suggest that the altmetric scores are generally lower than the citations. The Mann-Whitney U Test results emphasize prominent divisions in the relationship between altmetric scores and citations across distinct categories.

Table 7 summarizes the differences in altmetrics and citations between CA (subscription-based) and OA journals across various categories. CA articles outperformed OA articles in news mentions, with a higher average rank (951.96 vs. 1,024.79) and total performance (1,122,356.50 vs. 801,384.50). A similar trend was observed for blog mentions, where CA articles also had a higher average rank (958.37 vs. 1,015.12) and total performance (1,129,919.50 vs. 793,821.50). However, OA articles performed better in Twitter mentions, with a higher sum rank despite a lower mean rank (915.03). For overall citations, OA articles had a higher average rank (1,040.96 vs. 941.23) but lower total performance (814,030.00 vs. 1,109,711.00), suggesting greater visibility and impact for OA publications. Lower values for both metrics generally reflect better performance, emphasizing the increased reach of OA articles within scholarly and online communities.

Table 8 provides a comprehensive assessment of publication metrics for diverse toxicology journals, enabling a thorough comparison of their performance in terms of visibility, impact, and accessibility. Notably, the top 10 journals exhibit significant contributions from the Netherlands and the USA, both recognized as leading exporters of agricultural products. Among these, the journal Ecotoxicology and Environmental Safety, based in the USA, emerges as the foremost contributor, marked by extensive citations and a notable H-Index (161). Despite its prominence, the journal predominantly publishes under CA, showcasing a robust Impact Factor of 7.129. However, it is observed that this journal has a limited focus on social media engagement. Conversely, the UK-based Food and Chemical Toxicology stands out with the highest altmetric score of 2,544, complemented by substantial H-Index values (192) and a commendable Impact Factor of 5.572. The journal’s prominence on social media is remarkable, evident through a significant altmetric score, and its contributions, though predominantly in CA journals, receive the highest citations (1,766). An overarching observation is that these impactful works are predominantly featured in subscription-based journals, indicating widespread subscription among various institutions. This information is invaluable for researchers, institutions, and readers keen on understanding the influence and reach of these journals within the field of toxicology.

Table 9 unveils a curated list of the top 10 research articles concerning the impact of pesticides on human health. The most cited article, “Pesticides and human chronic diseases: Evidence, mechanisms, and perspectives,” authored by S. Mostafalou and published in the journal Toxicology and Applied Pharmacology in 2013, has garnered the highest citation count of 708. Despite its citation success, the article has a relatively modest altmetric score of 251, suggesting untapped potential for greater social media visibility and impact are highlighted in bold letters. A pattern emerges where top articles, primarily published in the journal Toxicology, receive comparatively lower altmetric scores. This indicates an opportunity for these impactful research works to enhance their visibility and societal impact by leveraging social media platforms. Such efforts could contribute to increased awareness and engagement among both researchers and non-academic audiences. These findings offer valuable insights into the performance and characteristics of leading toxicology journals, as well as the potential for maximizing the societal impact of influential research articles on pesticide-related human health impacts.

Table 10 represents the top articles that received more altmetric scores. The article “Exposure to glyphosate-based herbicides and risk for non-Hodgkin lymphoma: A meta-analysis and supporting evidence,” published in 2019, received the highest altmetric score of 2,928 with the least citation number of 156. This suggests that the article’s amalgamation in social media has created more impact than in traditional citation within the short period. The article “Heavy metals and pesticides toxicity in agricultural soil and plants: Ecological risks and human health implications,” which was published in the recent year 2021, received more citations and received a moderate altmetric score specifying the quality of work. It can be merged into social media to get more visibility. Most journals specializing in toxicology-related subject areas occupy top-ranking positions. The highest altmetric-scored articles are recently published in the years 2018 to 2023, with a mere number of citations showing the dominance of social media against traditional citation.