Bridging social network analysis and field theory through multidimensional data analysis: The case of the theatrical field

G Model POETIC 1261 No. of Pages 15

Poetics xxx (2016) xxx–xxx

Contents lists available at ScienceDirect

Poetics journal homepage: www.elsevier.com/locate/poetic

Bridging social network analysis and field theory through multidimensional data analysis: The case of the theatrical field Marco Serino* , Daniela D’Ambrosio, Giancarlo Ragozini Department of Political Science, University of Naples Federico II, Via L. Rodinò 28, 80138 Naples, Italy

A R T I C L E I N F O

Article history: Received 25 May 2016 Received in revised form 5 December 2016 Accepted 7 December 2016 Available online xxx Keywords: Bourdieu Field theory Theatre co-productions Social network analysis Multiple factor analysis Blockmodeling

A B S T R A C T

In this study, the theatre industry is conceived of as a field of cultural production, and analysed in the framework of Bourdieu’s field theory and social network analysis (SNA). The theatrical field is formalized as an affiliation network of companies participating in stage co-productions in Italy’s Campania region, over four theatre seasons. Differently from Bourdieu, but similarly to other works in the sociology of culture and the arts, the study focuses on relational and attribute-based dimensions of theatre production, presenting a novel way to combine field theory and SNA. By adopting the positional approach of SNA through Multiple Correspondence Analysis (MCA), Multiple Factor Analysis (MFA) and blockmodeling for affiliation networks, the study reveals oppositions among companies by reason of their differential partnerships in co-productions, and a combination of hierarchy and segmentation characterizing the network structure of the field. These oppositions also appear in the ‘objective’ social space defined by the attributes of companies and coproductions (positions and position-takings), showing the unequal distribution of symbolic cultural capital among theatre producers and the latter’s inclinations towards different theatrical styles and genres. © 2016 Elsevier B.V. All rights reserved.

1. Introduction Artistic production systems usually provide producers with differential opportunities for engaging profitable collaboration or experiencing isolation. These opportunities often take the form of partnerships in artistic projects that may give rise to a space of relations, i.e. a social space of cultural producers who derive their position in this space from the projects in which they participate. In social network analysis (SNA, hereafter), this implies the logic of the positional approach to affiliation networks (Burt, 1976, 1980; Doreian, Batagelj, & Ferligoj, 2005; Scott, 2000; Wasserman & Faust, 1994). However, from a rather different perspective, the same social space can be conceived of as a field of cultural production, “a space of objective relations among positions” (Bourdieu, 1993, p. 181) where ‘positions’ do not refer to concrete alliances among producers but to the latter’s location in an analytical space based on the distribution of capital forms. Albeit distinct, both these two perspectives allow us to see such social space as a space of opportunities where producers could fight or cooperate to gain relevant resources and artistic recognition. In this paper, thus, we propose to combine Bourdieu’s field theory and SNA for analysing the theatrical field as a social space in both the above-mentioned senses. Our approach is clearly different from Bourdieu’s own application of field theory

* Corresponding author. E-mail addresses: [email protected] (M. Serino), [email protected] (D. D’Ambrosio), [email protected] (G. Ragozini). http://dx.doi.org/10.1016/j.poetic.2016.12.002 0304-422X/© 2016 Elsevier B.V. All rights reserved.

Please cite this article in press as: M. Serino, et al., Bridging social network analysis and field theory through multidimensional data analysis: The case of the theatrical field, Poetics (2017), http://dx.doi.org/10.1016/j.poetic.2016.12.002

G Model POETIC 1261 No. of Pages 15

2

M. Serino et al. / Poetics xxx (2016) xxx–xxx

(that refuses ‘interactionist’ approaches [e.g. Becker, 1982] and SNA [see Bourdieu & Wacquant, 1992; Bourdieu, 1983, 1989]), but builds upon previous research that advocates the inclusion of manifest relations in field analysis, through SNA (e.g. Bottero & Crossley, 2011; Bottero, 2009; De Nooy, 2003). Our analyses rely on the potential encounter between the different notions of position in Bourdieu’s field theory and SNA. Although positions in networks must not be confused with Bourdieu’s field positions, these latter have a theoretical meaning in connection to the former.1 In fact, while network positions represent the structural location of groups of agents in a relational space by virtue of the relational patterns characterizing them, this also means that those agents may have similar interests and dispositions and be proximate in terms of social conditions (Bottero & Crossley, 2011), which relates to Bourdieu’s concept of position. Besides the ‘objective’ structure defined by the distribution of capital in the theatrical field, we posit that the latter also has its network structure, defined by the differential affiliation of producers in joint artistic projects. Positional analysis in SNA can thus be used to distinguish equivalent classes of producers with similar (relational) opportunities. In addition, as both theatre producers and productions have their own categorical attributes, this relational space, by means of multidimensional [38_TD$IF]data analysis, may also reveal agents’ distinctive properties and artistic inclinations. Bourdieu himself often utilizes the theatrical world as an example of artistic field (see Bourdieu, 1983, 1984, 1993), though without paying further attention to it. Thus the paper expands on field theory and positional analysis to shed light on the way theatre producers distinguish themselves by reason of their activity and [39_TD$IF]on how these distinctions matter for understanding the theatre production system as a “field of struggles” (Bourdieu, 1983, p. 312). The network structure of the theatrical field is formalized as a two-mode network in which theatre companies (the first mode) are involved in stage co-productions (the second mode). This permits us to operationalize field positions and position-takings (Bourdieu, 1983) through the network positions of companies and their affiliation in co-productions, these latter being ‘manifestations’ of companies as agents in the field, i.e. recognizable forms of position-taking (cf. Bourdieu, 1983, p. 312). We focus on the co-productions released by companies located in Italy’s Campania region (Serino, 2015), over four theatre seasons. In line with past research, and proposing a novel way to treat of the multidimensionality of this object of study, the positional approach of SNA is pursued by means of Multiple Factor Analysis (MFA) (Escofier & Pagès, 1998) and Multiple Correspondence Analysis (MCA) the key tool of Bourdieu’s methodology (Bourdieu, 1984, 1988) both applied to two-mode networks (D’Esposito, De Stefano, & Ragozini, 2014; Faust, 2005; Ragozini, De Stefano, & D’Esposito, 2015). The paper is organized as follows. [40_TD$IF]Sections 2 and 3 illustrate the background and our theoretical proposal and hypotheses, followed by data and methods in Sections 4 and 5. Then, Section 6 presents the results of each step of our analysis. In a first explorative step, we analyse the relational patterns of companies and co-productions through MFA, providing a preliminary indication of network positions. In a second step, network positions are analysed through generalized blockmodeling performed “in a confirmatory (deductive) mode” (Doreian et al., 2005, p. 234), following an explorative phase, in which we used this tool inductively. Finally, network positions are displayed in the factorial map along with companies and coproductions’ attributes (field positions and position-takings). Interviews with three ‘key informants’ provide a knowledge complement to the analysis. [41_TD$IF]Finally, Section 7 discusses the theoretical and methodological implications of our approach. Additional material concerning the application of factorial methods and blockmodeling to two-mode networks is provided in Appendix A, along with supplementary results in Appendix B. 2. Field structure and network positions Bourdieu’s field theory is intended to apply a relational view to the social space of cultural producers, described as “a space of objective relations among positions” (Bourdieu, 1993, p. 181), which corresponds to “the structure of the distribution of the capital of specific properties which governs success in the field” (Bourdieu, 1983, p. 312). Regarding theatre companies, the forms of capital defined by Bourdieu (1986) can be conceived of as follows. Economic capital refers to material resources such as theatre spaces, scenery, etc., and monetary resources deriving from box-office revenues or public and private funding. Cultural capital is best understood in the form of symbolic cultural capital deriving from the recognized value of a company’s prestige as institution (public acknowledgment, rewards, etc.) or for the ‘cultured’ character of its productions (depending on genres and performing styles). However, also economic capital can take a symbolic form by its recognition (Bourdieu, 1986, 1989; see Section 4). Social capital is “the aggregate of the actual or potential resources which are linked to possession of a durable network of more or less institutionalized relationships of mutual acquaintance and recognition or in other words, to membership in a group” (Bourdieu, 1986, p. 248). In this case, social capital is an asset managed by company members exploiting their social contacts in the art world of theatre (Becker, 1982). Artistic positions of companies in the field relate to the recognized possession of specific capital in the social space. In addition, the space of ‘position-takings’ is inseparable from that of artistic positions: it deals with “the manifestations of the social agents involved in the field” (Bourdieu, 1983, p. 312), i.e. artistic works like theatre productions. Finally, [42_TD$IF]Bourdieu’s conception of field includes the notion of habitus, which constitutes the system of actors’ dispositions towards the “objective

1 In order to avoid confusion between the terminologies of Bourdieu’s field theory and SNA, we shall speak of ‘field positions’ in the sense of the former, and of ‘network positions’ with respect to the latter.

Please cite this article in press as: M. Serino, et al., Bridging social network analysis and field theory through multidimensional data analysis: The case of the theatrical field, Poetics (2017), http://dx.doi.org/10.1016/j.poetic.2016.12.002

G Model POETIC 1261 No. of Pages 15

M. Serino et al. / Poetics xxx (2016) xxx–xxx

3

probabilities” of gaining profits in the field and, thereby, mediates the relation between positions and position-takings (Bourdieu, 1983, p. 342, 344). Field structure “is different from the more or less lasting networks through which it manifests itself” (Bourdieu & Wacquant, 1992, p. 114), and Bourdieu has, in fact, explicitly rejected SNA and any focus on “visible interactions” (Bourdieu, 1983, p. 311). Nonetheless, some authors criticize [43_TD$IF]Bourdieu’s relational method and point out that observed relations are important for ‘extracting’ field structure, highlighting habitus formation, and defining indicators of capital forms, and SNA is one suitable approach apt to do so (Bottero & Crossley, 2011; Bottero, 2009; Crossley, 2009; De Nooy, 2003; Emirbayer & Johnson, 2008; Mohr, 2013). One way to take concrete relations into account in line with Bourdieu’s idea, avoiding an ‘interactionist fallacy’ (Emirbayer & Johnson, 2008), is to consider network [45_TD$IF]positions (Burt, 1976). More broadly, this means putting together network positions and field positions. For two or more companies, being proximate in social space means occupying the same position and sharing “similar structural relations to economic and cultural resources” (Bottero & Crossley, 2011, p. 101). Transposing this idea into SNA, speaking of similarity of positions in social space would lead to embrace the “positional approach” (Burt, 1976, 1980; Doreian et al., 2005), which highlights the underlying network structure resulting from the patterns of relations partitioned into equivalent classes of social actors, in contrast with the relational approach focused on observed relations and network cohesion (Burt, 1978, 1980). Actually, this opposition is somewhat similar to that of interaction vs. structure in field theory (Bourdieu & Wacquant, 1992). Two actors occupy the same position in that they share similar patterns of relations with other members of a network, even if no direct links exist between them (Burt, 1976; Lorrain & White, 1971; Sailer, 1978; White, Boorman, & Breiger, 1976). It is not surprising, therefore, that the positional approach is usually adopted to fulfil the analysis of network structure within the framework of field theory, often using both blockmodeling (White et al., 1976) and correspondence analysis (CA), a necessary complement for analysing fields with their categorical attributes (Anheier, Gerhards, & Romo, 1995; Bottero & Crossley, 2011; De Nooy, 2003; Gerhards & Anheier, 1989; Giuffre, 1999, 2001). By means of CA, the social space of cultural producers based on their attributes can be appreciated as a space divided up into regions (Bourdieu, 1989), while blockmodeling provides a suitable partitioning of its network structure in clusters named ‘positions’, i.e. network positions (Doreian et al., 2005).2 We build upon this research line proposing our distinctive contribution in both theoretical and methodological terms. 3. Affiliation networks of stage co-productions Most fields of cultural production can be conceived of as dually structured systems that constitute of creators and their creative products or their affiliations in events or institutions (De Nooy, 2002; Giuffre, 1999, 2001). In the theatrical field, the connection between theatre companies and their co-productions can be represented by an affiliation (two-mode) network (Wasserman & Faust, 1994). A relational account of such field can thus be obtained through this non-dyadic data structure in which companies and works are linked insofar as subsets of companies are connected to the works they jointly realized (Faust, 1997). Duality is clearly implicated in this kind of networks since the two sets of entities are mutually related (Breiger, 1974, 2000). Co-productions are project-based activities in which two or more companies cooperate, for a limited time period, in order to produce and perform a theatre play. Hence, they are also partnerships that offer chances to establish or reinforce linkages among the organizations involved, beyond the mere formal contract of co-production (Serino, 2015), meaning that the relations we analyse are not concerned only with ‘interactions’ among participants in a set of co-productions. Instead, we deal with more subtle relations that link cultural producers, even if not directly connected to one another, by virtue of common third parties who are members of some joint projects. If a company shares one or more co-productions with certain other companies, it might also be potentially in contact with organizations that have linkages with its partners and, by contrast, it is inevitably disconnected to other different companies. The positional approach of SNA helps address these issues, in that theatre companies turn out to be positioned in the network by reason of their participation in co-productions and by virtue of the specific structural patterns characterizing such participation. Further, we apply positional analysis through generalized blockmodeling, relying upon a combination of exploratory and confirmatory strategies (Doreian et al., 2005; cf. Prota & Doreian, 2016), which allows us to verify hypotheses about the network structure of the theatrical field. In artistic fields, like the organizational field of American resident theatres (DiMaggio, 1986), or the German literary field (Anheier et al., 1995; Gerhards & Anheier, 1989), the structure of a network of producers can exhibit segmentation and/or hierarchy, with their corresponding structural patterning. The distribution of capital in the field may be related to the emergence of such structural models. As alliances among companies are necessarily selective, a network of co-productions will obviously have a basic degree of segmentation. Theatre companies may be inclined to enter more or less exclusive alliances in order to maximize their respective economic capital, e.g. by sharing the costs of a theatre-making venture, or by mutually compensating for the lack of specific financial, material and human resources (Gallina, 2007). Similarly, two distinguished cultural producers active in different sectors might participate in a joint project to reinforce their symbolic cultural capital. In either case, the

2

In the following, we shall use the words ‘cluster’ or ‘network position’ as synonyms.

Please cite this article in press as: M. Serino, et al., Bridging social network analysis and field theory through multidimensional data analysis: The case of the theatrical field, Poetics (2017), http://dx.doi.org/10.1016/j.poetic.2016.12.002

G Model POETIC 1261 No. of Pages 15

4

M. Serino et al. / Poetics xxx (2016) xxx–xxx

[(Fig._1)TD$IG]

Fig. 1. Affiliation networks of the four seasons: bipartite graphs (circle = company; square = co-production).

co-production is intended to exploit some interdependent resources. In addition, cooperation is more likely to occur among companies with similar artistic purposes. For example, two companies can be interested in co-producing highly popular theatre plays or, instead, in staging more unconventional works, with dissimilar earning opportunities in terms of capital forms. Indeed, a typical form of segmentation is the separation between sectors of ‘high’ and ‘low’ culture (Anheier et al., 1995, p. 865), by which subgroups of companies will converge upon different market segments. This will result in a network partitioned into distinct clusters (network positions) of companies involved in the same co-productions, with very few or no linkages among these clusters. We thus expect that: H1. [46_TD$IF](segmentation)Companies with interdependent economic or symbolic resources, or with similar artistic motivation, will participate in co-productions in relative isolation from other partners. [47_TD$IF]In a segmented structure thus hypothesized, reciprocity in resource exchange does prevail. However, different alliances can occur if some partners hold a greater amount of resources than others. Material and financial means can be offered as a support by a large organization to other small companies, but there could also be an explicit wish for a company to work with someone (an actor, a director, etc.) and his group, due to his/its higher prestige or valuable (and recognized) competence (Gallina, 2007). Hence, an unequal distribution of economic or symbolic capital among network positions may result in a hierarchically structured network when, for example, minor companies that are not able to rely upon each other because of their scarce resource availability will seek access to these [48_TD$IF]resources by participating in co-productions with other powerful institutions. Thus, certain linkages between network positions may consist in dominance or gatekeeping, i.e. when (major) companies participate in a large number of co-productions that involve other (minor) companies connected only via those co-productions. Therefore, we hypothesize that: H2. [49_TD$IF](hierarchy)Companies with higher resource availability will attract companies less endowed with capital to be exceptionally involved in some of their projects. [50_TD$IF]4. Data Our network data rely upon information on the ties between companies and the co-productions which they participated in. As we are interested exclusively in productions that translate into relations (partnerships) among different companies, we

Please cite this article in press as: M. Serino, et al., Bridging social network analysis and field theory through multidimensional data analysis: The case of the theatrical field, Poetics (2017), http://dx.doi.org/10.1016/j.poetic.2016.12.002

G Model POETIC 1261 No. of Pages 15

M. Serino et al. / Poetics xxx (2016) xxx–xxx

[(Fig._2)TD$IG]

5

Fig. 2. Distribution of co-productions by the number of companies involved.

do not consider individually produced plays. We focus on Italy’s Campania region, which holds a local theatre system resembling the national one (Serino, 2013). All types of company acknowledged by the Italian Ministry of Cultural Heritage (MIBACT, in Italian) up to 2014 do exist in the region. Data collection began in autumn 2013 by selecting a set of theatre companies operating in [51_TD$IF]the Campania region during four seasons (from 2011/2012 to 2014/2015) and running an own theatre venue located in this area. This eligibility criterion derives from the idea that companies with stable location better express the local cultural industry than touring companies. We thus obtained a list of 33 theatre companies: five producing houses belonging to the category of Teatro Stabile, the prominent form of official, publicly subsidized resident theatre established in Italy since the late 1940s, and 28 building-based companies that run a venue without being authoritatively acknowledged as producing houses. This is the initial population of focal companies linked to other non-focal organizations through a chain of co-productions. Then we gathered information on the co-productions the focal companies released during the four seasons by means of web-based questionnaires filled by companies’ staff, and referring to companies’ websites as informative support. We considered only the co-productions performed in co-producers’ venues located within Campania,[52_TD$IF] which count for the activity in this region even when extra-regional co-producers take part in the project. Excluding 13 focal companies not involved in co-productions in any season,3 the whole network dataset contains 20 focal companies (5 producing houses and 15 building-[53_TD$IF]based companies) and 80 non-focal organizations (of which 32 are non-theatrical) participating in a total of 157 co-productions collected for the entire period.4 The final longitudinal dataset consists of four affiliation networks related to the four seasons; each network comprises a distinct number of companies and co-productions (see Table B.3). The bipartite graphs of these networks (Fig. 1) are obtained by the software NetDraw 2.138 (Borgatti, 2002). The distribution of the co-productions by the number of partners involved (Fig. 2) is heavily skewed, and around 80% of the co-productions involve two organizations only. For each focal company we collected data on the main theatre venue. The latter’s seating capacity is a measure of size, while theatre spaces are classified into three main types to be found in Italy. Historic theatre is the traditional eighteenth- or nineteenth-century boulevard theatre; modern theatre corresponds to the twentieth-century theatre model designed under modern architectural and technical standards; non-theatrical building is a space refurbished to become suitable for hosting live performances. As for non-focal companies, we consider the aforementioned attributes and also the type of organization and the location within or out of the region. Some attributes are used as proxies for operationalizing capital forms. Authoritative acknowledgment of producing house functions as a “legal consecration of symbolic capital” and a sort of “title of nobility”; i.e. a credential officially recognized and sanctioned by relevant institutions (Bourdieu, 1989, p. 21–22). The size of the venue is a proxy for economic capital. Larger theatres may likely provide higher income than smallest ones, either as box-office revenues or, in Italy’s system, as public subsidies. Indeed, producing houses usually hold high portions of both economic (large venues and subsidies) and symbolic cultural capital, the latter being related to the amount of consecrated cultural capital guaranteed by the state to them as having the status of Teatro Stabile i.e. the state’s meta capital (Bourdieu & Wacquant 1992, p. 114). In addition, even possessing a given theatre space can be an indicator of economic capital converted into symbolic capital, as is the case of historic theatres, which are often exemplars of the distinguished ‘teatro all’italiana’ model but are also more expensive than other space types. Social capital is more difficult to measure for we deal with

3 4

These companies do not differ from others in the dataset. In particular, they are similar to minor companies present in the network. See Appendix B, Tables B.1 and B.2, for an extensive description of the characteristics of companies and co-productions.

Please cite this article in press as: M. Serino, et al., Bridging social network analysis and field theory through multidimensional data analysis: The case of the theatrical field, Poetics (2017), http://dx.doi.org/10.1016/j.poetic.2016.12.002

G Model POETIC 1261 No. of Pages 15

6

M. Serino et al. / Poetics xxx (2016) xxx–xxx

organizations rather than individuals. Thereby, we discuss the role of this capital form at the micro-level of interpersonal relations, without formalizing it. Co-productions’ attributes are theatre genres, types of script (i.e. the literary or drama work on which a given production is based) and performing styles. We classify co-productions by eight theatre genres based on historical and content definition, constructed by an in-depth reading of informative materials on co-productions available on companies’ websites and other web sources (e.g. programmes, reviews, etc.). Scripts may be original, revised or not revised, while performing style is classified into two non-exclusive categories: admitting interactions with the audience, and combining multiple art or communication forms. Genres and performance characteristics express the symbolic cultural capital co-productions are likely to provide to companies i.e. the symbolic profit (Bourdieu, 1984). In particular, theatre ventures gain the maximum symbolic profit as opposed to economic profit (Bourdieu, 1983, 1993) from the rarest and mostly unconventional performances or from drama pieces that have not yet obtained enough popularity (e.g. the genres of experimental theatre and new drama trend). This means that companies chiefly concerned with such works may increase their amount of symbolic cultural capital (recognition). In order to gain additional insight into the network positioning of companies, interviews were conducted in autumn 2015 with three ‘key informants’ of the field, namely the directors of three companies that resulted as having the most peculiar relational patterns (see Section [54_TD$IF]6.1). We showed the network graphs to these key informants asking them for opinions about their work and that of other companies of which they knew of. In this respect, we utilize interview results to consider the role of capital forms in the constitution of alliances for co-productions. 5. Assessing field structure: a multidimensional joint approach In this section, we set forth our methodological proposal for the study of the theatrical field. Bourdieu sets SNA aside because he does not reckon it suitable for putting his “relational mode of thinking” in practice, “save by way of correspondence analysis” (Bourdieu & Wacquant, 1992, p. 114; our italics). Against this criticism, we argue that the two domains of network structure (defined by relations) and objective structure (defined by attributes), both considered as inherent parts of the field, can be joined into a comprehensive framework based on multidimensional [5_TD$IF]data analysis. Hitherto, CA has usually been performed in the framework of field theory and SNA using categorical attributes of network units (e.g. Anheier et al., 1995; Gerhards & Anheier, 1989). Our task is, instead, to analyse field structure, i.e. the objective structure defined by the distribution of capital (measured via specific categorical variables), through the patterns of relations, thanks to MCA and MFA applied to the two-mode network data of theatre co-productions. Secondly, we pursue a two-fold usage of the notion of position (in SNA and field theory), mapping network positions and categorical attributes through MCA, in conjunction with blockmodeling. This is what we call a joint approach by which to analyse the field in terms of both patterns of relations and attributes of companies and co-productions.5 5.1. Multidimensional [56_TD$IF]data analysis MCA (Blasius & Greenacre, 2006) is applied to two-mode network data (D’Esposito et al., 2014; Ragozini et al., 2015) and performed as an available factorial method within Multiple Factor Analysis (MFA) (Escofier & Pagès, 1998), a technique suitable to handle multiple data tables. Here we deal with four different tables (i.e. the affiliation matrices) each of which has the companies placed in rows and the co-productions placed in columns. When a given company has participated in a coproduction, the corresponding entry in the matrix is 1, and 0 otherwise. We also adopt the doubling perspective so as to apply MCA deriving the indicator matrix from the affiliation matrix through the complete disjunctive coding (D’Esposito et al., 2014). We also include [57_TD$IF]the categorical attributes of both companies and co-productions as illustrative variables in MCA and MFA. These analyses are performed through the software Spad 5.5, adopting a proper coding in order to apply usual factorial methods to network data. MCA and MFA for affiliation networks permit to appreciate positional equivalence on the factorial maps, showing the degree of structural similarity of companies and co-productions as revealed by the distances among points representing companies, and among vectors representing co-productions. Note that MFA is used to conduct both partial (on each time slice) and global (on the overall time span) analyses; for the sake of simplicity and clarity, we opt mainly for global MFA, in that it permits to gauge structural properties over the whole time span considered. 5.2. Positional equivalence and generalized blockmodeling Generalized blockmodeling for two-mode networks (Doreian, Batagelj, & Ferligoj, 2004) allows us to partition the patterns of relations of the network units (both companies and co-productions) into clusters (called [58_TD$IF]network positions) of equivalent units, basing on equivalence forms that go beyond structural or regular equivalences (Lorrain & White, 1971; Sailer, 1978; White & Reitz, 1983). The ties between clusters are called blocks. In two-mode networks, blocks have a clear dual

5 A very recent example of joint approach combining SNA and “a field-theoretic understanding of correspondence analysis” is the mixed-method approach of “comparative netfield analysis” (Sonnett, 2016).

Please cite this article in press as: M. Serino, et al., Bridging social network analysis and field theory through multidimensional data analysis: The case of the theatrical field, Poetics (2017), http://dx.doi.org/10.1016/j.poetic.2016.12.002

G Model POETIC 1261 No. of Pages 15

M. Serino et al. / Poetics xxx (2016) xxx–xxx

7

structure, as they are made of the ties between two clusters of units from two different node sets placed in a rectangular matrix. This means that “the partitions of rows and columns are not identical” and, therefore, the two clusters are partitioned at the same time, but in different ways (Doreian et al., 2005, p. 249). Structurally equivalent blocks will obtain only when companies are clustered together because they participate in exactly the same co-productions (these are called complete blocks, which are all covered by 1s, in contrast to null blocks where no ties are present and, thereby, they are all covered by 0s). The same holds for co-productions involving exactly the same companies (see Borgatti & Everett, 1992, p. 99). Instead, regularly equivalent or simply regular blocks will obtain when, for example, given a cluster of companies and a cluster of co-productions, each company participates in at least one co-production, and each co-production involves at least one company (cf. Doreian et al., 2005, p. 264). This is formalized by the presence of at least one 1 in each row and each column of that block. In addition, other block types based on the generalization of equivalences are available in this approach (see Section A.1). As usual, an image matrix shows the block types corresponding to the model of connections of the two clusters. In generalized blockmodeling, different ideal block types [59_TD$IF]may be defined and empirically ‘tested’ by measuring the discrepancy between them and the network data. Through an algorithm that looks for the best partitions by relocating the units in the clusters (local optimization), this procedure seeks to determine a blockmodel structure that minimizes a proper criterion function which is also a measure of fit [60_TD$IF]i.e. the divergence between the ideal model and the resulting one. This means to minimize the number of 1s or 0s that are not consistent with the ideal block; these errors are called inconsistencies. On the basis of such procedure, generalized blockmodeling can be used in two ways: exploratory (inductive) and confirmatory (deductive). “In the exploratory mode, only a set of block types defining an equivalence is declared in advance of using blockmodeling”, relying upon available relational data. In the confirmatory mode, researchers can also have “theoretical hypotheses concerning the structure of the ties between positions” to be put into a more or less stringent prespecification (e.g. location of block types, constrains upon clustering) of the blockmodel before the analysis and “prior to fitting” it, in order “to test that blockmodel and the knowledge that went into its prespecification” (Doreian et al., 2005, p. 233–235, 349). In this work we started with an inductive strategy by gradually prespecifying blockmodel partitions, until our theoretical and empirical knowledge on the network was put into the prespecification to move on with the confirmative mode. Note that empirical knowledge is derived from both the preceding exploratory analysis via MFA and the inductive blockmodeling steps, as we could not fully prespecify our blockmodels prior to analysing in detail the (sparse) network we dealt with. This makes our approach partially confirmatory (Doreian et al., 2005, p. 27; see Section A.4). Generalized blockmodeling is performed through the software Pajek (Batagelj & Mrvar, 1998). Finally, the network positions highlighted by blockmodeling are added to the final MFA, in order to analyse their association with companies and co-productions’ attributes. 6. Results 6.1. The space of companies and co-productions Fig. 3 shows the results of the global MFA performed for the entire four-season span, where the two spaces defined by the relational patterns of companies and co-productions are superimposed. The map shows only the first two factors explaining the highest percentages of inertia, whereas we consider the first four factors for a thorough analysis. In the map, each point represents a company, and its coordinates are the weighted average of the coordinates over the four seasons. Each coproduction is represented by two vectors that pass through the axes origin and correspond to the positive and negative poles of each co-production, i.e. [61_TD$IF]the participation or non participation of companies in these projects in line with the doubling perspective (Ragozini et al., 2015). The angle between the vectors is proportional to the correlation between the participation patterns of co-productions. Hence, two vectors forming a small angle are relationally similar. In most cases, the coproductions shown in the map are so similar that they collapse on the same segment: they involve exactly the same companies and are, thus, structurally equivalent. For this reason, labels are not displayed for all co-productions. Thanks to their structural similarity, we can distinguish five network positions (or clusters) of companies (see also Table 1). At the bottom right, the first network position includes Nuvole and ProgMuseo, while, in the panel above, StabileNapoli and CampaniaFestival constitute a second network position. Both network positions are characterized by the co-productions located nearby, and particularly by the longest vectors representing some of these co-productions. Interestingly, the long vector lying close to the horizontal axis and having the largest x coordinates might represent an ‘intersection’ between StabileNapoli and Nuvole: its midway location is due to the joint participation of these companies in such co-productions. ProgMuseo and Nuvole, along with StabileNapoli and CampaniaFestival, have the largest x coordinates and their relational patterns are the most peculiar in the network. We should also note that these four companies exhibit the highest number of partnerships in co-productions (Figs. 4 and 5), the highest centrality6 scores (Table B.4), and give the main contributions to the first and second axes. Hence, they best characterize the first two dimensions albeit the second dimension distinguishes the two pairs of organizations, which belong to two distinct partitions (Table 1). We thus decided to interview the directors of the two focal companies StabileNapoli and Nuvole as ‘key informants’. StabileNapoli is a producing

6

Centrality refers to the idea of importance, prominence, or prestige of actors in a network (Faust, 1997; Wasserman & Faust, 1994).

Please cite this article in press as: M. Serino, et al., Bridging social network analysis and field theory through multidimensional data analysis: The case of the theatrical field, Poetics (2017), http://dx.doi.org/10.1016/j.poetic.2016.12.002

G Model POETIC 1261 No. of Pages 15

8

M. Serino et al. / Poetics xxx (2016) xxx–xxx

[(Fig._3)TD$IG]

Fig. 3. The space of companies and co-productions: joint representation and global analysis. Map of the first two factorial axes; points represent companies, with regular labels, vectors represent co-productions, with bold labels indicating the title of the play and the corresponding season (e.g. [37_TD$IF]‘_11’= 2011/2012 season). Table 1 Companies with largest coordinates and contributions in the global MFA, along with corresponding clusters. Axis 1

Nuvole ProgMuseo StabileNapoli CampaniaF. Magazzini TeatrAzione Carrozza Toto Bracco

Axis 2

Axis 3

Axis 4

Clusters

Coord.

Contr.

Coord.

Contr.

Coord.

Contr.

Coord.

Contr.

(5 classes)

13.79 6.19 10.43 3.82 1.65 1.19 0.85 0.92 0.81

51.18 10.32 29.29 3.94 0.73 0.38 0.19 0.23 0.18

7.33 6.28 9.92 7.75 2.49 1.70 0.95 0.35 0.29

19.96 14.66 36.59 22.34 2.30 1.07 0.33 – –

0.34 0.84 2.52 1.83 12.74 7.87 4.06 1.29 1.10

– 0.27 2.43 1.28 61.86 23.61 6.28 – 0.46

0.25 0.21 0.84 0.23 0.82 0.47 0.09 10.22 8.67

– – 0.37 – 0.36 0.12 – 55.36 39.87

1/5 1/5 2/5 2/5 3/5 3/5 3/5 4/5 4/5

Note: all 91 companies with lower scores belong to cluster 5/5 (not shown in the table).

house and CampaniaFestival is the foundation that organizes the theatre festival ‘Napoli Teatro Festival Italia’. This is a special case of co-production, as the two institutions are formally distinct but were managed by the same artistic director, who says: [62_TD$IF]When I arrived here [as a new director] . . . I began to coproduce everything as there was no conflict of interest at all . . . there was then a convergence of interests that . . . [106_TD$IF]was convenient for both. (Artistic director of StabileNapoli)

These two institutions, due to their public acknowledgement and funding, are those better endowed with both symbolic and economic capital in the local field: CampaniaFestival holds public financial means to release a festival programme built on a large part of co-productions with StabileNapoli, which also benefits from state subsidies. The festival may sustain a production with a higher pay and guarantee the visibility of a première hosted in the course of the festival. Hence, producers that aspire to gain prestige may benefit, also economically, from their participation in a well-regarded festival. Instead, alliances between the producing house Nuvole and its non-focal partner ProgMuseo arise from artistic purposes that seek their cultural but non-theatrical counterpart: ProgMuseo is an association of art historians working in art museums . . . they provided scientific knowledge for the production ‘Caravaggio’ [and for other similar productions] . . . I say ‘well, I’m a director, not an art historian[107_TD$IF]’. (Artistic director of Nuvole)

Please cite this article in press as: M. Serino, et al., Bridging social network analysis and field theory through multidimensional data analysis: The case of the theatrical field, Poetics (2017), http://dx.doi.org/10.1016/j.poetic.2016.12.002

G Model POETIC 1261 No. of Pages 15

M. Serino et al. / Poetics xxx (2016) xxx–xxx

[(Fig._4)TD$IG]

9

45 40 35 30 25 20 15 10 5 0 p Na ile b a St

i ol

i le zin vo az Nu ag M

to To

A Ex

o sil

on co yl ac ab Br B

nt po Co ro nt no a r ic le El Sa

N TA

D

ia

na

l i e o ro ile do in aa lli n ch le za ab ul sb Be Po st To M na Vi e n In D Sa

S NT

ti Ar

Fig. 4. Overall number of co-productions released by focal companies over the four seasons (2011–2014).

[(Fig._5)TD$IG] 30 25 20 15 10 5

Ca

m

pa n

ia F Pr esti o g va T e Mu l at s e rA o P R z io OS ne Fe sti Ca PET va r ro lC z z i tta a O n Sp gT et ea tri Fo o F o Sca l l kt l z ea a te Pu rn te R ca St a g a a b zz T a ile G i v e T e ern n a a Ve t r o R Es t su o m vi o a te at Am r o Ar e ia t E na a Fe TC So s ti r o l e va na Pi lA vi ca ra g n n Po del o n s it lia a n na o Sa F e s St n C t a b ar ile lo T To T h ea t sc e a r iU tN n it at i Br Ti ux T o nao ur s V e b illo su n v ia Vi n a ca r ia

0

Fig. 5. Overall number of co-productions involving non-focal companies over the four seasons (2011–2014). Companies involved in only one co-production are excluded.

In so doing, these organizations have the opportunity of increasing their respective symbolic capital, as Nuvole will benefit from the recognition of the cultural value of its activity, while ProgMuseo will see its ‘scientific’ contribution taking the form of performed works. Other two network positions of companies characterize the third and fourth dimensions. They are not well represented by the first two axes, but their location on the left-hand side of the map is of interest. One cluster is made of the buildingbased company Magazzini and its non-focal partners TeatrAzione and Carrozza (non-building-[65_TD$IF]based companies), whose joint work focuses mainly on youth theatre. Due to its peculiar relational pattern, we considered Magazzini’s artistic director as a third key informant, who said: [6_TD$IF]Both Carrozza and TeatrAzione . . . have collaborated with me . . . in my own productions. Later, a kind of collaboration came into being.[67_TD$IF] Indeed, this sort of co-productions we make together.[108_TD$IF] (Artistic director of Magazzini)

Another network position features the focal building-based companies Bracco and Toto, which exchange their respective artistic personnel but also their capacity for organizing, producing and hosting theatre plays in the segment of vernacular comedies. Albeit different, the third and fourth clusters of companies are similarly characterized by a mutual compensation of economic capital (in either material or human forms). In particular, as reported by the artistic director of Nuvole, companies like Magazzini often have organizational means to stage theatre works but lack artistic personnel. They thus provide organizational support to younger theatre groups, i.e. Carrozza and TeatrAzione, which supply human resources. The relational patterns of the third and fourth clusters are opposed to those of the first two clusters, on the horizontal dimension. The vectors of co-productions that characterize the two pairs of network positions virtually lie on the same diagonal, but one pair is neatly in contrast with the other as having the positive poles on the two opposite sides of the diagram: companies that participate in co-productions on one side tend not to participate in those lying on the other.

Please cite this article in press as: M. Serino, et al., Bridging social network analysis and field theory through multidimensional data analysis: The case of the theatrical field, Poetics (2017), http://dx.doi.org/10.1016/j.poetic.2016.12.002

G Model POETIC 1261 No. of Pages 15

10

M. Serino et al. / Poetics xxx (2016) xxx–xxx

In addition, there is a fifth network position of companies with very sparse relational patterns, represented by [69_TD$IF]the points standing close to the axes origin. These companies are characterized by a small number of partnerships in co-productions (Figs. 4 and 5) and belong to a cluster distinct from the others (Table 1), whose composition is quite heterogeneous:[70_TD$IF] it includes, among others, a wide number of smaller organizations less endowed with capital than those in the other positions, plus several extra-regional producing houses. 6.2. Blockmodel partitions This section shows the results of the confirmatory blockmodeling in order to verify the hypotheses presented in [71_TD$IF]Section 3, along with a more detailed representation of the network structure of the theatrical field. Nonetheless, the present analysis focuses on the main findings relevant to our theoretical framework, setting other aspects of the ideal and empirical blockmodels aside. The visual representation of the final blockmodels is a matrix displaying blocks made of ties between companies (in rows) and co-productions (in columns), where a dark cell stands for the presence of a tie, blank otherwise (Fig. 6). The notation for [72_TD$IF]blocks is based on the number assigned to each cluster both in Figures and Tables (e.g. [73_TD$IF]‘1-3’ stands for ‘cluster 1 of companies and cluster 3 of co-productions’).7 Subsequently, the resulting blocks will be given proper labels in order to ease the interpretation of the final analysis of [74_TD$IF]Section 6.3. 6.2.1. Prespecification With the aid of our exploratory positional analysis ([75_TD$IF]Section 6.1), we now translate our hypotheses into empirical terms consistent with the blockmodeling, defining the latter’s [76_TD$IF]prespecification. This procedure leads to define the possible ideal blocks for each season (Table B.5) and then to obtain the final blockmodels with the resulting inconsistencies (errors) in each block and in total for each time span (Table B.6). Note that the detailed definitions of blocks in the ideal models also derive from the early attempts of exploration and progressive (re-)prespecification of the blockmodels. In this regard, as certain companies and co-productions have resulted as belonging to the same cluster in different seasons, the clustering is constrained in order to maintain this partial composition and the order of appearance of some clusters which is different from the order shown in Table 1 [7_TD$IF]– for the sake of comparison among them over time. The global MFA reveals five distinct network positions of companies with different compositions and contributions given to the factorial axes (Table 1). Moreover, we saw that the companies in one network position tend not to participate in the coproductions of another network position. However, we ignore to what extent these clusters are unrelated. Theoretically, we emphasize their separation because of the motivation and benefits companies have in terms of capital endowment from coproducing plays more or less exclusively with specific partners. Therefore, in line with the hypothesis of segmentation H1, we expect that such network positions will equally be distinct in the blockmodels. More specifically, we expect to see four clusters of both companies and co-productions, in every season, in blocks (1-1), (2-2), [78_TD$IF](3-3), and (4-4), located on the main diagonal. These blocks are specified as complete or dense blocks. Furthermore, as suggested by the preliminary blockmodels, the fifth network position is expected to be split into different blocks that may include companies with more or less sparse relational patterns. Thus, blocks (5-5) and (6-6), or (6-5) depending on the clusters resulting for each season are defined not only as null, complete or dense, but also as one of the variants of regular blocks (see Section A.1, Table A.1). The hypothesis of hierarchy H2 concerns the possibility of inter-group partnerships in co-productions that link distinct network positions with different endowment of capital.8 For instance, this may arise when prominent companies allow minor ones to participate in some projects: The proposal [of co-producing a play] arose almost always from us. Of course it happened that minor organizations made a proposal but, indeed . . . this was ‘I want make this thing’, and then we jointly developed the idea.[109_TD$IF] (Artistic director of StabileNapoli)

In particular, we expect a connection between companies better endowed with capital (the second network position) and some less advantaged ones (the fifth network position), these latter taking part in some of the joint projects always involving the former. More specifically, we expect that this does not occur anywhere in the blockmodel and just by chance, but rather in a specific location, namely block (5-3) in the 2013/2014 season and block (6-3) in the remaining seasons. 6.2.2. Empirical blockmodels The hypothesis of segmentation H1 is confirmed by the final blockmodels: the network structure of the theatrical field is for the most part differentiated in several relatively unrelated segments; the blocks on the main diagonal are well fitted, with few inconsistencies that range from 5 to 13 (see Table B.6).9 Fig. 6 shows the different composition of all blocks. The first four

7

Blockmodels do not include all companies and co-productions of each season, because of the sparse nature of the network. Hierarchy is formalized as a core-periphery model in the special case of two-mode networks, as described in [7_TD$IF]Section A.2. 9 According to Doreian et al. [80_TD$IF](2005, p. 353), there is “no definitive statement concerning thresholds below which blockmodels can be seen as fitting the data and above which they do not”. Thereby, we can accept the model and its fitting by reason of the minimization of the criterion function. 8

Please cite this article in press as: M. Serino, et al., Bridging social network analysis and field theory through multidimensional data analysis: The case of the theatrical field, Poetics (2017), http://dx.doi.org/10.1016/j.poetic.2016.12.002

G Model POETIC 1261 No. of Pages 15

M. Serino et al. / Poetics xxx (2016) xxx–xxx

[(Fig._6)TD$IG]

11

Fig. 6. Final blockmodels; companies are placed in rows, co-productions in columns.

blocks located on the diagonal correspond to the network positions shown in Table 1– [81_TD$IF]except for block (2-2) in the 2014/2015 season.10 It is worth noting that some ties between companies and co-productions in other blocks break an otherwise neat segmentation. While some of these ties reveal a hierarchical structure (see below), other meaningful ones link the companies in block (1-1), which pertain to the first network position, and the companies in block (3-3), namely the second network position, through the co-productions in block (3-1), in the last season. This linkage has been suggested by the explorative MFA in [84_TD$IF]Section 6.1 and leads to label the first network position as the ‘Insiders’. Blocks (2-2) and (4-4) maintain for the most part their own closure as separate segments; the corresponding third and fourth network positions can thus be named the ‘Outsiders’. Blocks (6-6) or (5-5), in the first three seasons, and (6-5) in the last season, represent the periphery of the regional theatre system, as it includes most of the loosely connected small companies belonging to the fifth network position, which can be named the ‘Fringe’. The final blockmodels also lead to confirm the hypothesis H2, except for the first season. A hierarchical structure results from the links between the core blocks (3-3) and their peripheries in blocks (6-3), in the second and fourth seasons, and (5-3) in the third season. However, such hierarchical structure is subject to two substantive interpretations. In fact, those

10

[82_TD$IF]In this season, the composition of block (2-2) is different in that [83_TD$IF]the companies Bracco and Toto cease to collaborate.

Please cite this article in press as: M. Serino, et al., Bridging social network analysis and field theory through multidimensional data analysis: The case of the theatrical field, Poetics (2017), http://dx.doi.org/10.1016/j.poetic.2016.12.002

G Model POETIC 1261 No. of Pages 15

12

[(Fig._7)TD$IG]

M. Serino et al. / Poetics xxx (2016) xxx–xxx

Fig. 7. The space of companies and co-productions’ attributes (empty circles and squares) along with network positions (solid squares).

peripheries consist of co-productions that involve not only intra-regional, minor companies but also extra-regional producing houses which partly belong to the fifth network position. On the one hand, the amount of capital held by the institutions in the core attracts region’s peripheral companies less endowed with capital, which clearly confirms H2. On the other hand, these structures are also indicative of the power held by prominent companies in the core for attracting foreign, powerful partners to be involved in the region’s activity. Such collaboration also matters as a source of social capital (emerging linkages between respective staffs), economic capital (shared production costs) and symbolic cultural capital (the prestige of the participation of such extra-regional producing houses). By and large, this is an overall evidence of the prominence of the second network position lying in block (3-3), which manages access in key events within the region. We thus name this network position the ‘Gatekeepers’, while its peripheries are named the ‘Guests’. 6.3. The space of network positions, field positions and position-takings Fig. 7 shows the spaces of field positions and position-takings [85_TD$IF]defined, respectively, by the attributes of companies and co-productions, along with the companies’ network positions, on the same factorial plane (the global MFA).11 The coordinates of network positions and attributes result from the mean of the coordinates of companies and co-productions belonging to such positions and holding those attributes. In addition, network positions are denoted by the labels we gave them in [87_TD$IF]Section 6.2.2. The horizontal axis distinguishes the prestige and cultural vocation of producers and productions, i.e. it is the axis of symbolic capital. Differently from Fig. 3, the x-axis is reversed in order to put the positive pole of symbolic capital on the lefthand side of the map. This graphic solution emphasizes the relative “field autonomy” (Bourdieu, [8_TD$IF]1983, p. 319) of the ‘Gatekeepers’ and the ‘Insiders’, in contrast to the “heteronomous” principle that orients the activity of the ‘Outsiders A’. In fact, classic and contemporary plays, but also youth theatre and new drama, mean the propensity for more serious themes and elaborate styles and languages, or for educationally-driven pieces, which relates to a (partial) autonomy from the laws of the market. Clearly, these genres diverge from comedy and cabaret or music, which mainly focus on pure entertainment. The same holds for the opposition between producing houses, along with cultural institutions, and building-based companies: the former pertain to the cultural establishment, with the entitlement to give value to their products or services; the latter are more inclined to market their artistic outcome irrespective of its recognition. This also concerns the “degree of specific consecration” and the state’s meta-capital (Bourdieu, [89_TD$IF]1983, p. 320; Bourdieu & Wacquant, 1992) guaranteed to the network positions on the left-hand side of the map although this also means that such positions are not completely autonomous from the field of power. Conversely, the ‘Outsiders B’, despite their work in the segment of educational and youth theatre, are far less recognized in this regard. All these distinctions also matter for the segmentation of the network structure discussed

11 Gerhards and Anheier (1989) and Anheier et al. (1995) provide a similar joint visualization and analysis of blocks and attributes placed in the factorial [86_TD$IF] plane, albeit their usage of correspondence analysis and blockmodeling is different from ours.

Please cite this article in press as: M. Serino, et al., Bridging social network analysis and field theory through multidimensional data analysis: The case of the theatrical field, Poetics (2017), http://dx.doi.org/10.1016/j.poetic.2016.12.002

G Model POETIC 1261 No. of Pages 15

M. Serino et al. / Poetics xxx (2016) xxx–xxx

13

in [90_TD$IF]Section 6.2.2, as the ‘Gatekeepers’ and the ‘Insiders’ can represent the ‘high culture’ segments opposite to the ‘low culture’ segments occupied by the ‘Outsiders’. It is also worth noting that the location of the ‘Gatekeepers’ is subject to a significant change in the 2014/2015 season. Here, CampaniaFestival loses its prominent role of gatekeeper, perhaps because the social capital held by its former director is no longer available for that institution: Well, in a joking manner, I would say that here the festival [CampaniaFestival] loses its centrality simply because I no longer manage it! [10_TD$IF](Artistic director of StabileNapoli)

This social resource of which that director spoke of in the interview remained in the hands of the holder of a position of power into the field. The fact that he is still the director of StabileNapoli might be one of the reasons why the network position of the ‘Gatekeepers’ maintains at least in part its leading role. The vertical axis is, instead, the axis of economic capital, for it puts large venues in contrast with very small, small and medium ones. The same holds for historic theatres vs. non-theatrical buildings, the latter being less expensive than the former. Indeed, a symbolic recognition of economic capital has to be seen in the availability of historic theatres (see Section [92_TD$IF] 4), but also in the “economic conditions for the indifference to economy” that producing houses gain from public subsidies [93_TD$IF] and that permit them to pursue even “the riskiest positions in the intellectual and artistic avant-garde” (Bourdieu, 1983, p. 321). Therefore, the vertical dimension reveals another, more subtle distinction between the consecrated theatrical culture, or ‘orthodoxy’, on the upper side of the map, characterized by classic and contemporary drama or by the revision (or the nonrevision) of secure, well-known plays, and the new wave of theatre making that proposes original or new drama plays, i.e. the ‘heresy’, on the lower side of the figure. Here one may recall Bourdieu’s assertion: “The history of the field arises from the struggle between the established figures and the young challengers” (Bourdieu, 1983, p. 339), this struggle being inevitably a symbolic struggle. Furthermore, it is worth noting, at the bottom-left of the map, the combination of scarce economic capital and high symbolic cultural capital, this latter owing its recognition from the unconventionality of new or unpopular works which denote the so-called “art for art’s sake” (Bourdieu, 1983, p. 321). This is especially concerned with most of the companies in the ‘Fringe’ network position, which lies very close to the axes origin and cannot be seen in the map. A seemingly odd location is that of experimental theatre, which stands in the establishment area, but this is due to a connection with the mainstream sector of producing houses and festivals, which are committed to include this kind of productions in their programmes. Reflecting such location, the ‘Guests’, which are more clearly visible in the map for the 2012/2013 season, are those peripheral companies which are linked to the ‘Gatekeepers’, in line with the hierarchical structure revealed by our analyses. 7. Discussion and [94_TD$IF]conclusions In this work we proposed a joint approach combining, in a novel way, blockmodeling and multidimensional [5_TD$IF]data analysis, in order to investigate the theatrical field by conceiving of it as an affiliation network made of companies involved in coproductions. We have conducted our analyses advocating the application of SNA in the framework of Bourdieu’s field theory, despite the sharp differences between the two approaches. Considering the huge debate regarding Bourdieu’s work, we relied upon theoretical criticism about it and other empirical endeavours similar to ours, so as to sustain the rationale of the present work but also to take a step upward with respect to past research. We adopted the positional approach of SNA through MFA, in such a way that it classifies network units by their location in a relational structure, and also by their attributes, which denote various forms of capital. These latter, according to Bourdieu, define an ‘objective’ social space which, in our view, lays upon a basic stratum of relations among network units. We thus contend that SNA reveals a relational structure that sustains the objective one: the former stands ‘under’ the latter and furnishes a sort of frame to it. Therefore, network structure and objective structure parallel each other and, by MFA, are kept together in a unique framework. In more technical terms, the map of the attributes is directly linked to the relational structure, in that the location of a given attribute is computed on the basis of the mean of the location of the companies or coproductions having that specific attribute in the relational space. However, this does not mean that a causal nexus links the objective and network structures. Although the former results at least formally from analysing the latter which seems exactly opposite to Bourdieu’s view (cf. Bourdieu, 1989) the one does not determine the other, or vice versa. Instead, a dialectical relationship and a mutual implication link the two structures, as the distribution of capital and the formation of network ties operate simultaneously and imply the existence of one another. In more substantive terms, two companies join each other because of their relative field position, i.e. their respective location within the structure of the distribution of capital, but in this way they also reinforce their own field positions. This is the reason why, as we argue, a joint analysis of objective and relational structures may improve the study of cultural fields. On the one hand, by focusing solely on the objective space, information regarding the affiliation of theatre producers on the basis of shared goals or common artistic interests the homophily principle [95_TD$IF]implicit in Bourdieu’s approach (Bottero, 2009) is missed. Further, such a one-sided analysis would not account for the social and symbolic struggles among producers in the space of available memberships in co-productions: “Because symbolic capital is connected to groups or to the names of groups, alliances with groups through social ties are effective weapons in this battle” (De Nooy, 2003,

Please cite this article in press as: M. Serino, et al., Bridging social network analysis and field theory through multidimensional data analysis: The case of the theatrical field, Poetics (2017), http://dx.doi.org/10.1016/j.poetic.2016.12.002

G Model POETIC 1261 No. of Pages 15

14

M. Serino et al. / Poetics xxx (2016) xxx–xxx

p. 321–322). On the other hand, analysing the relational structure of the co-production network alone would deprive the study of any insight into the oppositions between network positions within the spectrum of opportunities available in the field, namely the evidence that a network position is incompatible with other ones that occupy the opposite area of the social space. In fact, the insertion of the network positions of companies into the final map of Fig. 7 helped us show these oppositions with respect to the distribution of capital within the field, and vice versa. After all, participating in co-productions is not a neutral enterprise: it implies definite conditions and motivations for choosing (or being chosen by) specific co-producers, which have consequences for organizations’ positioning in the field. For this reason we considered co-productions as ‘manifestations’ or ‘position-takings’ of a given company (Bourdieu, 1983). The hypotheses presented in [96_TD$IF]Section 3 concern the structural patterns related to such position-taking efforts. The analysis reveals a combination of hierarchy and segmentation in the network structure of the field, meaning that the latter exhibits social and cultural differentiation and dominance. This result seems related to the predominance of symbolic cultural capital and of economic capital taking a symbolic form, in line with Anheier et al. (1995). Our findings thus expand on field theory as they help understand the symbolic struggles among producers from a relational point of view. [97_TD$IF]Although our work is akin to other previous studies, we follow a truly different strategy. First, we apply a field-theoretic view directly to the patterns of relations among producers. Second, focusing on an affiliation network of producers involved in co-productions, without converting it into a one-mode network, permits us to preserve the duality of such network (Breiger, 1974) and to consider the interplay between this duality and the corresponding duality of positions and positiontakings typical of Bourdieu’s cultural fields. The present study is possibly biased because we disregarded individually produced plays, which would perhaps be indispensable in an orthodox field analysis performed by a conventional MCA or MFA applied to companies’ attributes, including the characteristics of their plays. Nonetheless, this choice directly relates to our focus on a social space made of relations of affiliation into theatre projects, whereas using the productions of single companies would make our own approach diametrically different. Future research should pay attention to the micro-level of the interpersonal ties among theatre workers in order to highlight the role of their social capital and habitus in the affiliation in co-productions. [98_TD$IF]Appendices A and B. Supplementary content Supplementary content associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j. poetic.2016.12.002. References Anheier, H. K., Gerhards, J., & Romo, F. P. (1995). Forms of capital and social structure in cultural fields: Examining Bourdieu’s social topography. American Journal of Sociology, 100(4), 859–903. Batagelj, V., & Mrvar, A. (1998). Pajek program for large network analysis. Connections, 21(2), 47–57. Becker, H. S. (1982). Art worlds. Berkeley and Los Angeles: University of California Press. Blasius, J., & Greenacre, M. (2006). Correspondence analysis and related methods in practice. In M. Greenacre, & J. Blasius (Eds.), Multiple correspondence analysis and related methods (pp. 4–40).London: Chapman & Hall/CRC Press. Borgatti, S. P., & Everett, M. G. (1992). Regular blockmodels of multiway multimode matrices. Social Networks, 14, 91–120. Borgatti, S. P. (2002). NetDraw: Graph visualization software. Harvard: Analytic Technologies. Bottero, W., & Crossley, N. (2011). Worlds, fields and networks: Becker, Bourdieu and the structures of social relations. Cultural Sociology, 5(1), 99–119. Bottero, W. (2009). Relationality and social interaction. The British Journal of Sociology, 60(2), 399–420. Bourdieu, P., & Wacquant, L. (1992). An invitation to reflexive sociology. Oxford: Polity Press. Bourdieu, P. (1983). The field of cultural production, or: The economic world reversed. Poetics, 12(4), 311–356. Bourdieu, P. (1984). Distinction. Cambridge: Harvard University Press. Bourdieu, P. (1986). The forms of capital. In J. G. Richardson (Ed.), Handbook of theory and research for the sociology of education (pp. 241–258).New York: Greenwood Press. Bourdieu, P. (1988). Homo academicus. Stanford: Stanford University Press. Bourdieu, P. (1989). Social space and symbolic power. Sociological Theory, 7(1), 14–25. Bourdieu, P. (1993). The field of cultural production. Cambridge: Polity Press. Breiger, R. L. (1974). The duality of persons and groups. Social Forces, 53(2), 181–190. Breiger, R. L. (2000). A tool kit for practice theory. Poetics, 27(2), 91–115. Burt, R. S. (1976). Positions in networks. Social Forces, 55, 93–122. Burt, R. S. (1978). Cohesion versus structural equivalence as a basis for network subgroups. Sociological Methods & Research, 7(2), 189–212. Burt, R. S. (1980). Models of network structure. Annual Review of Sociology, 6, 79–141. Crossley, N. (2009). The man whose web expanded: Network dynamics in Manchester’s post/punk music scene 1976–1980. Poetics, 37(1), 24–49. D’Esposito, M. R., De Stefano, D., & Ragozini, G. (2014). On the use of Multiple Correspondence Analysis to visually explore affiliation networks. Social Networks, 38, 28–40. De Nooy, W. (2002). The dynamics of artistic prestige. Poetics, 30, 147–167. De Nooy, W. (2003). Fields and networks: Correspondence analysis and social network analysis in the framework of field theory. Poetics, 31, 305–327. DiMaggio, P. J. (1986). Structural analysis of organizational fields: A blockmodel approach. Research in Organizational Behavior, 8, 335–370. Doreian, P., Batagelj, V., & Ferligoj, A. (2004). Generalized blockmodeling of two-mode network data. Social Networks, 26(1), 29–53. Doreian, P., Batagelj, V., & Ferligoj, A. (2005). Generalized blockmodeling. Cambridge: Cambridge University Press. Emirbayer, M., & Johnson, V. (2008). Bourdieu and organizational analysis. Theory and Society, 37(1), 1–44. Escofier, B., & Pagès, J. (1998). Analyse factorielles simples et multiples: Objectifs, me´thodes et interpre´tation. Paris: Dunod. Faust, K. (1997). Centrality in affiliation networks. Social Networks, 19(2), 157–191. Faust, K. (2005). Using correspondence analysis for joint displays of affiliation networks. In P. J. Carrington, J. Scott, & S. Wasserman (Eds.), Models and methods in social network analysis (pp. 117–147).Cambridge: Cambridge University Press. Gallina, M. (2007). Organizzare teatro. Produzione, distribuzione, gestione nel sistema italiano. Milan: FrancoAngeli. Gerhards, J., & Anheier, H. (1989). The literary field: An empirical investigation of Bourdieu’s sociology of art. International Sociology, 4, 131–146.

Please cite this article in press as: M. Serino, et al., Bridging social network analysis and field theory through multidimensional data analysis: The case of the theatrical field, Poetics (2017), http://dx.doi.org/10.1016/j.poetic.2016.12.002

G Model POETIC 1261 No. of Pages 15

M. Serino et al. / Poetics xxx (2016) xxx–xxx

15

Giuffre, K. A. (1999). Sandpiles of opportunity: Success in the art world. Social Forces, 77, 815–832. Giuffre, K. A. (2001). Mental maps: Social networks and the language of critical reviews. Sociological Inquiry, 71, 381–393. Lorrain, F., & White, H. C. (1971). Structural equivalence of individuals in social networks. Journal of Mathematical Sociology, 1, 49–80. Mohr, J. W. (2013). Bourdieu’s relational method in theory and in practice: From fields and capitals to networks and institutions (and back again). In F. Dépelteau, & C. Powell (Eds.), Applying relational sociology (pp. 101–135).New York: Palgrave Macmillan. Prota, L., & Doreian, P. (2016). Finding roles in sparse economic hierarchies: Going beyond regular equivalence. Social Networks, 45, 1–17. Ragozini, G., De Stefano, D., & D’Esposito, M. R. (2015). Multiple factor analysis for time-varying two-mode networks. Network Science, 3(1), 18–36. Sailer, L. D. (1978). Structural equivalence: Meaning and definition: Computation and application. Social Networks, 1, 73–90. Scott, J. (2000). Social network analysis. A handbook, 2nd ed. London: Sage. Serino, M. (2013). Theatre provision and decentralization in a region of Southern Italy. New Theatre Quarterly, 29(1), 61–75. Serino, M. (2015). Reti di collaborazione tra teatri di produzione in Campania. Sociologia [10_TD$IF]del Lavoro, 138, 121–137. Sonnett, J. (2016). Ambivalence, indifference: Distinction: A comparative netfield analysis of implicit musical boundaries. Poetics, 54, 38–53. Wasserman, S., & Faust, K. (1994). Social network analysis. New York: Cambridge University Press. White, D. R., & Reitz, K. P. (1983). Graph and semigroup homomorphisms on networks of relations. Social Networks, 5, 193–234. White, H. C., Boorman, S. A., & Breiger, R. L. (1976). Social structure from multiple networks. I. Blockmodels of roles and positions. American Journal of Sociology, 81(4), 730–780. Marco Serino was awarded his PhD in Sociology, Social Analysis and Public Policies by the University of Salerno, Italy, in 2010. He is now participating in the research activities of the Department of Political Science [10_TD$IF]at the University of Naples Federico II. His current research interests are concerned mainly with the sociology of culture and the arts,[102_TD$IF] and social network [103_TD$IF]analysis.

Daniela D’Ambrosio received her PhD in Statistics from the University of Naples Federico II [104_TD$IF]in 2016. Her main areas of research interest are statistical methods for the social sciences, social network analysis, social policies. In particular, her work is now focused on the encounter between multidimensional data analysis and network analytic techniques.

Giancarlo Ragozini is Associate [105_TD$IF]36Professor of Statistics at the Department of Political Science, University of Naples Federico II. His research interests range from statistical methods for social network analysis to computational statistics and multivariate methods for data analysis. He is member of the Italian Statistical Society, the Classification and Data Analysis Group of the Italian Statistical Society (CLADAG), and the International Network for Social Network Analysis (INSNA).

Please cite this article in press as: M. Serino, et al., Bridging social network analysis and field theory through multidimensional data analysis: The case of the theatrical field, Poetics (2017), http://dx.doi.org/10.1016/j.poetic.2016.12.002