International Collaboration in Latin American Psychology Through the Analysis of Co-authorship Networks

More than half of a century has passed since a paradigmatic change in the scientific endeavour took place: from science that advanced through the discursive interchange of geographically dispersed research findings and reflections, to highly interconnected and collaborative efforts that transcend borders (Wagner & Leydesdorff, 2005, Research Policy, 34:1608–1618). These new “invisible colleges” have been recognized since the end of the twentieth century as the network structures that emerge from co-authorships in scientific publications. These network structures that also inspire further production of knowledge, diffusion of concepts, methodologies and theories, circulation of scientific metaphors, and, above all, knowledge transfer, transformation, and appropriation. The expected impact of scientific collaboration on the consolidation of national capabilities for producing and absorbing scientific knowledge has resulted in the implementation of diverse instruments that aim at encouraging collaboration, particularly among peers from different countries. Psychology has also seen this increased international collaboration, and to evaluate its impact, we used bibliometric techniques on Latin American articles published between 2016 and 2020 in journals indexed in the Web of Science (WoS) under the subject categories related to psychology, to explore the characteristics and conditions of scientific collaboration in this corpus.

Bibliometric studies enable the creation of overviews of scientific production within a given timeframe and field of research through indexes derived from analysing different bibliographic materials such as scientific articles, books, proceedings, and programs (Morgado-Gallardo et al., 2018;Quevedo-Blasco & López-López, 2010). These studies can be used as tools to improve the quality of research practices and create and implement data-based development strategies that can be used to guide, for example, funding decisions (Quevedo-Blasco & Ariza, 2013). They can also be used to diagnose and understand scientific practices. Some of the commonly studied aspects in these studies are theoretical trends, productivity, nationality and gender of the authors, collaboration networks, among others (Gallegos et al., 2020).
One of the more interesting items that bibliometric studies can shed light on is internationality, that is, the level of collaboration amongst teams from different countries, and the nature of the research questions addressed by these teams, taking the contextual particularities of each country into account. Citation databases have historically had a much larger proportion of scientific output from North America, Western Europe and, to a lesser extent, Australia, and New Zealand, than other regions. This is explained by cultural and socioeconomic reasons such as a higher proportion of national wealth invested in scientific infrastructure and English being the lingua franca of science (Hermes-Lima et al., 2008); however, it also reflects inequalities in terms of indicators such as the number of training and research facilities, trained scientists, and dissemination of knowledge, all relative to population numbers.
Latin America is one such region, which has been steadfastly growing in terms of scientific infrastructure (Adams et al., 2021), without a corresponding growth in scientific visibility and collaboration. According to López-López (2019), in Latin America, international collaboration takes place amidst stark knowledge asymmetries, which impacts scientific development and communication. Hence, scientific output is affected by the financial, economic, and political sways of Latin American countries , besides other traditional variables that have a detrimental impact on regional means for scientific communication such as lower journal impact factors (IF), language (Spanish and Portuguese), and the countries where the research is conducted and where the journals are edited (Quevedo-Blasco & Ariza, 2013;Salas et al., 2019;Tortosa-Pérez et al., 2019). In top of this, publication pressure in constrained publication outputs opens the door to the involvement of aesthetic variables when selecting those contributions that could eventually be published (see Giner-Sorolla, 2012 for a detailed analysis).
For these reasons, the dynamics of knowledge production and consumption in Latin America differ from international trends (Alonso Gamboa & Sánchez Islas, 2005;Delgado, 2011). One example is the scientific output in Psychology in Latin America, where the creation and maintenance of cooperation networks has been difficult. Besides a long-standing tradition of single-author publications (López-López et al., 2011), studies have shown indications that international collaboration is scarce in this field. For instance, non-international collaborations between 2005 and 2007 accounted for 92% of all scientific output in Latin American Psychology . And whenever the number of collaborative papers increases, Trends in Psychology (2023) 31:503-519 collaborations often take place between authors from the same country (Barboza-Palomino et al., 2021).
This relative lack of international collaboration in Latin American Psychology (Chinchilla-Rodríguez et al., 2010;García et al., 2015García et al., , 2017García et al., , 2019 has a negative impact on the visibility of the knowledge produced in Latin America and therefore on its accessibility and citation (Ochoa Henríquez, 2004). This situation is in stark contrast with the high popularity of psychology in Latin America and the potential global relevance of its contributions, given that the region has highly qualified researchers working on significant topics such as poverty, inequality, prejudice, corruption, and violence (Ardila, 2018). Therefore, more collaborative work based on the diversity of socioeconomic and sociopolitical realities of the continent is needed, through exchange and collaboration (Hwang, 2008;Koller et al., 2008;Wagner, 2006).
In this context, professional associations have emerged with the goals of favouring scenarios of collaboration, such as the Iberoamerican Federation of Psychology Associations (FIAP for its Spanish acronym) or the Interamerican Society of Psychology (SIP), which have managed to serve as important spaces where research and collaboration networks that link academics from several countries can be established (Cudina et al., 2017;. Supplementary infrastructures for the dissemination of scientific knowledge have also appeared, such as SciELO (Scientific Electronic Library Online), Redalyc (Latin American and Caribbean network of scientific journals), and Latindex, which have strengthened and fostered publication and communication of Latin American scientific output (Buela-Casal & López, 2005;Ochoa Henríquez, 2004). And finally, there is evidence that the creation of training programmes has an impact on scientific output, as they allow for researcher mobility and contribute doctoral theses to the literature pool .
These movements are key because this collaborative scientific work, besides increasing visibility, can have a positive effect on the creation and strengthening of an identity of Latin American science. However, there are other implications, both positive and negative, of this type of work for developing countries, which we will turn to next.

Advantages and Disadvantages of International Collaboration for Developing Countries
International collaboration is favourable for developing countries since partnerships with institutions and countries with higher rankings in citation databases help increase the former's visibility in an international context (Barboza-Palomino et al., 2021). On average, collaborative products have 1.22 times more impact than local ones, and international scientific cooperation in general increases impact by a factor of 1.59 in comparison to national-level cooperation (García et al., 2019). This is because they attract more queries and citations, which in turn gives them more visibility (Barboza-Palomino et al., 2021;López-López, 2010;. These data agree with Schmoch and Schubert's (2008) suggestion that international collaboration exposes papers to a broader audience. It also promotes higher productivity, in terms of both quantity and quality, and this makes it more likely that they will have better impact (Barboza-Palomino et al., 2021;García et al., 2015;Stallings et al., 2013). With increased cooperation, there is also a strengthening of activities aimed to address common and compatible goals amongst countries, nourishing collaborative research networks, improving access to international funding, and reinforcing the exchange of knowledge, resources and experiences that impact training programmes, researcher careers and research itself in the end (Catalá-López et al., 2014;García et al., 2016;. In line with this, UNE-SCO (2010) has stated that increasing the scientific capabilities of the developing world goes together with a larger number of international co-authorships.
Increased collaboration also helps research groups devise and tackle more complex questions that benefit from assembling multidisciplinary teams (Catalá-López et al., 2014), and allow for more diverse datasets, more sophisticated analysis techniques and access to more specialized equipment and incentives (Bozeman & Corley, 2004;. Local researchers get the opportunity to study different social and cultural realities, and this contributes to better, richer science, and increases the likelihood that communities that are more diverse will benefit from scientific output, ultimately affecting societal well-being (Buela-Casal et al., 2007).
However, there are risks that have been described by Kreimer (2016). For instance, teams from developed countries seem mostly interested in the results of research, and less so on the continuing local concerns in developing countries. This goes in line with concerns regarding the promotion of certain research agendas and the potential homogenization of methodologies and scientific practices that might come with international collaboration as well.  have acknowledged that scheduling data collection and writing can be sources of difficulty for collaboration, as well as the lack of physical interaction amongst research teams, because it seems to decrease trust (García et al., 2015). There are other obstacles in terms of physical infrastructures, bureaucracy, and lack of a vibrant culture of innovation and collaboration (Vílchez de Salazar & Flores Urbáez, 2004).

Increasing and Maintaining International Scientific Collaboration
Scientific collaboration increases through many factors such as better access to the Internet and other technologies, the internationalization of training programmes, higher incentives to collaborative work, the creation of regional citation databases, and more interest in editorial processes (García et al., 2015;López-López et al., 2015.) Specially, López-López et al. (2011) state that spaces where scientific publication can be discussed and debated benefits regional production and cooperation networks; these are also strengthened by creating and nurturing scientific communities that can breed collaborative research programmes and projects.
However, expanding international collaboration also involves more local, personal factors. For instance, a sense of self-organisation based on preferential bonds mediated by acknowledgement, recognition, reward, assertive communication and the pursuit and achievement of common goals (García et al., 2013). Previous contact and selection of collaborators, joint definition of research topics and goals, equitable allocation of research tasks, and bilateral agreements are also key factors in setting up and running international collaborative projects (da Silva, 2007).
Once collaboration networks have been established, they are maintained by scientific production and visibility and impact . Hence, it is useful that research management systems and national scientific policies in developing countries assist these efforts (García et al., 2015). This can be done through an ecosystem perspective on the creation, management, and dissemination of scientific knowledge like the one described by , such that external models that disregard local contextual factors are not applied as-is and damage the burgeoning international collaboration.
With that in mind, this paper analyses scientific collaboration, conceptualised as the production derived from co-authorship amongst researchers from different institutions or countries (Chinchilla-Rodríguez et al., 2010); that is, processes where two or more authors conduct research together (Aleixandre-Benavent et al., 2017). The main advantages of co-authorship as a measure of collaboration are its verifiability, invariability across time, availability, and ease of interpretation (Bozeman & Corley, 2004;Zanotto et al., 2016). It is also regarded as one of the most tangible ways of measuring cooperation, since it is well documented (Lemarchand, 2016), and other researchers may reproduce results more easily (Zanotto et al., 2016). This paper expands on a small number of previous studies, such as Russell and Ainsworth (2013).

Materials and Methods
Our analysis includes the set of scientific publications in journals indexed under any of the subject categories in the Core Collection of the Web of Science (WoS-Core henceforth) with authorship from Latin American countries. The collection includes all journals in the traditional indexes: Science Citation Index Expanded (SCIe), the Social Science Citation Index (SSCI), and the Arts and Humanities Citation Index (AHCI). WoS-Core also includes documents in journals in the Emerging Sources Citation Index (ESCI), the Science and Social Science & Humanities Conference Proceedings Citation Indexes (SSCHCP) as well as books and book chapters included in the Science and the Social Sciences & Humanities Book Citation Index (SSHBCP).
We used the Advanced Search option in the Web of Science (WoS) to identify all documents published from 2001until 2020 within the subject area of psychology. 1 We identified 11 relevant subjects for our study: (1) Psychology; (2) Psychology multidisciplinary; (3) Psychology Clinical; (4) Psychology Experimental; (5) Psychology Developmental; (6) Psychology Applied; (7) Psychology Social; (8) Psychology Biological, (9) Psychology educational; (10) Psychology Mathematical; and (11) Psychology, Psychoanalysis. We filtered our results to account only for those with at least one author with an address from a country in Latin America. We retrieved 23.169 documents with the following distribution ( Fig. 1): We downloaded this information in text format to create co-authorship networks using VOSViewer, a free software that reads text-formatted bibliographic files and creates co-authorship visualisations (https:// www. vosvi ewer. com/). To illustrate the growing complexity of the networks, we used four time slices, which also represent a growing internationalization in the set of documents. Almost 85% of documents were written in collaboration, and international collaboration is present in 40% of all documents considered. For further analysis we used ISI.exe, freely available at http:// www. leyde sdorff. net/ softw are/ isi/ index. htm. This software converts WoS text files to spreadsheet files.

Results
After data cleaning and processing, we started describing the corpus and trying to get a bird's-eye view of Psychological scientific output in Latin America during the twenty-first century. Figure 2 shows a sustained growth of Latin American presence in the journals covered by the WoS-CC, especially in the last 5 years. However, this growth does not necessarily come with increased visibility. In fact, in the aforementioned period, the median percentage of cited documents (Mdn = 61%) is lower than the same figure in the 2005-2015 window (Mdn = 77%). Even when removing the last year, which is currently understandably lower because these documents have not had much time to be cited, the median percentage of citation does not change much. This means that even though more papers are being written in collaboration with international partners, they are not necessarily having an increased impact, which is concerning given that one of the main reasons to promote and engage in international collaboration is precisely more visibility. Let us turn to what explains this growth and the lack of a corresponding surge in citations. Figure 3 below shows how psychological scientific output has increased in the Latin American countries with the largest scientific capabilities in terms of numbers of psychologists, universities, and research facilities. Since 2001, Brazil, Mexico, Chile, Argentina, Colombia, and Peru have been expanding their international psychological production. Brazil being the largest and richest country in the region is naturally the main driver, followed by Mexico and Argentina. After a near plateau in the first few years, Colombia has also been showing sustained growth and is now comparable to Argentina. Peru has been doing better since around 2013 and continues a slow but continued trend of more international collaboration.
However, what Fig. 3 shows is probably a case of increased visibility. In 2015, the ESCI was introduced, and it covered many journals from the developing world where burgeoning international collaboration was being published. The sudden added coverage might help explain the updrift after 2014. If this is the case, it means that a small but steady flow of international collaboration in Latin American Psychology was waiting to be discovered before this, but also that it was probably not reaching the databases' top tiers -in other words, either that work was not being submitted to the highest-impact journals, or it was being mostly rejected by them.
To supplement the analysis, we will now turn to authorship practices. Figure 4 shows the evolution of the number of authors in the dataset. Figure 4 starts with the relatively high number of single authorships that has been a feature of Latin American psychology (and social and human sciences in general), especially before the turn of the century. From a high of 30% of papers written by a single author, the number has been slowly decreasing throughout the examined period, with the lowest figures in 2013 and 2014. The authorship profile has been shifting to one closer to that of health and physical sciences, where papers are often signed by two to ten authors. A slow and steady increase in papers with more than 10 authors is also noticeable.
The growth in multiple authorship is driven by an increase in international collaboration. Figure 5 below shows the yearly change in single authorship, intra-institutional co-authorship, national-level co-authorship, and international co-authorship.
The decline in the number of papers written by a single author is confirmed in Fig. 5, but it also shows a steady increase in international collaboration. The other shrinking category is national collaboration (all authors in the same country). This is Single author Two to five authors Six to ten authors More than 10 authors probably an indication of how authors have slowly realized that they need to team up to produce stronger research and have been doing so by reaching out to colleagues in their own institutions, and then probably taking advantage of international funding calls to do research with people in other countries.
To understand the practice of assembling research groups, we counted the number of papers according to the authors' regions, which we present in Table 1 below.
Clearly, psychology researchers have formed groups mainly with other countries in Latin America and the Caribbean (LAC), so that most papers feature collaborations with other people in the region. This is not necessarily like what happens in other fields but is at least partly explained by the heavy influence of Liberation Psychology and Participatory Action Research in Latin American Psychology (Adams et al., 2015); a core feature of these approaches is the need for contextually relevant research focused on the disadvantaged, in contrast to the traditional interest of cognitive and experimental social psychology on psychological processes at the individual and group levels. Hence, it would be expected that Psychology, and other social sciences, fostered more collaboration efforts with like-minded psychologists in Latin America, where they identified similar issues and more easily navigated linguistic and cultural obstacles.
However, in the last 10 or 15 years, as more researchers took advantage of both scholarships and institutional training programmes to pursue advanced education in Europe and the USA, more collaborations started to emerge between returned Latin American researchers with higher degrees and colleagues they met during their studies. This is what probably accounts for the second row of the table that shows 16% of papers written by Latin American psychologists together with teams in Europe. The main destination would have been Spain, where many people got (and are still getting) their advanced degrees. These collaboration networks can be further studied by visualizing production by country and representing the strength (i.e., the number of collaborations) in the connections between them. We can also explore the evolution of these collaboration networks, starting with Fig. 6 below. As mentioned before, the significant role of Brazil, Mexico, and Argentina as the main drivers of Psychological scientific output in Latin America during the first years of the twenty-first century is very clear. Countries that would grow in the following years such as Colombia and Chile are comparatively small. Brazil benefits from collaborations with diverse European and Oceanic countries, and the USA serves as a sort of linking hub for Latin American countries. It is important to highlight that it is in this period when Redalyc, arguably the largest repository of scientific literature in Spanish and Portuguese at the time of this writing, is founded. Redalyc (which stands for "Latin America and the Caribbean Scientific Journals Network") was committed to open science from the beginning, which helps it grow exponentially throughout the following years.
We can look at what happened in the following 5 years in Fig. 7: Interestingly, although Brazil and Mexico continue to grow, countries like Colombia and Spain become larger players and links between Latin American countries thicken and the countries gather closer together. Chile starts behaving similarly to Brazil in terms of collaboration, and emerging actors start to show up, such as El Salvador and Bolivia. Two possible explanations for this are the changes in institutional incentive practices in Colombia and Chile, which drive up the interest in publishing in scientific journals, and the increased flow of Latin American researchers to Spain to get doctoral training, which is promoted by universities as they want to further qualify their faculty and has the distinct linguistic advantage (see for example Feld et al., 2013). Arguably, the "Redalyc effect" of increased visibility (hence citation, calls for collaboration) and growing interest in publishing also starts to take hold (Fig. 8). Figure 9 clearly shows a clustering of Latin American countries. The USA continues to provide strong opportunities for collaboration, but the collaborative gateway is clearly through Brazil, which at this point has been not only founding many new journals, but also seeing a spectacular increase in the number of scientists trained at the doctoral level through policy-related incentives (Feld et al., 2013). Interestingly, Colombia drifts slightly away from the Latin American cluster (even though it maintains strong research links with them and the Iberoamerican world), and this has to do with the booming interest in studying its internal conflict, which promotes significant international collaborations.

Discussion
Latin American psychology has been experiencing an increase in scientific output and international collaboration. This is especially visible in countries such as Colombia, Chile, and more recently, Peru. Clearly, countries with stronger management and incentive infrastructures have played a leadership role (Brazil, Mexico, and Argentina for instance).
From the perspective of knowledge ecosystems (López-López, 2020; López-López & Rooryck, 2021), we can look at the interactions between different actors in the scientific world and how those interactions influence scientific output and practices -international collaboration in our case. For instance, database inclusion policies such as the creation of the ESCI and its coverage within the WoS-Core helped give visibility to an undercurrent of international collaboration that was not making it into the top tiers of the publication system. Of course, other factors come into play, such as institutional policies of faculty qualification that enable knowledge transfer processes which impact the type of research that scientists embark on. This, in turn, changes the dynamics of international collaboration as well.
In this sense, it seems evident that collaboration and production are intertwined, and that the authorship profile has been changing towards multi-authorship. The most collaborations and general output are still being driven by Brazil, Mexico, and Argentina, and this is probably related to the size of their institutions, their funding and governance infrastructures, their training facilities, and the sustainability of all these infrastructures. The relative deceleration of Argentina and Mexico is explained both by the features of their ecosystems (affected by sociopolitical and socioeconomical developments in the past few years) and by the growth of other countries such as Colombia, Chile, and Peru. On balance, there is a relationship between increased co-authorship, collaboration, and production.
As we have seen, bibliometric data can be mapped to financial, political, and cultural changes in national, regional, and international scientific dynamics. Through studies like this, scientific communities can delve into their collaboration practices and relate them to science and technology policies, evaluation and accreditation systems, the pressures they put on institutional governances, the impact of international organisations and rankings, and how all these have an influence on those practices.