The data were collected from the ISI Web of Science database. All business and management journals available were considered. The keyword business school* was used in the Topic field of the database to select the articles. The asterisk (*) was used so that any variation of the wording might be considered. The search outcome was examined to check if all articles in the sample were important to the analyses. The final sample considered the whole database search and contained 493 articles published in 19 journals (Table 1).

Table 1.

Periódico	Fator de Impacto 2017	Total de artigos na amostra	% amostra
Harvard Business Review	4.374	178	36
Academy of Management Learning and Education	2.866	146	29
British Journal of Management	3.059	34	6
Organization Science	3.027	24	4
Journal of Management Studies	5.329	23	4
Academy of Management Journal	6.700	14	2
Academy of Management Review	8.855	14	2
California Management Review	3.302	14	2
Organization	2.701	13	2
Long Range Planning	3.221	10	2
Administration Science Quarterly	5.878	6	1
Journal of Management	8.080	4	0
Academy of Management Annals	9.281	3	0
Academy of Management Perspectives	4.686	3	0
Asia Pacific Journal of Management	2.474	2	0
Management Science	3.544	2	0
Business Strategy and the Environment	5.355	1	0
Management and Organizational Review	1.462	1	0
Strategic Management Journal	5.482	1	0
		493	100

Source: Elaborated by the authors.

To reduce bias and ensure the validity and reliability of the selected studies, we set criteria for selecting the final sample (Vogel & Güttel, 2013). The works had to have B-Schools as central theme. The authors read the title, the abstract, and the introduction of each document to validate the sample independently. Some documents were read in full to clarify occasional doubts. We obtained a Cohen-Kappa coefficient of .93 inter-reviewer agreement.

Bibliometrics helps control biases brought by subjectivity which is usual in qualitative reviews performed without the help of empirical tools (Ramos-RodrigueZ & Ruiz-Navarro, 2004). Moreover, through statistical analyses, large amounts of bibliographic data can be considered, bringing a broader scope than in purely qualitative reviews. (Vogel & Güttel, 2013; Zupic & Čater, 2015). Bibliometrics may be used as a prior step to guide the analysis during the systematic review. In this study, we used two bibliometric techniques - co-citation and bibliographic pairing.

The co-citation enables the measurement of the frequency in which a pair of articles is cited together (McCain, 1990). It is useful for detecting paradigmatic changes and schools of thought (Zupic & Čater, 2015).

Bibliometric coupling is a measure of similarity based on the frequency that two documents of the sample share at least one common reference, i.e., the documents of a sample are grouped in accordance with the overlapping of its bibliographies (Vogel & Güttel, 2013; Zupic & Čater, 2015). The higher the number of shared references by two documents in the sample, the greater the similarity between them (Vogel & Güttel, 2013).

Co-citation analysis was used first. With the aid of Bibexcel, a co-occurrence matrix (Bernard & Ryan, 2010) of the references of each pair of articles in the sample was generated. The co-citation analysis was performed using the references of the sample of articles used for the bibliographic coupling. To run the bibliographic coupling, we reduced our sample and only articles with at least 6 couplings (ties ≥ 6), with at least one document (node ≥ 1) were preserved. The resulting reduced sample was composed of 76 articles.

This procedure enabled us to verify the intellectual structures that specifically influence the actual mainstream research agenda represented by these articles. For the bibliographic coupling analysis, using the same sample of 76 articles, an exploratory factor analysis, aiming to group the articles by their similarity, was performed. We performed a citation analysis of the references used in the sample (File link: https://drive.google.com/file/d/16-KfNKpXdbhhTxRHdrV2pvBRjoSY4x0s/view?usp=sharing).

For the bibliographic coupling analysis, the co-occurrence matrix generated was used and it was complemented by factor analysis and the network diagram. The former employed Varimax rotation (Lin & Cheng, 2010). Factor loads of approximately 0.4 or higher were considered (Guerrazzi et al. 2015).

All the articles that compose each factor, for both bibliographic coupling and co-citation analyses, were read to enable the understanding of their content. We used two clustering methods to increase robustness (Zupic & Čater, 2015). Network analysis allows the visualization of the intellectual structure of a field through a network diagram, with nodes representing publications and loops representing their sharing relationship. Therefore, the factors were used and overlapped in the network diagram provided by Ucinet software (Lin & Cheng, 2010). The network shows the proximity between articles and their connections, representing the relationship and dimension of the number of articles all together.

We also reported density, cohesion, and centrality, which are usual metrics for network analysis (Vogel & Güttel, 2013). Density represents the extent to which a factor has common conceptual bases, and its maximum value is obtained when all possible links between references of the same factor are exhausted (Vogel & Güttel, 2013). It is an internal indicator of the factor. In co-citation, the maximum density indicates that references grouped in the same factor have independence in the expression of an approach, as they are referenced together. In pairing, it indicates that items of a factor, having their focus defined by grouping, employ common references.

Cohesion equals the density of a factor in its interconnection with other groups, indicating the degree of independence that the factor represents (Wasserman & Faust, 1994). It is an inter-factor indicator. For both citation and bibliographic matching, cohesion complements the explanation of density with respect to the power of convergence between items of the same factor. Network centrality means groupings formed around the most cited articles. It is measured by the number of links each item establishes on the network - the more links the more central the item., covering the whole network. The centrality highlights the relevance of this work throughout the network as it is more often cited in conjunction with another (co-citing) or shares references in large numbers with other articles (pairing).

Focusing on an investigation of the influence of the literature structure on the ongoing works about B-schools, we conducted a jointly analysis with co-citation and coupling results. To achieve this aim, a co-occurrence matrix was created for the references that constitute the co-citation factors and for the articles from the coupled analysis. In this matrix, co-citation references were the lines and the bibliographic coupling articles were the columns, both grouped into the previously obtained factors. The crossing between lines and columns was binarized, and we filled it with “1” if such a reference was used in any of the articles and “0” if not. This matrix made it possible to gauge the influences of the intellectual roots (co-citation) in current research (bibliographic coupling) and present the values and relationships graphically (Figure 1).

Figure 1. Source: Elaborated by the authors.

To verify the influences of the references (co-citation) in the articles (bibliographic coupling) we made a cross-analyses of references in the articles. Since they constitute articles and references, and these structure the approach of an article, this analysis makes it possible to gauge the direct influence between the basic approaches with the streams of research used to investigate our theme.

Factor CC1 was labeled “Theory and Practice Gap”. The central works of the network are present in this factor: Starkey & Madan (2001); Van de Ven & Johnson (2006); Van de Ven (2007). These articles present arguments regarding the lack of relevance of B-School research to practice.

The authors in this factor argue that knowledge is increasingly produced and distributed socially (Gibbons et al., 1994) and hence, by its nature, management research should be applied (Tranfield & Starkey1998). The main focus is on the need to rethink the science (Nowotny et al., 2001) and develop the link between theory and practice (Hambrick, 1994). This creates a discussion regarding the Kuhnian model of normal science (Kuhn, 1962) and problem-solution research, which is more socially distributed (Gibbons et al., 1994; Tranfield & Starkey, 1998; Anderson, Herriot & Hodgkinson, 2001). Shapiro, Kirkman and Courtney (2007) add that the gap should be interpreted as the result of two distinct translation problems: “lost before translation”, where the knowledge produced does not have any chance to make an impact on practice and, a “lost in translation” problem - where knowledge production is not adequately translated into practice.

Nevertheless, the creation of knowledge is a complex process (Grey, 2001), and has to be considered in different paradigms (Pfeffer, 1993), with a thought for scientific rigor in the link between theory and practice (Van de Ven, 2007). However, the weak union of theory and practice is brought into question, with several studies deemed as being of little relevance to managers (Van de Ven & Johnson, 2006). The principles from research evidence should be translated into practice to solve organizational problems (Rousseau, 2006), and to help train managers to deal with real challenges throughout their careers (Leavitt, 1989). In B-Schools, managers should collaborate with their knowledge (Starkey & Madan, 2001).

Factor CC2, is named “Social Relevance”. The articles reinforce the relevance and social role of B-Schools, and the criticisms of the impact of rankings and accreditations on them (Navarro2008; Gioia and Corley2002; Adler & Harzing, 2009). Concerns over B-Schools’ relevance and social role are not new (Cheit, 1985; Gordon & Howell, 1958), and the need for change is emphasized by the authors in this factor who argue for trends to be considered in relation to both supply and demand in the education of managers (Friga, Bettis & Sullivan, 2003). There is a call for B-Schools to consider a global perspective to have social relevance, as well as thethe development of leadership and soft skills, and greater integration of disciplines to improve decision-making skills (Datar et al., 2010; Bennis & O’Toole, 2005). It is claimed that B-Schools should be more relevant to practitioners (Porter & McKibbin, 1988).

The fact that certain theories are treated as actual practices, irrespective of their social impact, has been questioned (Ferraro, Pfeffer & Sutton, 2005; Ghoshal, 2005). The ethical scandals provoked by managers of organizations show that B-Schools need to start considering ethical aspects and go beyond focusing on financial results (Ghoshal, 2005; Bennis & O’Toole, 2005). Changing the curriculum at B-Schools should mean multidisciplinary integration (Navarro, 2008; Bennis & O’Toole, 2005; Porter & McKibbin, 1988) and the development of B-Schools as forums for citizens and places for social networking (Starkey & Tiratsoo, 2007).

We labeled Factor CC3 as “Curricula and Practice”. B-Schools are criticized because of the pressure to achieve financial results (Khurana, 2007), which is fostered by the student’s belief that what is taught at B-Schools will aid an individual’s personal and professional growth (Pfeffer & Fong, 2004). The relationship with practice is necessary (Mintzberg & Gosling, 2002; Rynes, Bartunek & Daft, 2001). They advocate the need to create knowledge for management (Starkey, Hatchuel & Tempest, 2004). Curricula should address real problems (Pfeffer & Fong, 2004; Starkey & Tempest, 2009), and focus on managers’ competence development (Rubin & Dierdorff, 2009). B-Schools should consider experience and meaning in the learning process, which transcends the classroom (Mintzberg & Gosling, 2002), and implies changes for how managers are taught (Mintzberg, 2004) e.g., through evidence-based teaching (Pfeffer & Sutton, 2007) that enables “managerial decisions and organizational practices informed by the best available scientific evidence” (Rousseau & McCarthy, 2007, p. 84). B-Schools should consider the balance between acquiring knowledge through research and exploiting knowledge through instruction (Trieschmann et al., 2000).

Factor BC1 is labeled as “B-School Relevance”. The challenge posed by the authors in this factor is to prepare people with the necessary competences for the market and society (Walsh, 2011). B-Schools emphasize theory at the expense of relevance to practice (McGrath, 2007). Some criticisms argue that B-Schools cannot fulfil this important role to prepare leaders (Benjamin & O’Reilly, 2011), and so make no difference to their careers (Pfeffer & Fong, 2004). Others stress that the leadership developed in B-Schools reflects negative values and actions without concern for society, and that there is a need for gender diversity for more humanized executives (McTiernan & Flynn, 2011). Business schools are the potential workspaces, not only to improve knowledge, but also to influence managers’ identity for such humanized leadership (Petriglieri & Petriglieri, 2010; Petriglieri, Wood, & Petriglieri, 2011).

To overcome these criticisms, several models were proposed (Ferlie, McGivern, & Moraes, 2010; Thomas & Wilson, 2011): the “Agora” as an open platform for discussion and development (see Starkey & Tiratsoo, 2007); the schools in the professional model (see Bennis & O’Toole, 2005); and the business school public interest model, which operates to the interest of society (Ferlie, McGivern & De Moraes, 2010).

Some authors have suggested that B-Schools need to revise their purposes, curricula, and pedagogy, despite the institutionalized organizational form and search for legitimacy that inhibit change (see Wilson & McKiernan, 2011). The challenge of relevance is a question of behavioral competencies (Rubin & Dierdorff, 2011), which are valued by organizations (Safon, 2007). This means that the nature of knowledge produced and taught in B-Schools should be reviewed (Chia & Holt, 2008), and scholarly impact should be evaluated not only through academic publications, but also through its impact on students (Morgerson & Nahrgang, 2008).

The authors make several suggestions, such as the inclusion of disciplines like humanities (Bennis & O’Toole, 2005; Schoemaker, 2008), requiring a distinct teaching curriculum and pedagogical approach (Harrison, Leitch, & Chia, 2007). Vaara and Fay (2011; 2012), for example, presented a Bourdieusian perspective for management education. The development of leadership skills (Kork, 2011) should include business ethics (Rutherford et al., 2012), decision-making (Henisz, 2011), and sustainability (Slater, Dixon & Fowler, 2010), as well as how to perform in emerging markets (Dhanaraj & Khanna, 2011), supported by Evidence-based Management (Charlier, Brown, & Rynes, 2011) in a B-School’s curriculum.

However, some authors counter the criticisms of B-Schools’ relevance, for example, as posed by Pfeffer & Fong (2004). O’Brien et al. argue that the: “research conducted at business schools is relevant and valuable to practitioners as evidenced by the considerable long-term economic value added to MBA student salaries.” (O’Brien et al. 2010, p.648)

Further, Slater, Dixon & Fowler (2010) argue that executives who graduated from B-Schools present positive results in regards to sustainability. Arbaugh (2010) advocates that the article of Peng and Dess (2010) “not only challenges the notion that management scholarship is deficient but argues that in fact the systems and assessment mechanisms presently in place to evaluate the worth of our publications serve a positive role in motivating us to produce better work than what we might have otherwise.” (Arbaugh, 2010, p. 280).

Factor BC3 was named “Relevance to practice”. The discussion of relevance is important in the US and UK (Butler, Delaney, & Spoelstra, 2015), but in Factor 3, there is a different focus from Factor BC1. It seems to be a UK research focus that discusses the academy-practitioner divide and evaluates possible remedies. Some authors claim that research is not relevant to practice (McKelvey, 2006). Bell, den Ouden and Ziggers argued that “it is neither built upon a minimal consensus of paradigmatic beliefs, resulting in […] an academic gap, nor capable of providing practitioners with adequate answers, what we call a managerial relevance gap.” (Bell et al. 2006, p. 1615)

Other authors defend research and believe it can be relevant in practice (Sandberg & Tsoukas, 2011). A central idea is to achieve engagement and collaboration between academics and practitioners (Hughes et al. 2011; - see co-citation Factor CC1 for Van de Ven, 2007, Van de Ven & Johnson, 2006; Starkey & Madan, 2010). The collaboration between academics and practitioners would be a possibility for co-creation of relevant knowledge (Antonacopoulou, 2010). This relationship, despite other tensions, should consider the need for impartial research against short-term results demanded by practitioners (Tushman et al. 2007). Therefore, some authors advocate that the academy-practitioner divide will remain unsolved, as science and practice have different meanings for researchers and practitioners (Kieser & Leiner, 2009, 2011; Nicolai & Seidl, 2010). Kieser and Leiner (2009) argue that the relationship between academics and practitioners would not be adequate for research but would be feasible to solve practical problems. Academics may orient towards practitioners “in ways that contravene their academic identity and research ethos” (Butler, Delaney, & Spoelstra, 2015, p. 732).

In general, the authors agree with the research relevance gap and the need to better understand relevance (Priem & Rosenstein, 2000; Learmonth, Lockett & Dowd, 2012; Parnell et al. 2012). This led authors to discuss the suggestion of possible remedies and their effectiveness. This is highlighted in the UK discussion of methodological alternatives to Mode 1 knowledge (Knights, 2008) like Mode 2 (Stiles, 2004; Knights, 2008) and other possibilities like Mode 3 research, engaged scholarship, pragmatic science, and evidence-based management (Romme et al. 2015).

Factor BC4 was labelled “Curricula Relevance”. The factor has a low explained variance with 5.17%. Here, B-Schools are questioned about their lack of relevance to business practice and the authors defend models that bring practice to the student, as in medical schools (Pfeffer & Fong, 2004). Pfeffer and Fong remarked that this: “entails focusing research on phenomena and problems of enduring importance, and building curricula that are evaluated, in part, by how well they actually prepare students to be effective in practicing the profession.” (Pfeffer & Fong, 2004, p. 93)

This implies a curricular program closer to the real-life challenges facing professionals and the development of competences for their professional activities (Navarro, 2006; 2008). Clinebell and Clinebell (2008) argue that there are two challenges facing B-Schools here: the shortage of faculty members with doctoral degrees and the lack of relevance. One of the remedies for resolving this would be to solve the tensions concerning the integration and use of executive professors or professional qualified faculty members.

Factor BC5 was identified as “B-School evolution and influence”. Here, recognition of the evolution of B-Schools appears in discussions between the authors (Augier & March, 2005; 2007; Augier & Teece, 2005; Khurana & Spender, 2012), as well as the discussion of fashionable theories (Abrahamson, 1996; Bort & Kieser, 2011). The evolution of B-Schools showed a transition from conducting research driven by the real world to theory driven research almost detached from its problems and contexts (Augier & March, 2005; 2007; Augier & Teece, 2005; Khurana & Spender, 2012). The evolution reflects the ideal of Simon, where “knowledge from practice” and “knowledge from the disciplines” are integrated (Khurana & Spender, 2012). B-Schools are where researchers are, and researchers create fashion (Bort & Kieser, 2011). B-Schools should be the primarily fashion theory setters. Scholars should research and learn how to settle fashionable theories (Abrahamson, 1996).

Factor BC6 is named “Reputation”. This factor includes studies that directly (Baden-Fuller, Ravazzolo, & Schweizer, 2000) or indirectly consider the influence of rankings on B-Schools (Rindova, et al. 2005; Macdonald & Kam, 2007; Rindova, Williamson, & Petkova, 2010). The media rankings, which include the MBAs of B-Schools, have a strong influence on reputation and consequently, the possibility of charging premium fees (Rindova et al. 2005). The indirect effect of the rankings of periodicals on publications (Rindova, et al. 2005; Macdonald & Kam, 2007) and works of research is also recognized (Ofori-Dankwa & Julian, 2005). B-Schools, as part of a HEI, would also suffer the effect of this relationship (Rindova et al. 2005). The research of Rindova et al. (2005) provided empirical support for the positive effect of academic research on B-Schools’ reputation.

Factor BC2 was named “Teaching Relevance” with an aim to make a difference to society (Tsui, 2013; Walsh, 2012). Most of the articles focus on the teaching of management; testing tools (Wright, Paroutis, & Blettner, 2013); studying evidence-based management (Trank, 2014; Gamble & Jelley, 2014; Klimoski & Amos, 2012); design thinking (Glen, Suciu, & Baughn, 2014); or, combining design science with critical realism (Willmott, 2012). Other articles focus on management and sustainability (Shrivastava, Ivanaj, & Persson, 2013; Aguinis et al., 2014; Akrivou & Bradbury-Huang, 2015); addressing concerns over how B-Schools are teaching management (Knights & Clarke, 2014; Costigan & Brink, 2015); and, making suggestions to improve teaching (Waddock & Lozano, 2013).

Other authors question the metrics used to evaluate scholarly performance and the legitimacy paradox that B-Schools face (Alajoutsijarvi, Juusola, & Siltaoja, 2015). As mentioned by Aguinis et al. (2014), the “majority of inquiry on scholarly impact has focused almost exclusively on one particular stakeholder and one type of measure: academics and citations.” (Aguinis et al., 2014, p. 624).