Patterns and trends in engineering education in sustainability: A vision from relevant journals in the field

1 Patterns and trends in Engineering Education in Sustainability: a vision from relevant journals in the field Abstract Purpose To identify patterns and trends taking place in engineering education in sustainability, through analyzing the evolution of research conducted in relevant publications in the field of engineering education for sustainability in the last decades. Design / methodology / approach Firstly, a bibliometric approach has been applied, adopting a co-word analysis based on cooccurrence of the keywords (300 items) in articles from three indexed journals related to engineering, education or sustainability. The selection of the articles has been based on the appearance of the previous three terms in the topic and title fields of the journal, where journal scope (based in the categories of the InCites Journal Citation Reports) covered at least two topics and the third topic was applied in the search, as follows: - International Journal of Sustainability in Higher Education. Scope of the journal: sustainability and education. Keyword search: engineering (20 papers) - Journal of Cleaner Production. Scope of the journal: sustainability and engineering. Keyword search: education (122 papers) - International Journal of Engineering Education. Scope of the journal: engineering and education. Keyword search: sustainability (29 papers) Secondly, to identify topological patterns and their evolution a structural and temporal analysis of the network of keywords and a categorization of the keywords in thematic clusters (named categories) have been performed. Results 2 The most relevant categories in terms of corresponding number of keywords, even though these have decreased in recent years, are those related with institutional and policy aspects to embedding or applying sustainability in higher education. At the same time, categories related to the professional development of faculty members, implementation and use of learning strategies (i.e., real-world learning experiences, educational innovative initiatives/tools/techniques) and cross-boundary schemes (i.e., transdisciplinarity, ethics, networking, etc.) increase their relevance in the last five years, signaling some of the challenging fields of interest in Engineering Higher Education in Sustainability in the near future. Practical implications Knowledge of the trends in devising sustainability education in engineering allows for designing curricular schemes and learning strategies to achieve competences, which are key factors for the change towards sustainability. Originality / value This research has a strong strategic value since it indicates the focus of future research efforts and networking on some of the topics of greatest concern in engineering higher education for sustainability. 1 Introduction Actions for sustainability have been promoted from the different areas of environment, society and economy, with the longed for common aspiration to face multiple interconnected sustainability crises in a world that can no longer be conceived as “society without nature and nature without society” (Beck, 1992; Latour, 1993; van Breda et al., 2016). In September 2015 countries adopted the sustainable development goals to end poverty, protect the planet and ensure prosperity for all as part of the 2030 Agenda for Sustainable 3 Development: Transforming our world, with specific targets to be reached. In this context appropriate technologies are just another instrument to achieve sustainability as a “holistic concept that requires the strengthening of interdisciplinary linkages in the different branches of knowledge” (A/RES/72/223). When dealing with interconnected problems, finding integrated solutions is a complex undertaking, reaffirming the need to tackle them with a transdisciplinary approach (van Breda et al., 2016) aimed at integrating expert or academic knowledge with the practical or traditional knowledge from actors outside of academia (Scholz et al., 2006; Brundiers et al., 2010; Brown, 2014) to co-produce outcomes that could be both socially useful for transitioning and scientifically innovative to formulate new guiding principles (Jahn et al., 2012; Lang et al., 2012). From the academic side, scientific journals and conferences continue to be important forums for exchanging scientific information on research findings, concepts, policies and innovative procedures or technologies. In reference to the former consideration of journals as exchanging forums,, 2017, Zhou et al. characterized the Journal of Cleaner Production (JCLP) publications, identifying “low/no-fossil-carbon transformations” as the main related issue in three of the four areas, in which the JCLP historical of publications focused, namely: a) industrial applications; b) environmental management initiatives; c) governmental environmental policies and regulations; d) education, training and facilitation of changes toward SD. In the present article, authors focus on the fourth identified area, while extending the array to two more indexed journals (introduced afterwards), related to engineering, education or sustainability. The International Journal of Sustainability in Higher Education (IJSHE) is a fullyrefereed academic journal aiming since 2000, to document and disseminate what universities and colleges are doing to pursue the path of sustainable development. The JCLP is a transdisciplinary journal founded in 1993, focusing on cleaner production, environmental, and sustainability research and practice. The International Journal of Engineering Education 4 (IJEE) is an independent journal, serving as an archival forum of scholarly research related to engineering education since 1994. On the other hand, referring to conferences as exchanging forums, Segalàs and colleagues (Segalàs et al., 2018) analyzed the evolution since 2002 and future challenges of the Engineering Education in Sustainable Development Conference, as an exchanging platform of concepts, policies and strategies to enhance a sustainable education, through the characterization of the published papers, the identification of topic categories and interviews to experts. Relevance declining categories were related with environmental topics and management and policies. On the contrary transdisciplinarity, topics from humanities and circular economy increased their relevance. Additionally transdisciplinarity was considered crucial to improve engineering education in sustainability, in parallel to real implementation at universities and networking. In the literature review priority similar considerations can be distilled towards the academic need to transition from a focus on technical issues to more ‘messy’ problems that require an integrative, adaptive, collaborative, ultimately transdisciplinar approaches in the interface science-society (Clark and Button, 2011; Balsiger, 2014; Vilsmaier and Lang, 2015; Tejedor and Segalas, 2018), which necessitate new methods and tools, but also the development and teaching of new engineering paradigmatic schemes (Byrne et al., 2013; Halbe et al., 2015; Remington-Doucette et al., 2013). This paper aims at go deeper into these last considerations with the use of complex networks science based on paper keywords co-occurrence analysis (Newman, 2010). We additionally examine the temporal evolution and penetration process of the fields of Interand Transdisciplinarity into dynamic networks with the goal to identify how these fields are affecting Engineering Education in Sustainability. The paper is organised as follows. In the Materials and Methods section, we present our database, the creation process of our network, following a co-word analysis of the keywords of papers related to Engineering Higher Education in Sustainability (EESD). We explain the 5 centrality measures used to characterize the topology of the network and to assess the evolution of its structural features. We also present and identify the thematic categories. In the Results section, we detect structural patterns and particularly central keywords, and relate them to our EESD categories, with the objective of explaining and detecting differences between transdisciplinar collaborative frameworks towards sustainability. We finish with the Discussion section, where we recap and identify some key trends for future work. 2 Materials and methods With the advent of the indexing and availability of scholarly documents and literature, we have begun to quantitatively understand the process by which scientific fields emerge (Börner, 2010, 2015; Bettencourt and Kaiser, 2015). These quantitative methods commonly include one or a mixture of population contagion dynamical models for the spread of ideas and the emergence and development of scientific fields (Bettencourt et al., 2006, 2008; Bettencourt and Kaur, 2011) and the structural analysis of networks. This can be about collaboration between scientists (Newman, 2001; Barabási et al., 2002; Liu et al., 2005; Bettencourt et al., 2009; Bettencourt and Kaiser, 2015). Here, two scientists are considered connected if they have co-authored one or more papers together, or connectivity between articles, in order to detect communities, fields or disciplines. Articles can be connected by means of the paper’s references list (i.e., if they are cited in the same bibliographic list; a methodology known as co-citation analysis (Small, 1973); if they share common references; a methodology known as bibliographic coupling analysis (Kessler, 1963). Finally it can be done by means of words (from the paper itself or its keywords); a methodology known as co-word analysis (Callon et al., 1983; Callon et al., 1991). This methodology assumes that a paper’s keywords constitute an adequate description of its content or, the links the paper established between problems (Ding et al., 2001). In the structural analysis of these networks, communities are detected by modularity measures 6 (Newman, 2002) and information about these communities is retrieved by means of qualifiers on the nodes. For this paper, we adopted co-word analysis. The co-word analysis is an important bibliometric approach based on co-occurrence analysis and has been widely applied to illustrate how concepts, ideas, and problems within a given scientific field interact and to explore the concept network within the relevant field (see Liu et al., 2016 and references therein). Here we consider a network formed by keywords as nodes. Two keywords are connected if they appear in the same paper, and the weight of the link between them depends on their probability of cooccurrence across the various papers. Thus constructed, the structural analysis of this network serves as a proxy for the conceptual structure of a specific discipline. It allows a time-series record of the changes that occurred in the conceptual space and it can reveal patterns and trends in a specific discipline by measuring the association weight of representative terms in relevant publications (Ding et al., 2001). 2.1 Identification of journals In order to define the scope of our investigation, a general search was made in WOS for TI=(engineering AND education AND sustainability) OR TS=(engineering AND education AND sustainability); refined by years 2001 to 2017 and Type of document: Articles, bringing 448 articles. The more contributing indexed journals were identified (Journal of Cleaner Production – JCLP; Sustainability – SUSTDE; International Journal of Sustainability in Higher Education – IJSHE; Journal of Professional Issues in Engineering Education and Practice – JPIEEP; International Journal of Engineering Education – IJEE) and further selected in terms of bringing together the research topics in pairs, as follows (selected journal in bold in Table 1). 7 Table 1. Compilation of data for the selection of the journals identified in the Web of Science database based on the joint appearance in pairs, of the research topics in the Journal Citation Reports categories. Selected journals are shown in bold. Regarding the choice between IJEE and JPIEEP, we realized that the first records contributor to our search was IJEE, what makes us likely to deepen in the journal metrics. When looking at the count as a percentage of the first ten countries contributions to each journal, JPIEEP is the only journal where more than 50% of the contributions come from USA (53%), being 44% for IJEE, 37% for IJSHE and 32% for JCLP. Since perceiving it slightly biased, we considered IJEE to be a more related journal to our area of expertise and influence that could better fit the scope of our research. 2.2 Network dataset To create the keywords network, a search has been conducted of the keywords in articles (171 items) from three indexed journals related to Sustainable Engineering (JCLP), Sustainability in Higher Education (IJSHE) and Engineering Education (IJEE). The selection of the articles has been based on the appearance of the terms engineering, education and sustainability in the topic and title fields of the articles in the journals IJSHE, JCLP and IJEE respectively, assuring that Engineering Education in Sustainability is all covered in the papers (see Table 2). 8 Table 2. Scheme for the search of articles in the Web of Science database and number of papers identified in each journal, based on the appearance of the terms in the topic and title fields of the articles. Redundant, capital or lowercase letter, and similar keywords have been filtered and reorganized into a coherent and unique list where those related to the same concept have been grouped in a unique chosen keyword, which maintains in the network the weight of the number of keywords that is assigned to it. For example, Higher education for sustainability includes Higher Education, Higher Education for Sustainability, and Higher Education for Sustainable Development. Our final set includes 300 keywords out of a total amount of 871 initially identified keywords 2.3 Network Metrics We quantified the structural features of the networks using the following network metrics (see Newman, 2010, and references therein): • Average degree and degree distribution. A key centrality measure of a node is its degree, 𝑘𝑘𝑖𝑖, representing the number of links node 𝑖𝑖 has to other nodes. In an undirected1 network with 𝑁𝑁 nodes (i.e., size of a network), the average degree is defined as: 〈𝒌𝒌〉 = 𝟏𝟏 𝑵𝑵�𝒌𝒌 𝒊𝒊 𝑵𝑵 𝒊𝒊=𝟏𝟏 = 𝟐𝟐𝑳𝑳 𝑵𝑵 (1) where 𝐿𝐿 is the total number of links. The degree distribution 𝑝𝑝𝑘𝑘 is giving the probability that a randomly selected node in the network has 𝑘𝑘 links. Since 𝑝𝑝𝑘𝑘 is a probability, it 1 Network connections can be undirected or directed. Undirected ones simply exist between two people or things, being mutual relationships. They cannot exist unless they are reciprocated. Directed ones have a clear origin and destination and may be reciprocated or not. 9 must be normalised, i.e. ∑𝑝𝑝𝑘𝑘= 1 ∞ 𝑘𝑘=1 . For a fixed network of size 𝑁𝑁, the degree distribution is the normalised histogram 𝑝𝑝𝑘𝑘=𝑁𝑁𝑘𝑘 𝑁𝑁, where 𝑁𝑁𝑘𝑘 is the number of degree 𝑘𝑘 nodes, and its cumulative degree distribution 𝑃𝑃𝑘𝑘 is giving the probability that a randomly selected node in the network has 𝑘𝑘 or more links. • Clustering coefficient. The local clustering coefficient captures the degree to which the neighbours of a given node link to each other (Watts and Strogatz, 1998). For a node 𝑖𝑖 with degree 𝑘𝑘𝑖𝑖 the local clustering coefficient is defined as: 𝑪𝑪 𝒊𝒊 =𝟐𝟐𝑳𝑳𝒊𝒊 𝒌𝒌𝒊𝒊(𝒌𝒌𝒊𝒊−𝟏𝟏) (2) where 𝐿𝐿𝑖𝑖 represents the number of links between the 𝑘𝑘𝑖𝑖 neighbours of node 𝑖𝑖. Here we use the average clustering coefficient of a network 〈𝐶𝐶〉 as: 〈𝑪𝑪〉 = 𝟏𝟏 𝑵𝑵�𝑪𝑪 𝒊𝒊 𝑵𝑵 𝒊𝒊=𝟏𝟏 (3) • Average path length and diameter. Most networks are built to ensure connectedness. In an undirected network two nodes 𝑖𝑖 and 𝑗𝑗 are connected if there is a path between them on the graph. A path is a route that runs along the links of the network, its length representing the number of links the path contains. The shortest path 𝑑𝑑𝑖𝑖𝑖𝑖 between nodes 𝑖𝑖 and 𝑗𝑗 has the fewest number of links. The average path length, denoted 〈𝑑𝑑〉, is the average shortest path between all pairs of nodes in the network. The network diameter, denoted 𝐷𝐷, is the maximal shortest path in the network. • Betweenness centrality. For every pair of vertices in a connected graph, there exists at least one shortest path between the vertices such that either the number of edges that the path passes through (for unweighted graphs) or the sum of the weights of the edges (for weighted graphs) is minimised. The betweenness centrality for each vertex is the number of these shortest paths that pass through the vertex (Barrat et al., 2004). This measure is formalized as follows: 16 Table 5 shows the current (i.e., year 2017) network metrics for journals IJSHE, JCLP, IJEE and the total network. From a global point of view, all networks show similar topological patterns, albeit some noticeable differences arise. Table 5. Current (i.e., year 2017) network metrics for the individual networks of the journals IJSHE, JCLP, IJEE and for the global network resulting of aggregating the three journals. 3.2.1 IJSHE In the most recent years of the publication period (2010 to 2018), the institutional implementation of sustainability is discussed (edges in blue), differently than in previous years, when were explored mainly topics related to the environment, the environmental adequacy of educational centers and their relation to sustainability (edges in orange). Concern for a more sustainable society and the methods/ tools/ frameworks for the transition occur in intermediate years (in purple). Interdisciplinarity appears in 2012 and ends with a degree of 17 (position 5 of 66, 7.5 above 100), while Transdisciplinarity appears in 2014 and ends with a degree of 10 (position 17 of 66, 25.7 above 100). In terms of vertex-specific metrics we find the following results for the network’ keywords: 17 a) Average degree and degree distribution. The higher degrees correspond to Sustainability in Higher Education (55), Engineering and architectural education for sustainability (41), University (34) and Education for Sustainability (29). The next is Interdisciplinarity and Education with community, both with a degree of 17, while Transdisciplinarity ends up with 10. Although Interdisciplinarity and Transdisciplinarity appear respectively in 2012 and 2014, their importance places them very high in terms of degree, indicating that they quickly connect with all the topics. The IJSHE higher connectivity is shown in terms of the graph density (i.e., 0.14), one order of magnitude over the other graphs’ densities. b) Average path length and diameter. This network has the lowest average path length (1.96) and diameter (3), a fact that characterizes its high connectivity. In this network, it is easier to move from one keyword to another following a shortest path, compared to the other networks. c) Betweenness centrality. The search process of the papers used in the Web of Science database makes some particular important keywords to be present at each stage of our search, a fact that generates a high degree for these particularly relevant keywords. For these keywords, the highest degree implies, at the same time, the highest Betweenness centrality (i.e., Sustainability in Higher Education, 889). For this centrality measure, Interdisciplinarity has 20 (11th position) and Transdisciplinarity 6,5 (i.e., 18th position) 3.2.2 JCLP The amount of publications in JCLP has increased from 2013 (orange edges), in the most recent years of the publication period (2003 to 2018) (purple edges). Transdisciplinarity appears in 2013 and ends with a degree of 14 (position 29 out of 233, 12 above 100), while Interdisciplinarity appears in 2014 and ends with a degree of 13 (position 32 out of 233, 13 above 100). In terms of vertex-specific metrics we find the following results: 18 a) Average degree and degree distribution. The higher degrees correspond to Sustainability in Higher Education (131), Higher Education for Sustainability (105), and Education for sustainability (82) and Reform of curriculum for sustainability at universities with a degree of 76. The following keywords in terms of degree are related to topics of different Categories as Institutional and policies, Curricular structure, Educational strategy and Competences/ behavioural aspects (few). Transdisciplinarity ends with a degree of 14, surrounded by Systems thinking and Higher Education Institutions, both ending with a degree of 14. Interdisciplinarity ends with a degree of 13, surrounded by Circular Economy systems, Future studies/visions and Quality assurance, all ending with the same degree. The degree distribution follows a power law (Barabási, et al., 2000), indicating that the connectivity is dominated by some few but very much connected keywords which act as hubs in the system (i.e., Sustainability in Higher Education and Higher Education for Sustainability). b) Average path length and diameter. This network has intermediate values of average path length (2.36) and diameter (4). The fact that it comprises most of the nodes of the global network implies a topology similar to this last one. c) Betweenness centrality. Due to the search process of the papers used in the Web of Science database, the star effect causes the keyword with the highest degree to be the one with the most Betweenness Centrality, being the keyword Sustainability in Higher Education with a degree of 10627,5, while the Average Degree is 159,1. Transdisciplinarity has 169,4 (27th position) and Interdisciplinarity 32,3 (59th position). 3.2.3 IJEE In this journal, the amount of publications seems to maintain or indeed decrease a bit in recent years. Apart from sustainability aspects in HEIs, aspects appear the most recent years related to students, their representativeness and learning strategies and techniques. Just before, fields of research revolve around educational frameworks relating different learning environments, included disciplines and cultures (Interand Transdisciplinarity, Systems thinking, Active learning, Real-world learning experiences, etc.). In addition unlike the other 19 journals, typical engineering topics (i.e., Mining, Petroleum, Water) often appear, even sometimes in relation to social learning (i.e., Cross-cultural education, Social Sustainability, Service Learning, etc.). a) Average degree and degree distribution. The higher degrees correspond to Sustainability in Higher Education (32) and Sustainability of campus (28), followed by a group of four keywords namely, Underrepresented students groups (15), Education for Sustainability (14), Real-world learning experiences (14), Reform of curriculum for sustainability at universities (14). Interdisciplinarity and Transdisciplinarity end up respectively with a degree of 8 (position 14 of 75, 20 above 100), and a degree of 5 (position 31 of 75, 41 above 100), showing a low connectivity level, which is related to the network’s disconnected topology. It should be noted that both appear in 2003, at the beginning of the publication period (2003-2017). b) Average path length and diameter. This network has the highest values of average path length (2.75) and diameter (6), indicating a much more disconnected topology, where it is more difficult to move from one keyword to another following the shortest path. c) Betweenness centrality. Due to the search process of the papers used in the Web of Science database, the star effect causes the keyword with the highest degree to be the one with the most Betweenness Centrality, being the keyword Sustainability in Higher Education with a degree of 1084. 4 Discussion Results shown in the previous section offer new strategies to assess and analyze the evolution of the research in this three journals, highly devoted to Engineering Education in Sustainability in the last two decades, both from an individual (i.e., journal) point of view, and a global one. The next sections explain in terms of the structural analysis of the networks, the evolution of each of the three individual journals networks (4.1) and the global one (4.2). Finally, section 4.3 features the main trends of the identified Categories of keywords, by means of clustering them from the three relevant points of view of EESD (see Section 2.3). 20 4.1 Journal networks evolution analysis 4.1.1 IJSHE This journal presents from de beginning (2010) concern on institutional strategies for sustainability learning and the relation with society. In 2012 there is a peak of publications mostly in the institutional and curricular area, bringing the keywords Complexity and Interdisciplinarity to the arena. Later on (2014) seems to appear a curricular transition towards society, when Transdisciplinarity keyword appears. In further years (2015 to 2017), the concern is growing to move to society (Social responsibility, Sustainable societies, Regional cooperation for sustainability…) by means of, both institutional strategies (Commitment and implementation of sustainability in HEI, Global partnership) and learning strategies (Service learning, Active learning, Volunteering…). Finally, case studies and industrial aspects (Industry competency model, Cleaner production, Energy engineering education for sustainability) seem to have the will of reinforcing relationships beyond the university. In terms of network shape, IJSHE presents an intermediate topology between JCLP and IJEE, with some keywords used in particular articles and others with greater connectivity in the network. 4.1.2 JCLP The JCLP presents a more connected topology, with a higher number of long distance connections between nodes which indicates a more global behavior in the sense that all keywords are much more mutually connected, with the same keywords being used in different articles. As a particular case and in order to ease the analysis of this populated web, we have clustered the big amount of keywords into the thematic Categories explained in section 2.3 (see Figure 3). In the first period predominate the institutional topics and those referring to the adequacy of the curricular structure for sustainability, although three areas appear that will be dripping constantly in the complete analyzed period, corresponding to the Categories: Td and collaborative networking (9), Topics: techno-environmental (4) and Academic/professional development (8). Keywords related to the Td and collaborative networking category (Interand 21 Transdisciplinarity, Networking, Social Impact, Sustainable societies, Holistic approach, Social responsibility, Public debate, Cross-cultural education, etc.), have been appearing with considerable betweeness centralities, showing a kind of “bridge” score, where a gap exists. Fig 3. Snapshots for the temporal evolution of the JCLP keywords’ networks in 2010 and 2017. Dark colours of nodes indicate high Betweeness centrality. The different numbers and shapes indicate the classification in Categories, namely: 1Institutional and policies, 2Curricular structure, 3Educational strategy, 4Topics: technoenvironmental, 5Topics: techno-economics and 6Topics: socio-cultural, 7Competences/ behavioural aspects, 8Academic/professional development, 9Td and collaborative networking In the second period, after 2013, the number of publications increases. Although all the Categories are represented, keywords related to Educational strategy (3) and Competences/ behavioral aspects (7) increase their number. Keywords related to Td and collaborative networking (9) continues to preponderate. 4.1.3 IJEE The different areas to which the keywords refer are uniform throughout the period of publication of this journal, from 2007 to 2017. The most of the keywords point to topics referring to institutional and curricular aspects of the introduction of sustainability in the engineering higher education. From the very beginning the keywords Interdisciplinarity and Transdisciplinarity appear but they are not very significant, although both and other keywords 22 related to new cross-disciplinary frameworks and collaborative paradigms appear later, On the other hand, relevance is given to real case studies of education for sustainability in engineering in different countries, with a North-South component. Another IJEE peculiarity is that presents the highest proportion of articles with keywords related to students representativeness, additionally with high relevance (for example the keyword Underrepresented students groups, with a Degree of 365, position 3 of 75). On the other hand, do not appear keywords referring to the training of academics (included in Category 8). Finally, many keywords refer to educational strategies (included in Category 3) have appeared in the last period of years. The linear network shape of IJEE point to a low connectivity, i.e. the necessity to travel longer topological distances to find a path between the different nodes (article keywords), thus suggesting that research is made in different areas separately, where articles use different keywords from one another. 4.2 Global network modularity analysis Modularity analysis provides information on the relationship between connections in a network. Networks with high modularity have dense connections between the nodes within modules but sparse connections between nodes in different modules (also called clusters, groups, or communities).To measure the strength of division of our network into modules, we make use of a modularity algorithm (Newman, 2002). In our case study, the network keywords are divided into nine different modules. The algorithm gathers certain keywords in a proximity grouping, so that the network is distributed in these nine automatically separated kind-of knowledge domains (i.e., modules characterized by connections between keywords that tend to appear more often than in a pure random case). By crossing this atomization in modules (one to nine), with the distribution of keywords in the thematic Categories, it is shown that some keywords are not clearly classified within a domain of knowledge. On the contrary they are used in a generalized way throughout the network. Specifically, only the keywords belonging to two of the nine Categories are present in all modules (shown in Figure 5, with 23 different shades of grey). Keywords belonging to these two thematic Categories, namely Institutional and policies and Transdisciplinary and collaborative networking, permeate the entire network, indicating that these topics have spread throughout all the areas of knowledge addressed by all journals. Fig 4Number of keywords by module (1 to 9, in grey scale) and category (x axis). Categories Institutional and policies and Transdisciplinary and collaborative approaches appear in all modularity clusters, indicating that they have spread throughout all the areas of knowledge of the network. 4.3 Research trends The evolving research trends have been explored along three periods of five years each: 2003-2007, 2008-2012 and 2013-2017, discarding 2018 to avoid bias. As explained in section 2.3, the nine thematic Categories has been analysed from three points of view present for years in the engineering higher education for sustainability, namely: categories related to different aspects of the educational system (1Institutional and policies, 2Curricular structure, 3Educational strategy, 7Competences/ behavioural aspects and 8Academic/professional development); categories that are considering the three pillars of sustainability (4Topics: techno-environmental, 5Topics: techno-economics and 6Topics: socio-cultural); and Category 9 referring to aspects related to transdisciplinar and collaborative approaches to bring together academia and society for co-creative learning schemes towards 24 sustainability. This last category has been analysed independently due to its cross-cutting effect and the relevance shown in previous studies (see Segalas, et al., 2018 and references therein; Balsiger, 2014; Muhar, et al., 2013; Rieckmann, 2012). 25 Figure 5. Results of evolution in percentage of papers published per Category in the three relevant points of view in the field of higher engineering education for sustainability: Educational system (5a, 5b); Three pillars of Sustainability (5c, 5d); Transdisciplinary and collaborative approaches (5e), during the analysed years. Figures 5a, 5c, 5e show the trend of number of publications in each period (2003-2008; 2008-2012; 2013-2017), while figures 5b and 5c show a comparative of annual publications.