The concept of competence in educational research: a knowledge map of main research traditions and topics Open Access

Driven by the fundamental question of how learners apply what they have learned beyond educational settings, and accelerated by the adoption of competence-based approaches in education systems worldwide, competence has been of major interest in educational research for over half a century (Brauer, 2021; Le Deist and Winterton, 2005; Mulder et al., 2007). Despite a growing body of primary studies offering valuable insights and practices, secondary studies on this concept remain extremely limited. This is especially disadvantageous for a field as broad and fragmented as competence research, where diverse conceptualizations, methods and application contexts make it difficult to track cumulative knowledge, identify dominant traditions or detect blind spots.

Despite this lack of overarching synthesis, a few targeted review studies have begun to explore specific areas of competence research. Müller et al. (2020) reviewed the assessment methods of social, emotional and intercultural competences in upper secondary education. Ilomäki et al. (2016) examined how digital competence was conceptualized in studies focusing on primary to upper secondary education. Van Ginkel et al. (2015) reviewed the learning environment characteristics related to the effective development of oral presentation competence in higher education.

These studies reveal the diversity of issues that concern competence research. They show that it (1) relates to different levels of education, from primary to higher education, (2) concerns competences relevant across life and professional domains and (3) addresses varied elements of education and training, such as learning environments and assessment. Regardless of these diverse premises, however, we argue that competence research should be viewed as a broad, distinct research domain, for its studies are faced with the same fundamental question: how to best organize education so that learners can apply what they have learned in real-world contexts, i.e. develop competence. Because existing reviews remain narrow in scope, they offer only fragmented insights. As a result, we still lack a comprehensive understanding of competence research as a whole, particularly its main traditions, development and core topics.

We address this gap by mapping, clustering and visualizing prior competence research, regardless of any specific education type, competence or aspect of educational implementation, to provide the first comprehensive overview of this research domain for researchers, policymakers and practitioners. We position our research as a mapping study, that is, a secondary study that thematically explores and classifies earlier research using quantitative and visualization methods (Kitchenham et al., 2011). Mapping studies provide structure to complex literature by clustering studies into groups and creating spatial representations of the classifications that emerge, analogous to geographic maps (Zupic and Čater, 2015). Compared to traditional literature reviews that synthesize evidence from empirical primary studies to answer highly specific research questions, mapping studies aim to identify and classify large quantities of research related to a specific domain, providing insights about its extent, trends and research topics (Fellnhofer, 2019). They are especially recommended when a field is broad, heterogeneous, still emerging or has not yet been extensively reviewed (Kitchenham et al., 2011). As competence research exhibits all of these characteristics, we propose that a mapping study offers a valuable review approach to advance future competence-based research, policy and practice.

Mapping studies typically include far more primary studies than traditional reviews. As such, they rely on methodologies suited to large datasets, where scale and diversity make manual coding unfeasible. To that end, advanced bibliometric methodologies have recently been utilized to review substantial educational research domains, such as educational technology (Shen and Ho, 2020), multimedia learning (Li et al., 2019) and entrepreneurship education (Fellnhofer, 2019). Among these techniques, citation analyses examine relationships among articles, authors and journals using quantitative and visualization methods (Fellnhofer, 2019). These methodologies presume that citations are valid and reliable indicators of scientific knowledge transfer, and that they provide insights into the relationships and distances of the concepts that the articles represent, i.e. their latent intellectual structure (Tang et al., 2014). Another common method is term co-occurrence analysis (TCA), which uses statistical analyses and visualizations to explicate patterns of association among core terms in a research domain, enabling the identification of research topics, wider research areas and developmental patterns over time (Cheng et al., 2014). The strengths of advanced bibliometric methods are that they offer high levels of objectivity and reliability, characteristic of quantitative research, and provide researchers an efficient way to analyse large datasets and identify patterns that would otherwise go unnoticed (Zeng and Chini, 2017). However, no such bibliometric synthesis has yet been conducted in the field of competence research.

To address this gap, we apply direct citation network analysis (DCNA) with mapping, clustering and visualization to identify the main competence research traditions and describe their scope, historical development and most prominent works and journals. We then extend the DCNA with TCA to uncover the core and emerging research topics that these traditions have focused on over time. The resulting competence research knowledge map is intended to support researchers, policymakers and practitioners in navigating this complex and rapidly expanding field.

In educational literature, the concept of competence typically includes two elements that have remained relatively consistent across definitions over the last 2 decades (Baartman et al., 2007; Koenen et al., 2015). First, competence consists of multiple integrated components – usually knowledge, skills and attitudes, and sometimes extended to include concepts such as values, motivation and emotions (Hoskins and Crick, 2010; Mulder et al., 2007). Accordingly, competence pertains to different domains of human functioning, such as cognitive, affective, psychomotor, psychosocial and motivational (Rieckmann, 2012; Wesselink et al., 2010). That the components are “integrated” mainly acknowledges the interrelatedness of these domains (Baartman and De Bruijn, 2011).

Second, competence is inherently contextual: It requires real-world situations for application (Baartman et al., 2007; Hoskins and Crick, 2010). A common distinction is made between generic (transversal, core, key) competences, which apply in various everyday situations, and professional (vocational) competences, which are tied to specific work contexts (Starkey, 2020). As such, competence has been described as operating “at the interface between the person and the demands of the real world” (Hoskins and Crick, 2010, p. 122) and as “a function of the context in which [it is] applied” (Wesselink et al., 2010, p. 22).

The above characteristics of competence have several general implications for educational practice. To ensure that the competences outlined in a curriculum are applicable in real-life contexts, they need to be anchored to real-world problems, tasks or demands. In vocational and professional education, for instance, field-specific competences are frequently defined through analyses of workplace demands with input from employers (Mulder, 2012). Across many countries, key competences promoted by the European Union (EU) and the Organisation for Economic Co-operation and Development (OECD) have been adapted to national curricula, spanning early childhood to higher education. Unlike traditional content-based curricula, these competences are cross-curricular and presume interdisciplinary teaching and learning.

Currently, competence approaches are associated with self-regulated, authentic and social constructivist learning philosophies (Cremers et al., 2016). Self-regulated learning emphasizes learners’ autonomy through goal-setting, planning and self-evaluation (Koenen et al., 2015). Teachers, in turn, take on the role of facilitators or coaches, offering ongoing, personalized guidance tailored to students’ needs (Van der Baan et al., 2022). As per the applicability requirement, competence approaches rely on authentic, simulated and interdisciplinary learning activities that are derived from practice and enable learners to gain hands-on experience and participate in communities of practice (de Bruijn and Leeman, 2011).

Competence also poses distinct challenges for assessment. Because competence involves integrated components applied in real-world contexts, assessment must go beyond traditional testing to evaluate learning and performance in authentic situations (Baartman and de Bruijn, 2011). While the development of robust assessment tools is still evolving, researchers have proposed a mix of formative and summative methods, including performance assessments, portfolios, personal development plans, conceptual maps and reflective diaries (Baartman et al., 2007).

This study aims to provide the first comprehensive overview of competence research for researchers, policymakers and practitioners. It provides a structured and visual knowledge map with an extensive library of further reading. We classify competence research at two analytical levels—citation patterns and research topics—using DCNA to map main traditions and TCA to explore their thematic focus. The research is guided by the following questions:

We used the Scopus database by Elsevier for document retrieval, because it offers the most journal coverage for citation analyses, particularly in the field of social sciences. Acknowledging the diversity of scopes and terminology related to competence research, we adopted the following broad criteria to identify suitable studies: (1) The term competence (or competency) and (2) a population (e.g. students or teachers), context (e.g. education or school) or aspect (e.g. curriculum or assessment) that implied education or training needed to appear in the title, abstract or keywords. To ensure the relevance of our search query, we searched Scopus for systematic competence reviews in the social sciences (62 reviews), extracted the search terms referring to competence or education (82 terms) and combined them into our own query (Mannonen et al., 2025). We applied three inclusion criteria using the following filters: (1) subject area: social sciences; (2) document type: article, review, book chapter, conference proceeding; and (3) language: English. The final search query was as follows:

This search strategy yielded 135,298 documents on 20 June 2025 (Mannonen et al., 2025). While comprehensive, this volume proved too large for the visual representation of the research domain. Since there are no established guidelines for sampling in citation network visualization, we experimented with various approaches—including selecting only the most cited documents and applying systematic random sampling—to identify which method produced the most coherent and robust results. Ultimately, we chose to base our sample on citation links (Section 4.2). This approach aligns with the idea that highly interconnected documents are more likely to reflect widely relevant concepts within the research domain (Tang et al., 2014). Following extensive testing of various sample sizes, we identified the 15,000 most linked documents as achieving the ideal balance between domain coverage and interpretability in visualization.

DCNA with mapping and clustering is a quantitative method that projects a network of documents in two dimensions and clusters similar documents based on their citation relationships (van Eck and Waltman, 2017). DCNA treats two documents in the network as linked if one of them has directly cited the other, and the links between documents are assumed to be indicative of their conceptual relatedness (Tang et al., 2014). The analysis produces a graph known as a direct citation network, where each node represents a document and the edges—lines connecting the nodes—indicate citation links between them. The sizes of the nodes are proportional to the number of adjacent edges: The larger the node, the more linked that document is to the other documents in the network (Waltman et al., 2010).

To perform DCNA and visualize its results, we used Scopus export files and VOSviewer software (version 1.6.20 from the website https://www.vosviewer.com/; accessed May 2025). The details on how VOSviewer performs the analysis are not repeated here but are well-documented in articles written by the developers of the software (e.g. van Eck and Waltman, 2017; Waltman et al., 2010).

Like DCNA, TCA is a quantitative method that relies on statistical analysis and visualization techniques (Cheng et al., 2014). While DCNA presumes that citation links between documents are indicative of conceptual relatedness between the documents, TCA assumes that the co-occurrence of terms is indicative of conceptual relatedness between terms (Li et al., 2019). The resulting term co-occurrence network graph is similar to the direct citation network graph, but the nodes present terms, and the sizes of the nodes are proportional to the number of their occurrences. The edges now represent the co-occurrence of two terms (Zeng and Chini, 2017). Based on the terms’ positions, co-occurrence links and clusters on the visualized graphs, research topics may be identified.

We analysed document titles and abstracts, which contain more nuanced information than keywords and thus yield more precise and reliable results (Zeng and Chini, 2017). For each research tradition, we included the 500 most frequently occurring terms in the visualization. Using binary counting, multiple instances of the same term within a single document were counted only once. The TCA was conducted using the same version of VOSviewer as the DCNA.

We found four main competence research traditions (clusters) based on the DCNA: Healthcare Education (HET), General Education (GET), Language and Intercultural Education (LIET) and Developmental and Educational Psychology (DEPT) (Figure 1; Supplementary Material 1; Mannonen et al., 2025). Encompassing 14,546 documents in total, these four research traditions formed the largest connected group of documents in the sample. Table 1 provides a summary of the research traditions.

HET was the largest research tradition, with 6,603 documents published in 1977–2025. These studies covered the education of healthcare professionals such as physicians, dentists, pharmacists and nurses. Most of the earliest documents focused on competence assessment (e.g. Harden and Gleeson, 1979), which has remained a staple topic in this research tradition to this decade (e.g. ten Cate et al., 2024). For example, in this research tradition’s second most linked document, Van Der Vleuten (1996) discussed developments in the assessment of professional competence in health sciences education. Many assessment-related terms, such as score, test and performance, were among the top terms and represented an older area of focus in the research tradition (Figure 2; Mannonen et al., 2025).

A clear shift towards competency-based curriculum in healthcare education can be observed from the documents that were published close to the turn of the millennium, when a steep upwards trend in the yearly number of documents also began. In one of the most linked documents published around that time, Carraccio et al. (2002) reviewed medical education literature to inform educators on how to implement competency-based education, which they, at the time, argued was the new paradigm of medical education. Other authors, such as Wenzel et al. (1998), referred to a curriculum revolution. Published soon after, the most linked document of the research tradition provided guidance on the development and implementation of a competency-based curriculum (ten Cate and Scheefe, 2007).

While competence seems to have reached mainstream adoption in healthcare education by the early 2000s, not all researchers embraced the trend. In a highly linked critical position paper, Talbot (2004) argued that the competency discourse tended to “limit the reflection, intuition, experience and higher order competence necessary for expert, holistic or well-developed practice” (p. 587). This criticism relates to the origins of competence in behaviourist theories of learning, which, according to multiple authors in this research tradition, reduced complex human attributes and professional expertise to simplistic, measurable behavioural objectives that neglected the whole, particularly in highly skilled professions (e.g. Hawkins et al., 2015). Furthermore, competency-based education was criticized for being more geared towards assessment, regulation of proficiency, and standards than actual teaching or learning (e.g. Morcke et al., 2013). Even so, the assessment of competencies that successfully reflected the complexity of unique circumstances proved challenging, potentially limiting curricula to only what is measurable (e.g. Malone and Supri, 2012).

Simulation-based learning was one of HET’s most rapidly emerging trends based on the TCA. It first started appearing in documents from the 1980s, gained steam with digitalization at the end of the 1990s and became a staple topic in the research tradition by the 2000s (e.g. Decker et al., 2008). In one of the research tradition’s most linked articles, Issenberg et al. (2005) argued that simulations had become widespread in medical education and reviewed the features and uses of medical simulations that lead to effective learning. Recently, there has also been increasing interest in competence review studies in HET, as exemplified by emerging terms such as systematic review, scoping review and database.

GET was the second-largest research tradition, with 3,976 documents published in 1977–2025. The studies within it focused mostly on teachers and learners in different types of formal education. GET was largely concerned with the question of how to prepare learners for an uncertain future characterized by digitalization and climate change, and what is required from the teacher in the process (e.g. Starkey, 2020).

Like in HET, multiple of the earliest documents focused on assessment (e.g. Messick, 1984). In the tradition’s first truly active decade in the 1990s, authors focused on conceptualizing competence and discussing its implications for educational practice (e.g. Hager and Gonczi, 1996). Notably, most of the documents from this time were extremely critical towards competence as implemented in the United Kingdom (e.g. Hyland, 1993). The concept was challenged on multiple epistemological, ethical, ideological and practical grounds, which are much in line with the criticisms presented in HET around the same time. First, competence was seen to reduce complex professional expertise and human capabilities into atomistic measurable tasks (e.g. Lum, 1999). Second, to the authors, competence represented behaviouristic and functionalist approaches to learning which undermined knowledge, understanding and values (e.g. Bridges, 1996). Third, the concept of competence itself was considered to be confusing and incoherent, e.g. in its relation to knowledge and skills (e.g. Westera, 2001). While authors in the GET continuously pinpoint issues related to competence to date, the yearly number of such documents has reduced considerably, and more favourable voices have appeared and argued for more holistic conceptions of competence (e.g. Mulder et al., 2007).

A major turn in the yearly number of documents in the research tradition took place in approximately 2005, around the time when the OECD (2005) and EU (2006) published their influential works on key competences. By the 2010s and to date, key competences have become a staple topic in GET. Most of the highly linked documents focused on either key competences in general (e.g. Voogt and Roblin, 2012) or specific key competences, such as digital competence (e.g. Falloon, 2020). Based on the TCA, top terms such as digital competence, digital tool and ICT are still heavily emerging in the research tradition (Figure 3; Mannonen et al., 2025).

Fuelled by climate change, research on sustainability education gained steam around 2010 and became one of the most popular topics of GET. Namely, the tradition’s four most linked documents focused on sustainability key competences (e.g. Rieckmann, 2012). For example, in the most linked document, Wiek et al. (2011) developed a framework of key competencies in sustainability to support the development of academic sustainability programs. TCA also showed sustainable development as an emerging research topic, as represented by top terms such as sustainability, sustainable development and sustainability education.

Except for the healthcare professions discussed earlier, GET also covered most vocational and professional education studies included in the sample. Focusing on the practical implementation and implications of competence-based education especially in the Netherlands, this branch of research started emerging more rapidly in the early 2000s, some years before the key competence branch (e.g. Mulder et al., 2007). For example, in their highly linked document, Biemans et al. (2004) identified pitfalls in the implementation of competence-based vocational education and training. This branch of research was represented by top terms such as professional, graduate and labour market.

LIET was the third-largest research tradition, with 2,173 documents published in 1968–2025. This research tradition was continuously concerned with the intertwined questions of second language learning and intercultural communication. Driven by globalization and increased migration, this research tradition saw a considerable hike in the yearly number of documents from approximately 2005 to the current day.

The competences studied in the language learning branch of the research tradition included, for example, second language, interactional and sociolinguistic competence. The most linked studies of this branch focused on second language learning, particularly in classroom contexts (e.g. Cekaite, 2007), and its assessment (e.g. Youn, 2015). Top terms representing this branch included second language, pragmatic competence and native speaker.

Closely related to the former branch, the intercultural communication branch covered different variations of cultural and communication competences. For example, the most linked document of LIET, by Hunter et al. (2006), conceptualized global competence with the Delphi method and developed a survey to measure it. Most of the competence concepts appeared in documents across decades, but two emerging concepts could be identified with the TCA (Figure 4; Mannonen et al., 2025): intercultural communicative competence and interactional competence. Recently, an emerging topic has also been the internationalization of higher education, as represented in the temporal map by the relatively new top terms of higher education, international student and intercultural learning.

Compared to the previous two research traditions, criticism of competence was much scarcer in LIET. There were critical voices, such as on how intercultural, communicative, language or global competence should be conceptualized and studied (e.g. Holmes, 2006), but competence itself was rarely outright rejected to the extent of the previous research traditions. Some issues were still highlighted. For example, Ku and Liu (2021) emphasized the importance of the development of identity in language teacher education and called for the integration of identities in competency-based approaches that, according to the authors, mainly focused on skills and behaviour.

DEPT was the smallest research tradition, with 1,794 documents published in 1967–2025. Represented by top terms such as child, preschool and early childhood, common target groups for these studies included young children in kindergarten, pre-school and primary education, often those with special needs. From the oldest documents to the newest, this research tradition concerned the development and socialization of young people and the role of socio-emotional competences in this regard (e.g. Raver and Zigler, 1997). In the most linked document of the current research tradition, Rose-Krasnor (1997) developed a model of social competence, which was also one of the research tradition’s most frequently occurring terms. Lately, the role of emotions in development and competence has gained more traction and become an emerging research area, as represented by top terms such as emotional competence, emotion regulation and emotional learning (Figure 5; Mannonen et al., 2025).

Most of DEPT’s frequently linked documents focused on the closely related concepts of motivation and self-determination (e.g. Ntoumanis, 2001). Both concepts had deep roots and were widely covered by the research tradition, as represented by multiple top terms, such as self-perception, perceived competence and competence belief. For example, Bandura and Schunk (1981) studied the role of proximal self-motivation in cultivating children’s competence and self-efficacy. Regardless of these deep roots, however, motivation and self-determination theory and multiple associated concepts were still highlighted as a trending research area by the TCA.

Overall, concepts such as social, emotional and perceived competence appeared to be well-established in DEPT. Like in LIET, documents outright rejecting the concept of competence were absent. However, like in the previous research traditions, the issues that were highlighted were related to assessment (e.g. Waters and Sroufe, 1983). Notably, Foschi (2000) argued that there were double standards in the assessment of a person’s competence, with those with lower status (based on gender, ethnicity and socio-economic class) facing stricter standards.

For over half a century, educational researchers have created a rich and diverse body of literature related to competence. Without a comprehensive synthesis of this research, however, it has been challenging to see the forest for the trees. One could argue that this has been particularly problematic for competence research, which has been characterized by conceptual inconsistencies and diverse practical approaches (Brauer, 2021; Le Deist and Winterton, 2005).

The aim of this study was to provide researchers, policymakers and practitioners with the first comprehensive and structured knowledge map of competence research. We first identified its main research traditions, of which there were four: Healthcare Education, General Education, Language and Intercultural Education, and Developmental and Educational Psychology (RQ1). Each tradition has seen a rapid growth in the number of documents in the 21st century. From the reviewed documents, it seems clear that the increased interest in competence has been driven by the megatrends of the rapidly changing world (e.g. digitalization, climate change and globalization) and policy efforts to adapt to those megatrends through education.

This study is not the first attempt to identify competence traditions. Le Deist and Winterton (2005) identified three dominant competence traditions that had emerged independently in different countries: the behavioural approach, the functional approach and the multi-dimensional and holistic approach. Similarly, Mulder et al. (2007) identified three main traditions to competence: the behaviourist, the generic and the cognitive. These studies classified competence traditions based on the theoretical conceptualization and practical usage of the concept in different countries. Our bibliometric approach complements them and provides a new classification that is based on knowledge transfer patterns in scientific literature.

We also uncovered the topics on which the research traditions had focused over time (RQ2). We identified several emerging topics in competence research, such as simulation-based learning, digital competence, intercultural communicative competence and emotions. We are not aware of any other studies that have mapped the research topics of competence research to the same extent. However, we have only scratched the surface, and multiple unresolved questions that were outside of the scope of this study remain. From the beginning, the purpose of this study was to provide a one-stop resource that enables others to dive deeper into the identified classifications. Some considerations for future research can still be provided.

The research traditions identified in this review have historically struggled with comparable questions that have been the centre of debates for decades: How should competence be conceptualized or assessed? How do we ensure that we do not lose sight of the complexity of human functioning and expertise when we focus on competences? What kind of values or notions of education do competence-based approaches represent? What kinds of implications do competence approaches have for the teaching profession and teacher education? What kinds of key competences should learners develop in future? These are all excellent directions for future reviews, for which we encourage other researchers to use our knowledge map as a starting point. The identified research traditions provide an interesting opportunity for cross-tradition comparisons.

Future competence research would also benefit from cross-tradition collaboration that combines the unique research perspectives of the four traditions. For example, as a safety-critical field, HET has deep roots in the topics of competence assessments and simulation-based learning. For a long time, this tradition has had to ensure, via assessment, that all graduates have the professional competences needed to treat patients safely. Similarly, simulations have offered a risk-free way to practice high-risk activities. Because of this long history and the extensive coverage of these topics, HET has likely made advancements that may prove to be equally useful in other contexts. Likewise, as the development of digital technologies, e.g. artificial intelligence, and climate change keeps advancing, researchers may want to consider the studies conducted in GET. In times of high international mobility, international crises and polarized world politics, the areas within LIET are becoming increasingly relevant in all contexts. With growing recognition of the importance of social and emotional competences (e.g. OECD, 2024), DEPT is well-positioned to contribute meaningfully to this evolving field of study. We would like to highlight that multiple such cross-tradition studies already exist, which readers can look for at the borders of the research traditions on the direct citation map.

We combined two state-of-the-art bibliometric methods to map the main traditions of competence research. Researchers may use the visual graphs (Figures 1-5; for interactive versions, see Mannonen et al., 2025), summary table (Table 1) and extensive bibliography (Supplementary Material 1) of this work as a competence research knowledge map to efficiently find the documents, authors, journals and research topics that are relevant for their work and derive novel ideas for investigation. They may dive deeper into the research traditions, use them as a starting point for further in-depth literature reviews and find directions for future research by identifying topics that have not yet been addressed. By making all our data openly available (Mannonen et al., 2025), we have ensured the replicability of this study and provided a baseline against which competence research trends can be tracked over time. Policymakers may use the knowledge map to evaluate the quantity and quality of research-based evidence about competence approaches for decision-making. Finally, by providing clarity and structure to complex literature in the form of an intuitive visual knowledge map, this overview may facilitate the transfer of findings, derived by educational researchers for over half a century, to practitioners, improving the research-based implementation of competence approaches in education.

The results of this study should be interpreted considering several limitations, which also point to directions for future research. First, due to the technical constraints of the tool used for DCNA (VOSviewer), we were limited to using the export file from a single database (Scopus). This is a common limitation in bibliometric network analyses (e.g. Li et al., 2019). Second, the analysis relied on two widely used tools—Scopus export and VOSviewer—to accurately reproduce the citation links of 135,298 documents. Given the scale of the dataset, manual verification was not feasible. However, both tools have been extensively tested and validated by the scientific community, and we found no evidence—either direct or reported—of inaccuracies in their outputs (van Eck and Waltman, 2017; Waltman et al., 2010).

Third, even though citations remain the oldest, most established and clearest indication of knowledge transfer between scientific works, citation practices exercised by researchers are biased by multiple factors, such as document age and self-citation (e.g. Zupic and Čater, 2015). Citation links should therefore be interpreted as indicators of conceptual relatedness rather than scientific quality or value. Older foundational works may be underrepresented if their impact is indirect or uncited within the domain. Future studies could build on our work by incorporating multiple databases to broaden coverage, applying complementary qualitative analyses to validate cluster interpretations, and using hybrid citation measures to mitigate known biases.

The supplementary material for this article can be found online.