Mapping leading universities in strategy research: Three decades of collaborative networks

Abstract This paper presents a longitudinal classification of the impact that universities have on strategy research from three decades of publications, between 1987 and 2016, by using bibliometric techniques and distance-based analysis of networks applied at the level of universities. Using the WoS database, this study proposes a general overview of three decades of strategic management research. Using these techniques we (i) categorize the last 30 years of academic production of research institutions in terms of strategy, evaluating their impact; (ii) analyze which universities are publishing the most in journals whose scope of publication covers strategic management; and (iii) map the network of collaboration structures among research organizations, determining its relationship and analyzing its evolution in those three decades. We found that the University of Pennsylvania was the most prominent institution throughout the years, showing the broadest network of citations according to our network analysis. There was also a remarkable presence of international universities from the UK, Canada, France and the Netherlands, however, the citation pattern among them is still low. We also observed evidence of inner knowledge flowing among different fields based on the deliberate multidisciplinary nature of research in strategy, as the strong coincidence with the ranking of the main journals in the marketing field when comparing the bibliometric studies of both fields. This analysis contributes to strategy research, first by delivering insights based on the impact of academic production and secondly through the evolution of collaborative network linkages in terms of strategy investigations undertaken to build collective knowledge.


PUBLIC INTEREST STATEMENT
The main objective of our research is to identify the main trends on strategy research as an investigation field from last 30 years of research, recognizing the most prominent universities and the evolution of collaborative network linkages to build collective knowledge. In order to present an integrated perspective, a longitudinal classification analysis was made by taking into account bibliometric indicators that are used to weigh the quality and impact of the investigations, and a graphic analysis based on visualization of similarities, which allows to detect the structure and configuration of exploration in the subject area. This kind of investigations are highly valued by people who does not know the topic in deep, or people that is new on the subject, and even for the development of university policies around research processes, allowing a quick overview of the matter by providing information based on scientific output.
inner knowledge flowing among different fields based on the deliberate multidisciplinary nature of research in strategy, as the strong coincidence with the ranking of the main journals in the marketing field when comparing the bibliometric studies of both fields. This analysis contributes to strategy research, first by delivering insights based on the impact of academic production and secondly through the evolution of collaborative network linkages in terms of strategy investigations undertaken to build collective knowledge.

Introduction
The dual character of scientific rigor and practical utility that strategic management research comprises has become a natural scenario that demands the quantity and quality of knowledge production (Nerur, Rasheed, & Pandey, 2016). All efforts to cover that demand for knowledge generate substantial volumes of scientific material and academic information each year. For instance, a basic topic search in the Web of Science (WoS) database, limiting the request to the word "strategy", can lead to more than 1.5 million academic products. Researchers from different fields have adopted bibliometric analysis to categorize all that information. Bibliometric techniques are helpful in finding trends, delimiting the existing structure of research and indicating clear potential future themes for investigation (Bar-Ilan, 2008).
When examining bibliometrics in strategy, it is evident that important studies have given greater relevance to the field. Ramos-Rodriguez and Ruiz-Navarro (2004) and Nerur, Rasheed, and Natarajan (2008), applying citation and co-citation analyses, found an increasing tendency to replace books for papers as a source of academic publication in terms of strategy and the prominence of the Resource-Based View theory of the firm as a foundation for upcoming developments. The work of Ronda-Pupo and Guerras-Martín (2010), based on a country network analysis, found that the discipline had three stages of evolution, with the current stage characterized by a stable growth of research links among countries. Based on journal co-citation analysis, Nerur et al. (2016) found a downward tendency in practitioner orientation and an increasing collaboration with fields such as finance and sociology, international business and entrepreneurship. The main assumption in these studies was that strategic management had evolved to become, of itself, a research topic; therefore, the analysis of its intellectual structure based on bibliometric techniques became a necessity. According to this requirement, these studies delineated research on strategy based on articles published in the Strategic Management Journal (SMJ), regarded as being the most relevant academic forum in the field (Hoskisson, Wan, Yiu, & Hitt, 1999).
Although SMJ represents a solid foundation for obtaining an overview of the field, the exclusion of important data misleads this representation. Essential information flowing from the increasing number of new and specialized journals in the academic network, as well as the existence of strategy and strategic management as subjects in the scope of publication in other widelyrespected journals, represents a serious bias when seeking a picture of strategy research. Therefore, the aims of this investigation-by using bibliometric techniques and distance-based analysis of networks applied at the level of universities-are (i) to present a general overview of strategic management research based on the WoS database, which gathers more than 12,000 journals; (ii) categorize the last 30 years of academic production of research institutions in terms of strategy, evaluating its impact quinquennially and globally; (iii) analyze which universities are publishing the most in journals whose scope of publication covers strategic management; and (iv) map the network of collaboration structures between research organizations, determining their relationship and analyzing the evolution across these three decades.
As noted by Daft and Lewin (2008), two kinds of "relevance" help solve particular needs, one to the practitioners' world and the other to the academic and scientific communities. This investigation is aligned with this second need, enriching the overview provided in previous works, by showing a longitudinal classification and mapping of institutions that represents research, vigilance of quality and diffusion of knowledge. This approach will be useful in research by other academics because it identifies the evolution of linkages, tendencies of research and similar profiles in investigations among institutions.
This work particularly shows the advances of science in the subject of strategy, appreciated as one of the most important and prestigious areas in the field of business due to its contribution to research. For these reasons, we conducted a bibliometric study that provided answers to the following six research questions: RQ1. What are the most productive or influential universities in strategy research?
RQ2. How have the most influential universities evolved over 5-year periods across the last 30 years?
RQ3. What are the most influential journals in strategy research, according to their H-Index?
RQ4. Which are the leading universities in strategy research that publish in the Top 20 most influential journals in this scientific field?
RQ5. How has the citation behavior of the leading universities in strategy research evolved over the 1987-2016 period?
RQ6. How, in 5-year periods, have collaborative network linkages evolved according to the bibliographic-coupling fractional-counting analysis performed on the leading universities? Therefore, the main objective of this article is to determine the contribution of the strategy area to scientific research and to show which have been the most influential universities and journals during the period between 1987 and 2016. Section 2 of this paper first presents a brief review of the literature available on the different bibliometric analyses that have covered subjects related to strategy. Section 3 describes the bibliometric methods used throughout this document, presenting the most prominent and prolific institutions and their evolution arranged per quinquennial, as well as their classification based on the most prominent journals in the field. Section 4 presents the bibliometric results extracted from the WoS Core Collection, and Section 5 provides a graphical analysis of 30 years of bibliographic production in strategy, mapping the network structures of collaboration among universities by decades. Section 6 is a brief report on the main findings and conclusions of this document.

Literature review
Bibliometric analysis can be defined as a quantitative method based on mathematics and statistics applied to published units to understand the nature and course of development of a field of research (Broadus, 1987;Pritchard, 1969). Bibliometric explorations are appreciated as instruments that analyze and organize large amounts of historical data to identify hidden patterns, and which is highly valued by researchers during their decision-making processes (Daim, Rueda, Martin, & Gerdsri, 2006). Bibliometric techniques allow the categorization of scientific production according to its impact on a field of knowledge. Different bibliometric studies have used different categories for their analysis, such as authors (Cancino, Merigó, Coronado, Dessouky, & Dessouky, 2017), journals (Valenzuela, Merigó, Johnston, Nicolás, & Jaramillo, 2017), universities (Linton, 2004), countries (Merigó, Cancino, Coronado, & Urbano, 2016), and even more specific areas with concepts such as Entrepreneurial Orientation (Andrade-Valbuena, Merigo-Lindahl, & Olavarrieta, 2018), which allow for the evaluation of the influence and relevance of academic products in the scientific community.
Bibliometric overviews have also been used in different scientific fields, determining the delineation of research domains through different bibliometric techniques (Hood & Wilson, 2001). From broad areas of research such as econometrics (Baltagi, 2007) or probability and statistics (Genest & Guay, 2002), to more specific domains of investigation such as New Product Development , or ant colony optimization (Deng & Lin, 2012), bibliometric techniques have been a valuable tool for identifying the configuration of the research subject area. Similar efforts have been used to analyze particular interests of investigation in business and management by using bibliometric analysis, such as mergers and acquisitions (Ferreira, Santos, de Almeida, & Reis, 2014), dynamic capabilities (Vogel & Güttel, 2013), competitive intelligence (Calof & Wright, 2008), and Nordic strategy research (Schriber, 2016).
As a quantitative methodology, bibliometric analysis is based on indicators that provide a representative and informative perspective of the data (Podsakoff, MacKenzie, Podsakoff, & Bachrach, 2008). Some of the most common indicators for measuring academic productivity are based on the number of publications and number of citations that a scientific work receives (Trieschmann, Dennis, Northcraft, & Niemi, 2000). Hirsch (2005) argues that the use of an arrangement of both indicators in one single measure is the optimal way of evaluating research, because it captures a wider perspective that the number of publications and the number of citations by themselves cannot achieve: the so-called H-Index. This indicator combines articles with citations, indicating the number of studies that have received at least the same number of citations (Hirsch, 2005). The H-index has been highly used and accepted as a bibliometric instrument to measure the relevance of a wide variety of research output from different actors, like universities, countries or even regions (Alonso, Cabrerizo, Herrera-Viedma, & Herrera, 2009). However, since each indicator receives a different dimension of productivity and impact of academic research, choosing between one and another can mean a misrepresentation of the field . In this sense, some complementary and mixed methodologies have emerged, as well as the inclusion of composite measurements, or the use of network graphs (Yan & Ding, 2012). The bibliometric mapping and clustering procedure is a bibliometric distance-based analysis of the Visualization of Similarities (Stands for VOS), by utilizing different measurement approaches as bibliographic coupling networks, citation networks, co-citation networks and coauthorship networks (Van Eck & Waltman, 2010). Studies of related subjects of research have been based on these methodologies, addressed a wide spectrum of topics, as resilience in business and management research (Linnenluecke, 2017), the revision of Human Resource Management (Markoulli, Lee, Byington, & Felps, 2017), the analysis of born global firms (Dzikowski, 2018), among others.
An interesting development to the study based on scientometrics is the bibliometric distancebased analysis based on the VOS-Viewer developed by Van Eck and Waltman (2010). This software allows the Visualization of Similarities, in which the distance between two items reflects the strength of the relation between them (Andrade- ). An interesting particularity of this development, is that allows the use of fractional counting, assigning a fraction of the authorship of each paper based on the number of authors of each publication. In this sense, only a fraction of the authorship of each paper is assigned, depending on the number of authors included in it. For instance, in a full counting method, the publication of one single paper by two authors ascribes one paper to each author, however, in a fractional counting method, the authorship allocation goes 50% of the paper each. Therefore, it allows to capture the knowledge structure, diffusion and influence from a more sensitive perspective (Servantie, Cabrol, Guieu, & Boissin, 2016).

Methods
Based on the bibliometric procedure proposed by Merigó, Gil-Lafuente, et al. (2015), this study uses single and composite indicators to present different results based on the same variables. Therefore, we consider that the number of citations, number of publications, the H-Index and thresholds ranking the number of documents above a number of citations, are the bibliometric measurements that best suit the objectives of this investigation (Laengle, Modak, Merigó, & Zurita, 2018;Martínez-López, Merigó, Valenzuela, & Nicolás, 2018). Furthermore, some general indicators proposed by other research institutions, such as QS Top Universities and the Academic Ranking of World Universities, are used to characterize different interests of research that allow the reader to detect strengths and opportunities (Tur-Porcar, Mas-Tur, Merigó, Roig-Tierno, & Watt, 2018).
The network mapping and clustering procedure was analyzed employing the Visualization of Similarities (VOS)-Viewer software, version 1.6.6, developed by Van Eck and Waltman (2010). This software provides a bibliometric distance-based analysis of the visualization of similarities, where the distance between two items represents the strength of the relationship between them. All the clustering analyses were based on fractional counting, which means that authorship was divided by the number of authors.
All the data in this research is available in the Web of Science (WoS) database, which is currently owned by Clarivate. This decision contributes to several objectives: first, the decision guarantees that all included material has received a critical peer review; second, it provides an ample inclusion of articles and journals; and finally, this database gathers journals that are forums of publications of different specialized fields, such as strategy (Nerur et al., 2016), thus enhancing the reliability and pertinence of the results.
Our search was based on the WoS Core Collection database for the period 1987-2016 and includes the following indexes: Social Sciences Citation Index (SSCI); Science Citation Index Expanded (SCI-EXPANDED); Emerging Sources Citation Index (ESCI); Conference Proceedings Citation Index-Social Sciences and Humanities (CPCI-SSH); Conference Proceedings Citation Index-Science (CPCI-S) and Book Citation Index-Social Sciences and Humanities (BKCI-SSH).
We integrated the results from well-known specialized journals and selected keywords validated by three different experts to ensure that no relevant work was left aside. The level of expertise was taken into account, including a minimum of 7 years of researching and teaching field-related courses. The results included the following journals: Advances in Strategic Management: A basic topic search complemented the journal results. This kind of search looks for records, which include specific words in the title and/or abstract, the author and suggested keywords within each document. We used specific keywords suggested by the experts: "strateg*" OR "SWOT" OR "competitive advantage*", which automatically include plural or singular variations of each word. These keywords were chosen by the experts, bearing in mind their inclusion of most of the papers related to strategy, without incurring an insertion of articles that were unrelated to the field.
All results were filtered under the WoS categories Business + Management to guide research as purely as possible to the specific production of scientific knowledge in the field. Following the procedure of Merigó, Mas-Tur, et al. (2015), only the categories Article + Review + Letter + Note were included in the study. Accordingly, 58,440 results were included in the data analysis, with 1,922,158 citations, averaging approximately 33 citations per paper per year.

Results
The results of the study based on bibliometric techniques focused on three main topics: (i) Leading Universities in Strategy Research, (ii) Leading Universities in Strategy Research by Quinquennials, and (ii) Journal Analysis of the Leading Universities. Each topic is reviewed as follows:

Leading universities in the field of strategy
The ranking of universities, as shown in Table 1, considers the H-index of each institution based solely on the field of strategy as the first classification item; the second classification item is the number of citations received, considering only articles published on strategy; and the third classification item is the total number of papers published in strategy research ("TPS"). Three citation thresholds are offered for analyzing the publications and citations in the field: 250, 100 and 50, which are shown in order to identify the most productive institutions and the citation structure during the 1987-2016 period. Additionally, Table 1 presents the results of comparison with two other well-known international rankings of universities: QS Business and Management World University Ranking of 2017 and the Academic Ranking of World Universities in Economics/ Business of 2015 ("ARWU"), also known as the Shanghai Ranking.
As may be seen in Table 1, the most influential institution during 30 years (1987-2016) of scientific publications in the strategy area was the University of Pennsylvania. This finding was confirmed first by the H-index, which highlighted that 133 articles had at least 133 citations each, with the total number of citations ("TCS": 72,882) that strategy papers received, as well as by the It is important to note that the Institut Européen d'Administration des Affaires (acronym INSEAD), from France, provided a very important contribution to strategy research, standing out as the only institution outside the USA ranked among the Top 10 universities.
In regard to the number of papers published in the Top 200, universities in the USA garnered important proportions of production (53.76%), followed by the UK (13.25%), Canada (5.56%), Australia (5.27%), the Netherlands (4.95%) and China (3.28%). It is important to emphasize the contribution that Erasmus University Rotterdam, from the Netherlands, made, since it was as productive as those universities ranked in the Top 10, while not necessarily having the highest H-index.
Georgetown University, Stanford University, the University of Chicago and the University of Pennsylvania achieved more than 100 citations per paper published in strategy research, jointly accumulating almost 7% of all the citations made in this field, according to the Top 200 group. In fact, the first 41 universities accrued almost half of the citations made in strategic management research over three decades of publications, which explains their high impact. The London Business School also presented similar results for the number of citations per paper published in strategy research to those at the Top 10 universities.
The average global H-index (Hirsch, 2005) of the Top 200 universities was 50. This finding means that at least 50 papers received 50 citations from each paper, on average. However, the Top 50 Table 1 universities as shown in Table 1 received an H-Index of 78. This result was expected due to the order of classification.
It is important to note that even when the Top 50 from Table 1 are compared to the Top 200 presented in Appendix A, important names and proportions from the most prominent universities continue to appear.
The last two columns show the ARWU and the QS Business and Management World University Rankings of 2017. Dissimilarities were noted, even though most of the Top 10 universities, based on the H-Index, were well positioned in these rankings. This finding is because the dimensions that each methodology captured in its measurement were also different. For instance, the prestige, size, or age of each institution were important variables for a student when searching where to study; therefore, the QS and ARWU are important indexes that capture these dimensions. However, if we wish to measure their scientific impact, these dimensions might appear as noise in the measurements. Despite the above, most of the universities listed in Table 1 were relatively well classified in those rankings.

Leading universities in strategy research by quinquennials
This section focuses on the evolution of research in strategy performed by the leading universities throughout the 1987-2016 period. This analysis is based on periods of 5 years, ranking the top 20 most prominent institutions according to their H-Index. Other indicators, such as the total number of papers, total number of citations received, and the number-of-citations/number-of-papers ratio during that period, were also shown.
The preponderance of USA institutions within the period was remarkable; however, it was noticeable that from 2000 on, the presence of universities from other parts of the world gained importance in the field, to the point that in the last quinquennial they represented almost 50% of the top 20 universities in terms of strategy research. Universities from Europe deserve special mention, where we found that the Netherlands, Finland and the UK had the same number of institutions in the ranking. This growth in the number of papers may be driven by the increasing number of journals specializing in strategy, particularly those that have emerged in Europe; this finding warrants further investigation in future studies.
In regard to productivity, the number of papers that the Top 20 institutions published in each quinquennial grew by approximately 30%; this result is clearly indicative of the relevance that strategy research has gained in the scientific arena.
Six universities figured in all the quinquennial rankings: the University of Pennsylvania, Harvard University, Indiana University and the University of Minnesota. It is remarkable that the presence of INSEAD from France, only in the last quinquennial, did not rank it in the Top 20. Table 2 shows the most influential universities in strategy research by quinquennial, classified by their H-Index, as the first classification item. The second classification criterion is the total number of citations, and the third is the number of papers published in strategy research.
The University of Pennsylvania was the most influential during the 1987-1991 period. Its production in number of papers over the three decades was also noticeable, increasing by almost 150% between 1987 and 2016. The number of citations per paper that the University of Stanford received was also remarkable and accounts for almost 14% of all the citations obtained in this period. This finding highlights the relevance to the field of this quinquennial of academic production by such institutions.
It is important to note the academic production results for the period between1992 and 1996 and the number of citations that are recorded for the University of Pennsylvania. In fact, five universities  Considering the number of citations per paper between 1997 and 2001 (on average, 134% of the citations received per paper in this period), the relevance of this quinquennial in the construction of knowledge and learning in the field cannot be understated. This kind of citation structure is also evident in different subjects, such as biology and physics, where what are known as "leaders of the field" are identified, thus setting the foundations for the formation of future knowledge. In this period, the Top 3 most productive institutions also had the most impact on the field measured in the number of citations, led by the University of Texas at Austin and followed by Harvard University and the University of Pennsylvania.
Academic production in strategy research during the 2002-2006 period highlights the role that institutions, in a global context, begin to have in the academic discussion of strategy. A total of 23% of the Top 20 institutions are from outside the USA. Note that the Erasmus University Rotterdam increased its production of papers on strategy through the three following quinquennials, by almost 132% between 2002 and 2016, receiving an average of 33.6 in the H-Index. Table 2 indicates that between 2007 and 2011 the number of publications of universities outside the USA was approximately 30% of the academic production; however, the number of citations received was still low (approximately 3% of all citations received during the period). Traditional names from the following institutions in the UK were relevant to the analysis: Manchester University, Nottingham University, Cambridge University and the London Business School.
In the last period of strategy research, from 2012 to 2016, half of the institutions that participated in the Top 20 were outside the USA, led by the UK and the Netherlands. Copenhagen Business School, Erasmus University Rotterdam and Aalto University produced almost 15% of the papers from the Top 20 universities, which suggests a remarkable contribution to the research of the field. The presence of the Hong Kong Polytechnic University was also notable, as it produced almost 4% of the papers and received a similar percentage of citations during this period, highlighting the relevance that China will achieve in the academic conversation.

Journal analysis of the leading universities
To analyze the relationship between research institutions and publishing institutions in strategy research, this section first applies bibliometric analysis to find the Top 20 most prominent journals according to their H-Index and integrates these results with those obtained in Table 1. This approach will allow for incorporating the representation of scientific vigilance of rigorousness to demarcate the academic space, evidencing a potential flow of knowledge among the different fields of knowledge. Table 3 presents a ranking according to the H-Index, based on material published in strategy research. Some influential names that are specialized in the field were expected, such as the Strategic Management Journal and the Academy of Management Journal. The relevance that other journals from other specialties have on strategy research is noteworthy, as among them the Journal of Marketing, Research Policy and Journal of Operations Management. This outcome was foreseeable based on the deliberative and multidisciplinary nature of research in strategy. In general, the results from Table 3 show eight fields of research and interests of publication: Strategy, Management, Innovation, Operations, Human Resources, Marketing and Organizational Behavior, evidencing strong collaborative ties and flow of knowledge among these disciplines. There was a notable presence of publishers from the USA, constituting almost 60% of the institutions from this country. The second most represented country in the Top 20 journals was the Netherlands (25%), followed by the UK (10%).

Cross-analysis of journals and institutions
To analyze the relationship between universities and journals more deeply, we looked at the Top 20 journals from Table 3 to see which universities published the most in each one. Tables 4 and 5 present the Top 20 leading universities that publish in the Top 20 most influential journals regarding strategy research.
Note that the institutions that ranked as leading universities in Table 1 are also those that published the most in these journals. The University of Texas at Austin leads in 11 of the Top 20 journals; Harvard University, the University of North Carolina and the University of Pennsylvania lead in 10; and Indiana University, the INSEAD Business School, Michigan State University and Pennsylvania State University are among the leaders in 9 of the Top journals in strategy.
Of the institutions outside the USA, INSEAD was the one that was the most present in most of the Top 20 Journals (9 Journals), followed by Erasmus University Rotterdam (8 Journals, the Netherlands), and London University (5 Journals, United Kingdom).
Some institutions that were not ranked in Table 1 of the university ranking are represented in Tables 4 and 5, such has the University of Mannheim, which had a significant number of papers published in subjects related to strategy in the Journal of Marketing.
There is evident leadership of the University of Pennsylvania. The number of papers published in strategy from this institution was outstanding. The number of papers published by INSEAD in both journals was also remarkable.
The Journal of International Business Studies ("JIBS") is more diverse, and many universities, not just from the UK, USA or the Netherlands, regularly publish their research there. This outcome was expected due to the nature of the specialization in IB. The role the Chinese University of Hong Kong plays in strategy research is noticeable, and it is the institution that is most published in the JIBS. Moreover, the presence of Chinese institutions was remarkable in these journals, evidencing the upward trend of research in strategy in that country.
The multiplicity of international institutions that regularly publish in the Journal of Management Studies ("JMS") related to strategy was remarkable, with 50% of the institutions coming from outside the USA, demonstrating the inclusive spirit and the openness to different methodological approaches that underpin this journal. Furthermore, the presence only of institutions from the USA in the JMS is evidence that this field is one of the strong areas of research in this country.

Network analysis of the leading universities based on VOS viewer
In order to map the leading publication and citation links from a general perspective, this work used VOS viewer software (Van Eck & Waltman, 2010) based on the fractional counting of bibliographic coupling, co-authorship and citation analysis. This software allows for a distancebased bibliometric analysis of the Visualization of Similarities (VOS). This approach means that the distance between two items reflects the strength of the relation between them, since the shortest distance represents the strongest relationship and vice versa. Moreover, the size of the vertices in the network represent the most prominent institutions in citations, co-authorship or bibliographiccoupled variables, depending on the type of analysis that is performed. Figure 1 presents a bibliographic data network based on citation analysis with a threshold of one hundred citations and one hundred of the most representative citing connections. The University of Pennsylvania was the most cited institution (20,244 citation links) and had the broadest network. Texas A&M University (citation links: 15,983), Harvard University (citation links: 15,368) and Michigan State University (citation links: 14,477) were also highly cited. The strongest relationships were evident at Harvard University, the University of Pennsylvania and Stanford University, which belong to the same blue cluster. The dispersion of the variables showed a stronger relationship among the universities located in the USA, which was not the case for the universities in the UK, Canada and China. This finding is based on the relatively recent weighting that these institutions are giving to strategy research, which is congruent with the results from Table 2. Figure 2 shows a bibliographic data map based on a co-authorship and fractional counting analysis, with a threshold of one hundred co-authorships and one hundred co-authorship links. The co-authorship map revealed the most important collaborative partners among the institutions. Five general clusters were evident in the network in accordance with the geographic positioning, shown in different colors. From the USA, there were three noticeable universities that were essential collaborators in the field: Michigan State University (co-authorship links: 279), the University of Pennsylvania (co-authorship links: 274), and the University of North Carolina (coauthorship links: 257), each conforming a different center and periphery cluster of scientific collaboration. Another cluster was formed by universities from the Netherlands, with the center at the Erasmus University of Rotterdam. From the UK, the University of Warwick played a principal role in collaboration in terms of strategy research. Finally, an Asian cluster was led by the Hong Kong Polytechnic University, which was the institution with the highest number of publications authored jointly in this cluster; however, the distance between vertices revealed dissimilar preferences in research.
Based on bibliographic coupling, Figure 3 shows a bibliographic map, with a threshold of three hundred institutions and three hundred bibliographic coupling relations. Bibliographic coupling helps find related research in different universities. This approach shows the intensity of relationships, given by the number of documents cited in common by two different papers. Affinity was inferred among bibliographic sources, revealing a similarity of thought. The dispersion among institutions in the map reveals the wide spectrum of investigation preferences based on strategy research. This network showed similar results to the co-authorship analysis; clusters were related   Note that most of the universities were in the USA, which is in line with the findings in Tables 1 and 2. Figure 5 shows the second decade of research in strategy, with a threshold of one hundred institutions and one hundred bibliographic-coupling relations. Institutions with the highest link strength were the University of Texas at Austin (12,584.19),Michigan State University (9,203.25), the University of Pennsylvania (8,349.16) and the University of North Carolina (8,191.86). Compared to the previous decade, Figure 5 shows a clearer structure among institutions, forming a comprehensive configuration of research, with defined central and peripheral roles and more demarcated preferences in investigation. Some institutions outside the USA are now connected to the different clusters, such as INSEAD, London Business School, University of Western Ontario and   versities (1987-1996), first decade. Size variation network visualization. Figure 5. Bibliographic data map, based on bibliographic coupling fractional counting analysis from leading universities (1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006), second decade . Size variation network visualization.
National University of Singapore. Note the emerging cluster from the UK, with its particular patterns of research in strategy; however, strong links of collaboration are not yet obvious. Figure 6 shows the last decade in research, with a threshold of one hundred institutions and one hundred bibliographic-coupling relations. Michigan State University exhibited the highest strength in bibliographic-coupled links (23,483.2), followed by the Erasmus University Rotterdam (21,718.75), Indiana University (20,856.25), Texas A&M University (20,495.02) and the University of Illinois (18,788.37). Note that the bibliographic links were less condensed in the USA, and stronger links among English universities have become clearer. Several international institutions are now central actors in some clusters, such as Warwick University (UK), Copenhagen Business School (Denmark), Tilburg University (the Netherlands), Aalto University (Finland) and the City University of Hong Kong (China). This result reinforces previous findings and tendencies seen in the bibliometric analysis in Table 2.

Discussion
The aim of this research is to achieve a longitudinal classification of the impact that universities have on strategy research based on three decades of publications, between 1987 and 2016. The more specific objectives of this investigation, using bibliometric techniques and distance-based analysis of networks, applied at the level of universities, are to (i) present a general overview of strategic management research based on the WoS database; (ii) categorize the last 30 years of academic production of research institutions in terms of strategy, evaluating its impact quinquennially and globally; (iii) analyze which universities are publishing the most in journals whose scope of publication covers strategic management; and (iv) map the network of collaboration structures between research organizations, determining their relationship and analyzing the evolution over those three decades. This study finds that the most productive and influential university is the University of Pennsylvania (H = 133), which had the broadest network of citations according to the network analysis (Table 1). There was also a remarkable presence of international universities from the UK, Canada, France and the Netherlands. However, the citation pattern among them is low. We also find that the most influential journals in strategy research are the Strategic Management Journal (H = 241), Academy of Management Journal (H = 182), and the Academy of Management Review (H = 154). An interest finding is that the main universities that publish in them are also between the most influential institutions: University of Pennsylvania, Arizona State University, University of Washington. Another interesting finding is that some universities in the UK, France, Denmark, China, Finland and the Netherlands are new relevant actors based on publications in European journals that specialize in strategy research, which highlights the importance of incorporating a wider spectrum of journals to define the academic domain of strategy.
Our longitudinal analysis demonstrated an evolution towards concentrating scattered subjects of research, which confirms the construction of common knowledge as shown in different subjects, such as biology, computing or physics. This outcome can be compared with the stages of evolution that the study of Ronda-Pupo and Guerras-Martín (2010), based on a country level analysis. The expansion/transformation stage proposed in their work almost a decade ago is more evident when the analysis is made based on institutions and journals, which allows a deeper comprehension of advancement in the field.
We also find an evolution of collaborative partners among institutions. If the perspective was focused on geographic relations, four clusters were evident in the networks in most cases: two from the USA, one from Europe (including the UK), and one from Southeast Asian countries. This seems to suggest that today's boundless knowledge flow and the globalization of management education have not weakened the location-specific interest in identifying and studying locally important phenomena, contributing positively to the development of the different currents in strategy research. In this sense, three noticeable universities were found as central collaborators in the field from the USA: Michigan State University, the University of Pennsylvania, and the University of North Carolina, each conforming a different center and periphery cluster of scientific collaboration. The second cluster was led by universities from the Netherlands, with a center in the Erasmus University of Rotterdam, INSEAD from France and the University of Warwick from the UK. Finally, the Southeast Asian cluster was led by the Hong Kong Polytechnic University, which is the institution that had the highest number of publications authored jointly in this cluster. However, the distance between vertices revealed dissimilar preferences in research.
We find under a different perspective, more evidence from the flow of knowledge with other fields, as found in the study of Nerur et al. (2016). Based on institutions and their classifications in the Top 10 most relevant journals in strategy research, as well as the growing collaboration among universities with strengths in different disciplines, we observed that the stream of information among different fields that cover strategy tended to grow. This finding is beneficial to the field because it uses new approaches and perspectives to solve similar problems. Additionally, we find evidence of the inner knowledge that flows among different fields based on the deliberate multidisciplinary nature of research in strategy. For instance, there was a strong coincidence with the ranking of the main journals in the marketing field when comparing bibliometric studies in this field. In Martínez-López et al. (2018), it was observed that some journals relevant to marketing research matched with those among the most influential in strategy research, as shown in Table 3. Among these journals are the Journal of Marketing, Journal of Business Research, Journal of Marketing Research, and the Journal of Product Innovation Management. When also comparing the results of leading universities in strategy publishing in the Top 20 most influential journals (Table 5) with the results of Valenzuela et al. (2017), we saw commonalities between both areas, given that both studies ranked influential journals such as the Harvard Business Review, Journal of Marketing and Journal of Marketing Research. These similarities demonstrate the strong connection between both areas. Some limitations in our study arise from the very characteristics of the methodology addressed here. For instance, when calculating the H-index and other indicators, the assignment of authorship considers a full counting that gives one unit to each co-author instead of fractional authorship. Despite fractional counting consideration in the network analysis, there is an evident lack of methodological procedures that would allow its introduction in one index following bibliometric processes. Another restriction in this paper is the exclusion of some variables that could have led us to a better description of the field, such as authors, or data gathered from another source of information accepted by scholars, such as specialized associations or conferences. Finally, despite our efforts to cover the largest and most representative sample of the articles, there is a possibility that some were not captured by our filters. Therefore, future research should include different variables, such as authors, and a wider variety of journals to identify tendencies in research and different perspectives and relationships that have contributed to the construction of the structures of collaboration and knowledge in strategic management research.
This article offers several contributions. First, this work allows for the inclusion of institutions in the academic discussion, thus enriching the overview provided in previous works. Second, the addition of prominent journals to the overview allows for the incorporation of the representation of rigorous scientific vigilance to demarcate the scholarly space. Third, mapping the research network of cooperation is valuable in identifying connections and similar profiles in research among institutions. Finally, the longitudinal network representation throughout 30 years of academic production is advantageous for understanding the dynamic that peripheral and central actors have played in the creation of the present strategic management network of knowledge and collaboration.

Conclusions
The aim of this research is to achieve a longitudinal classification of the impact that universities have on strategy research based on three decades of publications, between 1987 and 2016. This study finds that the most productive and influential universities have the broadest network of citations, where the presence of international universities from USA, the UK, Canada, France and the Netherlands are leading institutions, publishing in different journals across the globe. This highlights the importance of incorporating a wider spectrum of journals to define the academic domain of strategy.
Our longitudinal analysis demonstrated an evolution towards concentrating scattered subjects of research, which allows a deeper comprehension of advancement in the field. We also find an evolution of collaborative partners among institutions based on geographic relations, which show location-specific interests in identifying and studying locally important phenomena. We also find evidence from the flow of knowledge with other fields, as marketing and logistics.
As noted by Daft and Lewin (2008), two kinds of "relevance" help solve particular needs, one to the practitioners' world and the other to the academic and scientific communities. This investigation is aligned with this second need, enriching the overview provided in previous works, by showing a longitudinal classification and mapping of institutions that represents research, vigilance of quality and diffusion of knowledge. This approach will be useful in research by other academics because it identifies the evolution of linkages, tendencies of research and similar profiles in investigations among institutions. † Present address: School of Information, Systems, and Modelling, Faculty of Engineering and Information Technology, University of Technology, Sydney, Australia.