Abstract
Building on recent research emphasizing how legitimacy depends on consensus among audiences about candidates’ characteristics and activities, we examine the relationship between cultural producers’ (candidates) position in the social structure and the consecration of their creative work by relevant audiences. We argue that the outcome of this process of evaluation in any cultural field, whether in art or science, is a function of (1) candidates’ embeddedness within the field, and (2) the type of audience—that is, peers versus critics—evaluating candidates’ work. Specifically, we hypothesize that peers are more likely to favor candidates who are highly embedded in the field, whereas critics will not show such favoritism. We find support for these hypotheses in the context of the Hollywood motion picture industry.
Bourdieu’s (1993) seminal work on cultural fields emphasizes the role of relevant audiences as agents of consecration with the authority to produce symbolic capital. The ability to impose judgments of symbolic legitimacy, or the power to consecrate, in cultural fields allows participants to reproduce their positions—thus influencing the choice of (and return to) different aesthetic strategies (DiMaggio 2011). Because these judgments produce prestige hierarchies and affect field evolution, cultural fields are in a constant state of struggle between established and emerging actors who compete for symbolic distinction based on subjective rules of merit, and the vested interests and social objectives these rules embody. While incumbents work to defend and reproduce their views and impose consensus, challengers try to “break the silence of the doxa and call into question the unproblematic, taken-for-granted world of the dominant groups” (Bourdieu 1993:83). The structural outcomes of this struggle have been variously conceptualized as dichotomies that classify cultural producers into incumbents and dissidents, insiders and outsiders, orthodox and heretics, and core and peripheral players.
Several studies (e.g., Anheier, Gerhards, and Romo 1995; Faulkner 1983; Faulkner and Anderson 1987; Giuffrè 1999) have found empirical support for an oppositional structure permeating cultural fields, in which a relatively small number of established players have the necessary material and political resources to enforce norms and standards for evaluating cultural productions that conform to their specific interests, while a much larger number of peripheral players try to advance alternative views and voice their relevance to the field. Thus, the propensity to produce work that departs from a field’s canons and expectations is not randomly distributed but tends to map onto the field’s social structure: core players are more likely to defend orthodoxy in cultural production because “their symbolic capital was founded on already-established types of cultural production, such as the performance of canonical works” (Kremp 2010:1055). In contrast, peripheral players are more likely to produce work that departs from the field’s canons and expectations. Because cultural fields reproduce the power and privilege of incumbent groups, peripheral players have only limited chances of proving their worth, especially if their products contrast sharply with prevailing expectations (Fligstein and McAdam 2012). What mechanisms can help peripheral players break into the prestige hierarchies of cultural fields? What role do social audiences play in shaping the struggle for legitimacy between core and peripheral players?
We address these questions by proposing a framework based on Bourdieu’s intuition regarding the role of social audiences 1 in shaping the allocation of symbolic capital between core and peripheral members of a cultural field. Our strategy is to focus on two types of audiences that reward cultural producers—peers and critics. These audiences correspond to two distinct selection systems, each emphasizing different criteria to assign cultural legitimacy (Allen and Lincoln 2004; Kersten and Bielby 2012).
In the U.S. feature-film industry, our empirical setting, cultural producers have become increasingly concerned with professional and critical recognition, following the legitimation of film as an art form in the late 1950s and 1960s (Baumann 2001, 2007). After characterizing the particular discourse of value that regulates the practice of consecration within each of these audiences, we investigate how peers and critics differ in the extent to which their symbolic capital allocation—in the form of prizes and awards—is influenced by the socio-topographic location of those whom they evaluate. Our findings reveal that cultural producers who are core members of a field enjoy disproportionately higher esteem from peer audiences. Critics, on the other hand, do not favor core producers in their consecration decisions and may even give some preference to peripheral producers. These findings emphasize the importance of studying how different audience-level selection mechanisms influence the structure of symbolic capital allocation and ultimately affect field-level social structure and change.
The Hollywood feature-film industry provides an ideal context for testing the implications of our theoretical framework. Every year, several organizations of elite peers and critics bestow awards on those judged to have made exemplary cinematic contributions. These awards are important vehicles through which cultural producers may internalize collective social norms about what is sacred and what is profane, and who is an insider and who is an outsider (Rossman, Esparza, and Bonacich 2010; Watson and Anand 2006). In this way, award ceremonies are crucial mechanisms for social stratification because they “construct prestige hierarchies that both enable and constrain actors’ abilities to form relationships with others in a field” (Anand and Watson 2004:76). We exploit these industry features to study the relationship between social structure, audiences, and consecration in cultural fields.
Cultural Fields and the Struggle for Legitimacy
Social structure’s effect on the propensity to engage in innovative rather than conforming behavior has deep sociological roots (Simmel 1971). Marginal actors “with the least opportunity for full participation in the most valued activities of their own society . . . may be stimulated to make new responses which depart from the habitually required” (Coser 1962:179). Several studies by sociologists of science and culture emphasize that peripheral actors are in a unique position to champion dissenting ideas that challenge the natural order (e.g., Cattani and Ferriani 2008; Edge 1977; Kirschbaum and Vasconcelos 2006; Mezias and Mezias 2000; Mulkay 1972; Sgourev 2013; Tuchman 1989). Being less tied to norms that regulate actors in the core, peripheral actors possess what Zuckerman and Merton (1973:518–19) call “focused naïveté,” that is, “a useful ignorance of prevailing assumptions and theories that allows them to attack problems generally regarded as impossible or uninteresting by specialists.” For this reason, they may be more open to lines of inquiry that run counter to received wisdom. The challenge for peripheral players is that the same social position that enables them to depart from prevailing norms may also restrict their access to resources and social contacts that would facilitate the completion and legitimation of their work.
Bourdieu (1993) identified two mechanisms that enable peripheral players to establish their cultural authority. The first is exogenous changes, such as revolutions or other political crises, that open up space for alternative views. The second rests on the availability of a homologous reception space, namely an audience predisposed by its beliefs and tastes to consider the kinds of offers being proposed. Cultural sociologists and institutional theorists have extensively examined the role of exogenous shocks in subverting a field’s dominant logic (e.g., Collins 1987; Zucker 1988). However, the possibility of audience-level variations offering countervailing mechanisms to the social stratification process is understudied.
The understanding of legitimacy as a relationship with an audience rather than a possession of the actor is central to the theory of cultural fields. We take seriously Bourdieu’s (1993:95–96) assertion that “all the homologies which guarantee a receptive audience and sympathetic critics for producers who have found their place in the structure work in the opposite way for those who have strayed from their natural site.” The existence of a homologous audience is a critical precondition for peripheral cultural producers to climb the reputational ladder. Contributions that conflict with the dominant view “cannot be understood sociologically unless one takes account of the homology between the dominated position of the producers . . . and the position in social space of those agents . . . who can divert their accumulated cultural capital so as to offer to the dominated the means of objectively constituting their view of the world” (Bourdieu 1985:736).
Peripheral field members can benefit greatly from the presence of a homologous audience, one whose views, beliefs, and tastes are attuned to their own. Anand and Watson (2004) provide a powerful illustration in their study of the Grammy awards given by the National Academy of Recording Arts and Sciences. The Academy initially opposed recognition of progressive genres such as rock-and-roll and rap. But intergenerational changes in Academy demographics, particularly the entry of new members whose tastes and values were more attuned to progressive musicians, proved crucial in gaining legitimacy for peripheral actors. We explore and extend these ideas by examining a case in which the observed variation is not within—as in Anand and Watson’s account—but across different audiences. Specifically, we examine two audiences—peers and critics—that have quite different degrees of homology with respect to the producers they evaluate.
The Social Structure of Consecration in Hollywood
Choice of Hollywood as Field
We believe the Hollywood feature-film industry is an ideal setting for our analysis. This industry embodies key aspects of our theoretical framework: cultural producers are entangled in a “competitive struggle for advantage and advancement” (Faulkner 1983:69). Like many cultural fields, Hollywood is dominated by an elite, with only a small fraction of cultural producers successfully climbing the reputational ladder (Faulkner and Anderson 1987). Producers face severe disparities in control over resources, together with a “very large pool of peripheral participants who vie to establish a foothold in the industry” (Zuckerman et al. 2003:1036). The mechanisms by which marginal players acquire symbolic capital are therefore of particular interest.
Field players (e.g., producers and distributors) tend to invest in past successes and avoid recruiting professionals who do not have established reputations or projects that do not fit recognized categories (Zuckerman and Kim 2003). This pattern of investment yields a continually reproduced system of inequality whereby a small number of individuals and organizations are very productive, and often very conspicuous in terms of their connections and rewards, while a much larger group remains on the margins of the network (Faulkner 1983; Giuffrè 1999). Professionals advance their careers by accruing credits and connections that allow them to move from the margins of a network to the center. Projects (here movies) are the material means for making social announcements about one’s ability and identity.
In network analytic terms, the social structure of the feature-film industry is dominated by a core-periphery split in which “the inner core of actors tends to be a small world of tightly knit individuals, whereas the periphery is more open” (Peterson and Anand 2004:322), allowing individuals with skills and resources to gain an entry point into the system. This partition reflects a relentless pressure to become visible and “push one’s image, voice and credentials deeper into the industry’s culture” (Faulkner 1983:22). Yet, as Faulkner’s (1983) work on Hollywood composers clearly shows, the odds of a movie professional moving from the periphery to the core are extremely low. Moreover, admission into the core comes with liabilities in the form of tighter restrictions on productions that challenge recognized standards and categories. At times, such restrictions are so strong that some cultural producers may perceive them as onerous, perhaps even contrary to their self-understanding as artists (Anheier et al. 1995).
A second motivation for studying the feature-film industry is the opportunity to examine how different audiences’ evaluation criteria shape a professional’s chances of being culturally consecrated. Starting in the late 1950s and early 1960s, the transition from the studio system to director-centered production led to the legitimation of film as an art form in the United States. This was followed by the proliferation of film festivals and ceremonies intended to celebrate cinematic achievement through special prizes and awards (Baumann 2001). Such honors are bestowed by others directly involved in the film industry (i.e., peer professionals) and by movie critics. Results of these evaluations are made public every year by conferring prizes that celebrate exemplary achievements, thereby establishing a level of recognition in the field unattainable through other means. These accolades and rituals constitute “the symbolic capital that may serve as an alternative to economic capital” (Baumann 2001:406) and signal that film production has achieved a certain autonomy as a field.
Peers, Critics, and the Allocation of Symbolic Rewards
Cultural legitimation involves the use of aesthetic judgments to assign value to cultural producers and their products. Bourdieu (1993) identified three types of cultural legitimacy: public acclaim (popular legitimacy), professional recognition from peers (specific legitimacy), and critical evaluation (bourgeois legitimacy). 2 Different audiences typically employ different criteria. Here we focus on the selective judgments that express two distinct aesthetic logics embedded in the world of professional criticism and the world of film practice. 3 As Allen and Lincoln (2004:877) note, “the existence of competing discourses of value may explain why films that receive professional recognition from other cultural producers are not always the same films as those that receive critical recognition from critics.” Similarly, Kersten and Bielby (2012) suggest that critics’ evaluations vary systematically on several aesthetic dimensions across films ultimately consecrated by peers or critics. 4
In the U.S. feature-film industry, peers and critics are organized into distinct awarding organizations that reveal their preferences in annual ceremonies that play a critical role in constructing prestige hierarchies ( Gemser, Leenders, and Wijnberg 2008; O’Neil 2003). Like tournament rituals (Anand and Watson 2004), these ceremonies not only offer established professionals an opportunity to reinforce their authority but are also “the arena within which non-elites can express alternative and non-official attitudes and values” (Lukes 1975:299). Using information on professional societies’ membership, we classified Hollywood’s main awarding organizations into peer- and critic-based groups. Peer-based awarding organizations include the Academy of Motion Picture Arts and Sciences and various professional guilds (e.g., Directors Guild of America and Screen Actors Guild). Critic-based awarding organizations include the New York Film Critics Circle, the National Society of Film Critics, the Los Angeles Film Critics, and the Boston Society of Film Critics (see Table 1). These organizations have been in existence for several years, are widely regarded as reliable and competent, and grant annual accolades in most major categories of filmmaking expertise (Levy 2003; Simonton 2004).
List of Awarding Organizations (Audiences)
Peer Recognition
Peer audiences are members of the same community as the producers they evaluate, although they take on different roles (Debackere et al. 1994; Wijnberg 1995). This observation holds important implications for understanding how they evaluate cultural producers. First, peer evaluators are typically elite representatives of a field’s dominant canons. As such, they have the authority to determine the legitimate definition of a given type of work and, by extension, the authority to define which works configure the field’s canon (Bourdieu 1993). Accordingly, they tend to define excellence as “what is most like me” (Lamont 2009) and provide a disproportionate amount of material and symbolic resources to core members of the field who are more strongly associated with its dominant canons.
Second, evaluators and producers located in the core of a field’s social structure are more likely to share similar cognitive and social networks, resulting in a potential bias toward work emanating from the core. Indeed, producers can impress prominent peers if they share “with those whose judgment matters most a common repertoire of beliefs, assumptions, [and] background knowledge” (Gross 2000:856). In academic evaluation systems, for example, evaluators are usually established and highly embedded scholars who “are frequently asked to adjudicate the work of individuals with whom they have only a few degrees of separation,” and they “often favor their own type of research while being firmly committed to rewarding the strongest proposal” (Lamont 2009:8). Third, elite peers are prone to resist work that deviates from established normative expectations because they have a vested interest in guaranteeing “the continued reproduction of the legitimacy of those who produce or defend the canon” (Bourdieu 1993:20). As Peterson (1979:157) noted, when the evaluation criteria “are tightly held by an academy of peers, there is an orientation to traditional canon of arts (or religion or science).”
These characteristics of peer audiences are well exemplified by the Academy of Motion Picture Arts and Sciences, both in the process by which it accepts new members and in the decisions it makes about nominations and awards. These decisions tend to reflect the aesthetic preferences of the elite group in each branch—namely those who are more attuned or homologous to the core players in Hollywood. Since its founding in 1927, the Academy has been an association of Hollywood’s creative elite. New members are invited to join “when their services to the motion picture industry have been prominent enough to make the Academy members feel they would like to have them as brother members” (Levy 2003:46). Here, for example, are the official requirements for Directors Branch membership:
Membership shall be by invitation of the Board of Governors. Invitations to active membership shall be limited to those persons employed by motion picture producing companies, or credited with screen achievements, or who have otherwise achieved distinction in their respective fields of endeavor within the industry and who, in the opinion of the Board, are qualified for membership. (http://www.oscars.org/academy/members/requirements/producers.html)
The easiest way to become a member, however, is to get an award nomination—all Oscar nominees are invited to become members. Beyond that, each branch has its own criteria. In most branches, nominees must have several film credits, several years of experience, and sponsorship by two established members. Branch membership determines who has the right to vote regarding awards and nominations; members vote on nominees in their respective categories (e.g., directors for directors, actors for actors, and writers for writers). All voting members, however, are eligible to select the Best Picture nominees. Except for the Actors Branch, the relatively smaller size of the other branches means that relatively few voters determine a nominee. For instance, in the Directors Branch, which in 2002 consisted of 364 members (the number has not changed much over the years), a group of 20 or 30 members could nominate a director. Indeed, the Directors Branch is “not only small but cliquish as well” (Levy 2003:49).
Life membership and the relatively high average age of voting members may also affect the Academy’s decisions. The age difference between older members and those actively involved in filmmaking inevitably makes “the Academy vote more conservative, lagging behind the industry’s aesthetic and technical innovations” (Levy 2003:48). Indeed, the Academy vote is often criticized for displaying a conservative bias. Best Picture winners, for example, are often described as “soft, noble, middlebrow movies that reflect the dominant culture, steering clear of provocative issues or innovative experimental styles” (Levy 2003:48).
The professional guilds are much larger than the Academy branches. For instance, in 2002 the Academy’s Directors Branch consisted of 364 members, whereas the Directors Guild of America (DGA) had 12,400 members, including, in addition to filmmakers, TV directors, associate directors, stage managers, and unit production managers. Their larger membership base might suggest that guild decisions are less cliquish than the Academy’s. Yet guilds’ award decisions do not support this suspicion. For example, according to O’Neil (2003:797), the DGA award is considered “Hollywood’s most influential prize after the Oscar—for two reasons: directors are the reigning kings and queens of pix, so it matters a lot what their peers think, and second, Oscar voters defer most often to the DGA’s choices before inking their ballots.” From 1947 to 2004 (which includes our study period), the Oscars have embraced 51 of the DGA’s winners in the Best Director category, an agreement rate of 89 percent.
Decisions by the Screen Actors Guild (SAG) are also quite similar to those of the Academy. From 1995 (when the SAG awards were first introduced) until 2004, 27 of the 44 actors/actresses nominated by SAG were also nominated by the Academy. Similar patterns hold for the other professional guilds. Thus, despite having thousands of additional members, guilds and Academy voters “are obviously like-minded” (O’Neil 2003:794).
Critic Recognition
Several considerations suggest that critics’ evaluations of cultural producers do not mirror those of peers. Due to their specialized training and high levels of cultural capital, critics cultivate aesthetic dispositions that raise their legitimacy as arbiters of taste and agents of consecration (Janssen 1997; Van Rees 1987). They also have significantly more incentive than peers to discover new talents with the potential to rise to fame (Bourdieu 1984). Indeed, a critic is motivated to be “the first to come up with an assessment that other people . . . might somehow use as a peg for their own . . . response” (Van Rees 1987:286). In fact, it may be dangerous for critics not to embrace a new style, as they risk losing reputation if that style becomes popular. In the visual arts, for instance, critics played a crucial role in fostering the rise of “radical” painters who became known as the Impressionists. Systematically rejected by the Salon de Paris (the bastion of elite peers within the Académie des Beaux-Arts), which favored established artists, the Impressionists found a homologous reception space in the world of critics and painting experts, who sometimes openly proclaimed their disdain for the Académie’s judgments. These critics did not have a vested interest in perpetuating the status quo because they “derived their legitimacy and importance from being among the first to recognize the value of new entrants into the visual arts industry” (Wijnberg and Gemser 2000:324).
A similar contrast between peers and critics seems to characterize the film industry. For the award of Best Director, the National Society of Film Critics agreed with the Oscars in only three instances during our study period, and agreed only once with the DGA. The New York Film Critics agreed three times with the Oscars and two times with the DGA. The Los Angeles Film Critics agreed with the Oscars in five cases; the Boston Society of Film Critics agreed two times with both the Oscars and the DGA.
Critics have what Simmel (1971) called the objectivity of the stranger—he who is not bound by stable social ties to other group members. Although both critics and cultural producers are members of the same field, they are not embedded in the same professional community. In principle, this should promote an unprejudiced perception, understanding, and assessment of candidates’ work, allowing critics to make evaluations with more objectivity. Lacking an incentive to prefer one particular style over another, they may be more willing to support and consecrate offers that depart from the field’s dominant norms and standards—offers that typically (although not exclusively) emanate from a field’s margins (Collins 1987).
Consider the impact of critics on the career of iconic filmmaker Quentin Tarantino. As a former employee of a video rental store in Los Angeles and with no formal filmmaking education, Tarantino had practically no markers of credibility and was truly peripheral when he directed his 1992 debut feature Reservoir Dogs. First screened at the Sundance Film Festival, Reservoir Dogs became the festival’s most talked-about movie, and Miramax decided to distribute it. Over the course of that year, Tarantino “turned up at festival after festival, receiving lavish praise and awards from intellectual critics for making the hottest indie of the year” (Levy 1999:15). In just a few years, Tarantino rose from obscurity to fame, and the fact that “the film didn’t do well didn’t matter. It created enough of a stir to give Tarantino the clout to make his next film, Pulp Fiction, with a larger budget ($8 million) and a high-caliber cast” (Levy 1999:17). Notably, Reservoir Dogs received no recognition from the Academy, the guilds, nor any other purely peer-based audience. It might be tempting to regard Tarantino’s case as unique; in fact, it is highly instructive of how peripheral players can succeed by eliciting support from one homologous audience within a cultural field.
These arguments suggest not only that symbolic rewards for cultural producers are socially structured, but also that the saliency of this structure may vary with the type of audience bestowing the awards. One would expect peers to be more likely to consecrate core cultural producers as opposed to peripheral producers. We do not expect the same relationship for critics, however. Critics may have an incentive to consecrate the work of peripheral producers, or at least to not discriminate between core and peripheral producers. By providing peripheral producers with a more receptive space for their offers, critics may alter the stratification resulting from peers’ choices. Thus, we make two major predictions regarding award decisions by these audiences:
When peer organizations bestow accolades, they are likely to favor members in the core of a cultural field over those in the periphery, holding other factors constant.
When critic organizations bestow accolades, they are not likely to favor core members of a cultural field over those in the periphery, and they may actually favor those in the periphery.
Analytic Strategy
Dataset
Our data consist of the population of crew and cast members (hereafter “professionals”) who worked on at least one of the 2,297 movies distributed in the United States by the eight major studios—the seven historical majors plus Dreamworks (founded in 1994)—and their various subsidiaries over the period 1992 to 2004. Because we are interested in films made and distributed by Hollywood, we did not include documentaries, foreign-made films, short films, and compilations. Focusing on the major studios might suggest neglect of artistically oriented movies in favor of commercial ones, but these companies have numerous divisions (often carried over from previous acquisitions) that specialize in different types of films (see Table 2).
Hollywood Studios and Their Distribution Divisions
Note: Studios purchased distribution divisions at different times. Many of these divisions changed ownership during the study period (e.g., Focus Features is the art house films division of Universal Studios and originated from the 2002 divisional merger of USA Films, Universal Focus, and Good Machine; October was purchased by Universal in 1997; Screen Gems became a specialty film-producing arm of the Sony group in 1999; Samuel Goldwyn was purchased by MGM in 1997; and Castle Rock and New Line were purchased by Warner in 1996). In attributing film releases to major studios we accounted for the timing of all such transactions.
Some divisions specialize in small-budget niche films (they often have adjectives such as “repertory,” “independent,” or “classic” in their names) and are widely acknowledged as focusing on more artistically oriented movies and less visible talents. Thus, although we do not have data on the entire population of films, we believe the risk of underrepresenting the periphery of the Hollywood field is low. We gathered information on the composition of each movie’s production team and the consecration of its members’ work by recording awards and nominations (hereafter “accolades”) received. We focused on the following professionals: director, writer, leading and supporting actor/actress, editor, cinematographer, and production designer. Using the Internet Movie Database (IMDb), we identified more than 12,000 professionals. We also crosschecked this information with the Alan Goble Film Index (Goble 2003) to ensure data quality. Although we included producers in analysis of the Hollywood social network, we excluded them from the analysis of accolades because there is no specific award assigned to producers. The award for best movie goes to the movie, a collaborative venture, even though the award is handed to the producer.
Dependent Variable
We used a discrete-choice approach to model the audience-candidate evaluation process, in which peer- and critic-based organizations select candidates whose work is consecrated with an award or a nomination, from among the large set of candidates eligible in a given year. This approach seems particularly appropriate given that consecration, by its very nature, imposes discrete (rather than continuous) distinctions between candidates who deserve recognition and those who do not (Allen and Lincoln 2004). In our context, a decision consists of a choice by a particular organization, in a particular year, to award an accolade (award or nomination) to a particular professional for performance in a particular role. Each decision has a choice set that includes all the professionals who were potentially eligible for the accolade. For each member of the choice set, the dependent variable is coded 1 if a person received an award or nomination, and 0 otherwise. We estimated separate discrete-choice models for peer organizations and critic organizations.
Independent Variables
To distinguish between core and peripheral producers, we adopted a network analytic approach in line with Faulkner (1983), Faulkner and Anderson (1987), and Anheier and colleagues (1995). We first identified the social network of interactions between Hollywood professionals from the bipartite affiliation structure between professionals and movies. This means we presumed professionals working on the same movie had a tie to one another (see the Appendix for details).
In an idealized core-periphery structure, the core is a group of nodes connected to all other nodes of both the core and the periphery. The periphery is a group of nodes not connected to each other but only to the nodes in the core. Although no real social network conforms to this ideal, an algorithm is used to maximize the density within the core and to minimize the density within the periphery. Borgatti and Everett (1999) proposed the following method for partitioning a set of nodes into a core and a periphery. Let αij be a (0, 1) variable indicating the presence of a tie between node i and node j. Define δij to have a value of 1 if both i and j are in the core, 0 if both i and j are in the periphery, and “missing” otherwise. Then choose a partition to maximize the Pearson correlation between α and δ, where the correlation is computed over all non-missing pairs. This can be accomplished with a genetic algorithm.
Using the UCINET VI package (Borgatti, Everett, and Freeman 2002), we constructed the binary variable periphery, which takes the value 1 when individuals are in the periphery of the Hollywood network and 0 for those in the core. 5 This variable was based on network ties in a three-year moving window, but results did not vary much using different time windows.
The set of professionals in the core varies from year to year. Over the study period, approximately 346 different professionals (about 3 percent of the total) were in the core in at least one year. Professionals may stay in the core or the periphery for the entire period, or move from the periphery to the core (e.g., Billy Bob Thornton after All the Pretty Horses in 2000) or vice versa (e.g., Mira Sorvino after Mimic in 1997). The small size of the core reflects the unequal distribution of ties in the movie industry’s social structure. As Faulkner and Anderson (1987) observed, “the film community like most culture industry systems and like most high-performance systems is dominated by an active elite and manifests inequality in productivity, [and] cumulative resources in the form of ties.”
Control Variables
To rule out alternative explanations for our hypothesized relationships, we included several control variables in our models.
Role
As noted earlier, the analysis focuses on a restricted group of professional roles. Controlling for role is essential because the choice sets are specific to particular roles. Moreover, different organizations bestow awards for different roles, and the number of these has changed over time in some cases. While the Academy of Motion Picture Arts and Sciences tends to assign awards to all categories, the Los Angeles Film Critics Association assigns no awards for movie editing, and the various guilds give awards only to their members. Individuals performing roles with more award categories have more chances of receiving accolades. We adjusted for these possibilities by including a fixed effect for the role each professional performed in a given movie. This was accomplished either by stratifying on role in conditional logistic regression or by including role as a set of dummy variables.
Degree of control
It is quite possible that individuals performing multiple roles also enjoy more freedom in the pursuit of their goals and are in a better position to express their skills and talents. In most cases there was only one specialist per role, but a professional sometimes performed multiple roles in a single movie (e.g., Clint Eastwood was director, actor, and producer for Unforgiven in 1992), or the same role was collectively performed by multiple individuals (e.g., Joel and Ethan Cohen co-directed Fargo in 1996). We therefore created the variable degree of control to capture the extent to which professionals enjoyed enough latitude to express their creativity by measuring the average number of different roles each professional performed in the same movie in a given year. This variable is similar to what Baker and Faulkner (1991) call “role as resource.”
Commercial success
Participation in commercially successful movies both reveals and shapes perceptions of a professional’s place in the industry hierarchy. Indeed, Hollywood professionals’ bankability translates into stardom. Accordingly, we measured professionals’ commercial success by how well their previous movies fared at the box office. Specifically, we counted the number of top-10 box-office movies in which each professional worked during the three years prior to the focal year (results did not change much with different time windows). We also executed the analyses using the top-20 and top-30 box-office movies, but results were not substantially affected.
Artistic reputation
For most observers, a high number of accolades in an individual’s career would indicate exceptional talent and reputation. Previous research suggests that the most successful movie professionals enjoy preferential access to resources and information (Faulkner and Anderson 1987) and are offered more jobs (Zuckerman et al. 2003), further increasing their chances of receiving an accolade. Accolade recipients also enjoy greater media attention than do their lower-status counterparts. Accordingly, we created the variable artistic reputation, that is, the number of accolades each professional gained during the three years prior to the focal year (again, results did not change much with a different time window).
Team reputation
Movies are made through the complementary inputs of many professionals, whose efforts facilitate the work of other team members. In practice, this suggests professionals should perform better when surrounded by talented collaborators (Rossman et al. 2010). Besides skills, status may also spill over from accomplished team members (Benjamin and Podolny 1999). Thus, a professional can receive greater recognition by collaborating with highly reputed colleagues. We measured a team’s reputation as the average number of accolades team members other than the focal individual received in the three years prior to the focal year. If a professional made more than one movie in a given year, we used the average number of prior accolades for all team members she worked with in that year.
Movie sequel
Movie makers may strive for artistic originality, or they may focus on more formulaic content. Although sequels are sometimes critically praised (e.g., sequels to The Godfather and The Lord of the Rings), they tend to capitalize on a successful formula—which might reduce the likelihood of a professional receiving an accolade. Following previous studies (e.g., Ravid 1999), we created a dummy variable sequel that has the value 1 if a movie is a sequel and 0 otherwise.
Movie rating
Another potential factor affecting the receipt of accolades is the rating assigned by the Motion Picture Association of America (MPAA). Ratings signal the degree of sexually graphic sequences, violence, and strong language in a movie. Prior research suggests that features produced for mature audiences perform less well at the box office (Ravid 1999). Movies rated G, PG, and PG-13 have greater audience potential, and indeed movie theaters’ landlords sometimes contractually prohibit them from showing NC-17 films. As a result, studios quite often exert some pressure on directors (and producers) to ensure films receive a rating aligned with their market aspirations. This practice can obviously constrain creativity (e.g., in the treatment of controversial material or the choice of scenes to edit out of a final print). We accounted for this by including a categorical variable with five categories: G, PG, PG-13, R, and no available rating.
Movie genre
The likelihood of an accolade being bestowed could also depend on movie genre, on the premise that a movie’s artistic content might vary across different genres (e.g., action, drama, comedy, thriller, or animation). For instance, the thriller genre was seen as disreputable until the 1960s when U.S. critics began to take Alfred Hitchcock seriously as an artist (Kapsis 1992). Similarly, a professional working on an action movie is less likely to gain accolades, because action movies typically reflect more formulaic conventions. We created a categorical variable (with 18 categories) to control for each movie’s genre using data from the American Film Institute (AFI).
Number of movies
The chance of receiving an award or nomination also likely depends on the actual number of movies each professional makes in a given year. Because most professionals tend to work on a single movie per year, we created a dummy variable equal to 1 when a professional was involved in at least two movies during the focal year and 0 otherwise.
Awarding organizations
We grouped awarding organizations selected for this analysis into two distinct audiences, peers and critics, on the premise that each audience type tends to apply similar norms and standards when they evaluate and reward individuals’ creative work. To account for the impact of stable, unobserved differences between organizations within the same audience type, we either stratified by awarding organization or included a set of dummy variables in the models.
Year
Because we had no a priori expectations about possible trends over the study period, we controlled for unobserved factors (e.g., macro-economic trends, changes in taste or fashion, and other factors that might affect the movie industry) by stratifying by year or by treating year as a categorical variable.
Model
For a given role in a given year, we modeled the impact of a professional’s characteristics on the probability that an organization bestowed an accolade on that professional rather than any other who was eligible in that year. This can be framed as a series of discrete choices with a professional selected in each category (role) each year from a set of possible candidates. Let yij be equal to 1 if organization i (with i = 1, …, n) chooses professional j (with j = 1,…, Ji), otherwise 0; and let xij be a vector of explanatory variables describing professional j for organization i. The number of possible choices is Ji to indicate that different organizations may have different sets of candidates to choose from. The conditional logit model introduced by McFadden (1973) has the following general form:
where β is a vector of coefficients. This equation implies that the odds that organization i will choose professional j over professional k is given by exp{β(xij − xik)}. That is, it is a function of the difference in the vectors of explanatory variables describing the two professionals. 6 In the analysis, we split the sample into two subsamples, one of peers’ and the other of critics’ organizations (see Table 2). We then stratified by awarding organization, professional’s role, and year. We estimated the conditional logit model by maximum likelihood using PROC LOGISTIC in SAS (release 9.3).
Analysis and Results 7
We began by estimating separate discrete-choice models for peers and critics in which the only predictor variable was periphery, a dummy variable for whether a professional was in the periphery rather than the core of the Hollywood network. As previously noted, these models simultaneously stratified by year, role, and awarding organization, so each stratum corresponded to a choice set for an awarding organization in a particular year. Among peer organizations, the coefficient for periphery was −1.233 (p < .0001), corresponding to an odds ratio of .29. In other words, a professional in the periphery had 71 percent lower odds of being chosen for an accolade than did a professional in the core. For critic organizations, the periphery coefficient was −.508 (p = .03) with an odds ratio of .60. These results suggest that both peers and critics had a preference for members of the core, although the effect size and level of significance is much stronger for peers.
Next we introduced our control variables into the model, with results shown in Table 3. Model 1 displays results for peer organizations. Although we do not report coefficient estimates for movie genre and movie rating, the two sets of dummy variables both had a statistically significant overall impact. The coefficient for movie sequel was significant and in the expected direction: professionals working in sequels were less likely to receive an accolade. Professionals whose team members received accolades in previous years (team reputation) were more likely to receive an accolade, a result consistent with findings from Rossman and colleagues (2010). The quality of each professional’s human capital (artistic reputation) was significant and in the postulated direction. Professionals who previously worked in commercially successful movies (commercial success) were more likely to receive an accolade, whereas those who worked in more than one movie (number of movies) were less likely to receive an accolade. The only variable that was not statistically significant was degree of control, measured as the number of roles a professional performed.
Discrete-Choice Models
Note: Standard errors in parentheses.
p < .05; **p < .01 (two-tailed tests for all variables).
When all these variables were controlled, periphery had a coefficient of −.927, which is somewhat smaller in magnitude than the bivariate coefficient reported earlier but still highly significant (p < .0001). The corresponding odds ratio of .40 indicates that peripheral professionals had 60 percent lower odds of receiving an accolade compared to core professionals.
Results in Model 2 for critics show many similarities with the peer results. Again, the sets of dummies for genre and movie rating were highly significant. Like peers, critics tend to prefer producers with a strong team reputation and a strong personal reputation, although coefficients are somewhat smaller for critics. They also tend to dislike sequels, but the effect is much larger for critics. Unlike peers, critics are not swayed by whether a professional previously worked in commercially successful movies. Another exception is number of movies, which has a positive coefficient for critics and a negative coefficient for peers. The most important difference—and the one that confirms our second hypothesis—is that the coefficient for periphery was positive and far from statistically significant. Moreover, the difference between the periphery coefficient for critics and the coefficient for peers is highly significant (p < .001). Thus, in contrast to peers, we find no evidence that critics’ decisions are influenced by whether a professional is in the core or the periphery.
Unobserved heterogeneity
Although models in Table 3 control for several characteristics of both the professionals and the films in which they appeared, it is possible that unobserved heterogeneity could have biased either the coefficients or their standard errors. For many reasons, some professionals may simply be more attractive to both peers and critics, and the same is likely to be true for films. Due to computational limitations, we were unable to introduce unobserved heterogeneity directly into the discrete-choice models. However, we explored the potential consequences of unobserved heterogeneity by switching to a closely related specification. Instead of stratifying by year, organization, and role, we simply entered these three variables into the logistic regression models as sets of dummy variables. That allowed us to use standard methods for introducing random effects into logistic regression (Allison 2012, Ch. 8). We then estimated one model with a random effect for individual professionals and another model with a random effect for films. We used PROC GLIMMIX in SAS to estimate the models by maximum likelihood. Because the two effects are not nested, it was not computationally feasible to enter both individual and film random effects in the same model. Formally, the model is specified as
where pij is the probability that professional i receives an accolade for film j, xij is a vector of predictor variables describing professional i and film j, and β is a vector of coefficients. The random effect ϵij is assumed to be normally distributed with a mean of 0 and a variance σ2 and is statistically independent of xij. For the model with a random effect for individuals, the j subscript is deleted so that ϵ varies only across individuals. For the model with a random effect for films, the i subscript on ϵ is deleted.
Instead of estimating models separately for peers and critics, we estimated combined models for both audiences. This was essential to allow the unobserved heterogeneity to be common to both critics and peers. To distinguish the two groups, the combined dataset included a dummy variable critic with a value of 1 for critics and 0 for peers. The models also included an interaction between critic and periphery to test whether the effect of periphery varied between the two audiences. We included interactions between critic and three other variables—number of movies, sequel, and commercial success—to accommodate the apparent differences in effects of these variables in Table 3.
To ensure consistency with the discrete-choice models of Table 3, Model 3 in Table 4 shows results from a logistic model without any random effects. Careful inspection of the main effects and interactions reveals that these results are all quite similar to those in Table 3. Most important, the effect of periphery for peers (−.910) in Model 3 (the main effect of periphery) is quite close to the coefficient in Model 1 (−.927). Both are highly significant. The effect of periphery for critics in Model 3 is obtained by adding the main effect to the critic × peer interaction: −.910 + 1.094 = .184, which is not statistically significant. This compares with .089 in Model 2, which is also not statistically significant.
Logistic Regression Models with Random Effects (RE)
Note: Standard errors in parentheses.
p < .05; **p < .01 (two-tailed tests for all variables).
Model 4 of Table 4 introduces a random effect for individual professionals. There is strong evidence for unobserved heterogeneity at the level of professionals: the estimated random-effects variance is 7.822 with a standard error of .601. Results for two of the control variables are markedly different in this model, compared to models in Table 3. Commercial success had significant positive effects in Models 1 and 2 but was not significant in Model 4. Artistic reputation had highly significant positive effects in Models 1 and 2 but a highly significant negative effect in Model 3. Despite these differences, our results for the key variable periphery are essentially similar to those for the discrete-choice models. Periphery had a highly significant negative effect for peers (−1.423) but no detectable effect for critics. We obtained the coefficient for critics by adding the main effect and the interaction: −1.423 + 1.162 = −.261, which is not significantly different from zero. The odds ratio for peers was .24, translating to a 76 percent reduction in the odds of an accolade for members of the periphery compared with members of the core. The interaction between critic and periphery is highly significant, implying that the effects of periphery for critics and peers are significantly different.
Model 5 introduces a random effect for films. As in Model 4, there is strong evidence for unobserved heterogeneity, this time at the film level. In this model, results for the control variables are fairly consistent with those in the discrete-choice models. But the periphery results are somewhat different. Here the magnitude of the periphery effect for peers was much smaller (−.443) although still highly significant. For critics, the periphery effect was −.443 + 1.151 = .708, a highly significant positive effect corresponding to an odds ratio of 2.03. This means that, holding other factors constant, the odds that a critic organization will award an accolade to a periphery member are double the odds for a core member. This is our first evidence that critics may actually favor members of the periphery rather than simply being indifferent to core membership. Again, the highly significant interaction term tells us that periphery coefficients for peers and critics are significantly different.
We also estimated fixed-effects models analogous to the two random-effects models. These models have the attraction of actually controlling for all unchanging characteristics of professionals or films. Results essentially corroborated those in Models 4 and 5. However, we are not reporting the details because large fractions of the professionals or films had to be excluded due to the absence of variation on the dependent variable.
Scope Conditions
We further probed the previous results by estimating additional models to establish the conditions under which the identified patterns are more or less likely to occur. Three scope conditions seem especially important. One possibility is that peers’ preference for core cultural producers is restricted to the realm of elite peers who may have a stronger stake in conformity and reciprocity. Our analysis has so far pooled all peers’ awards irrespective of peers’ actual standing in the awarding body, but one could argue that not all awards are decided by elite peers. The Academy’s awards express the taste and preference of prominent representatives of the various cinematic professions, whereas the guilds encompass a much broader membership with some individuals only marginally attached to any given profession. To explore such potential award-specific effects, we estimated separate discrete-choice models for Academy awards and guild awards. The coefficients of the periphery variable were negative (−.927 in the model for the Academy and −.913 in the model for guilds) and statistically significant in both groups (p < .01), suggesting that peers’ preference for core members holds irrespective of the awarding body. The difference between the periphery coefficients for the Academy and the guilds is not statistically significant. Thus, our results suggest that elite and nonelite peer choices follow a very similar pattern.
Another possible objection is that substantial variation may exist among critics in their openness to peripheral professionals, with prestigious critics paying significantly greater attention to established cultural producers (Janssen 1997). Because of the lack of compelling empirical grounds for accepting or rejecting value judgments, a critic may risk her reputation by expressing a judgment that differs from those of her colleagues (Van Rees 1987). A critic’s reputation as expert is “inseparably related to the extent to which, over a period of time, his judgments have met with his colleagues’ approval” (Van Rees 1987:283). Because they have more to lose, established critics may be less inclined to support deviant offers advanced by peripheral players. For instance, in the field of literary criticism, Janssen (1997) found that more occasional, and therefore less established, reviewers tended to make more deviant choices.
To address this issue, we isolated the award choices of two organizations that stand in sharp contrast to each other with respect to their prestige and influence: the National Society of Film Critics (NSFC) and the Boston Society of Film Critics (BSFC). The NSFC was founded in 1966 as a high-brow association to counter middle-brow film circles, and it consists of 57 of the nation’s most prestigious critics. Every year the NSFC confers awards in 10 categories. Established in the early 1980s, the BSFC is one of the youngest professional associations of film critics, and its membership consists exclusively of Boston-based journalists, primarily freelancers with multiple affiliations and much less visibility than their NSFC counterparts. Every year the BSFC confers awards in 16 categories. To the extent that established critics act more deferentially, we expect critics’ tendency to favor peripheral producers to be stronger within the BSFC than the NSFC. Accordingly, we estimated separate discrete-choice models for the NSFC and BSFC. The coefficient of the periphery variable was positive (1.473) and statistically significant (p < .05) in the model for the BSFC, and positive (1.190) but not significant in the model for the NSFC. The difference between the two coefficients was not statistically significant. While these results suggest that nonelite critics may be more inclined to favor peripheral professionals, we are unable to draw any firm conclusions.
In the preceding analyses, our dependent variable was coded 1 if a professional received one award or nomination, and 0 otherwise. But the same professional could receive more than one award or nomination for her performance in a movie in a given year, especially if she performed multiple roles in that movie. Accordingly, Models 6 and 7 in Table 5 report results for logistic regressions in which the dependent variable was coded 1 if a professional received two or more awards or nominations from peers (Model 6) or critics (Model 7). 8 The goal was to capture the degree of consecration accruing to any professional. As before, we find that peer organizations are less likely to bestow two or more accolades to peripheral members, whereas critic organizations do not seem to favor core over peripheral members.
Scope Conditions
Note: Standard errors in parentheses.
p < .05; **p < .01 (two-tailed tests for all variables).
Discussion and Conclusions
A rich and vibrant tradition building on Bourdieu’s pioneering insights treats cultural producers as engaged in an ongoing struggle to secure notoriety, prestige, and esteem from colleagues. In this struggle to define what counts as culturally legitimate, the social audiences that control access to symbolic and material resources play a crucial role. Cultural consecration can be viewed as the most definitive form of cultural legitimation. By conferring honors, awards, and prizes, cultural consecration separates individuals and achievements that are worthy of admiration and respect from those that are not. Consecration is important in virtually every field of cultural production. Although the process of consecration may be governed in part by objective criteria associated with merit, achievement, and performance, the relationship between observable differences and the attainment of symbolic capital is not straightforward.
Following Bourdieu, we sought to understand the consecration of cultural producers and their work as a joint result of socio-structural conditions at the individual level and judgments made by audiences about cultural producers and their offers. We framed cultural producers’ claims for recognition as an ongoing tension between the core and periphery of the social field: on the one hand, core actors have an interest in continuity, identity, and reproduction; on the other, peripheral actors are more prone to depart from established norms and standards. Audiences are central to this oppositional struggle because they define, elaborate, and, most important, bestow differential value on some producers and their claims while devaluing others.
After characterizing the discourse of value that permeates the social practice of consecration within two of the principal audiences in the film industry—peers and critics—we examined how these audiences differ in the extent to which their consecration choices are affected by the socio-topographic location of the producers they evaluate. Consistent with our hypotheses, we found that professionals in the periphery have lower odds of receiving an accolade from peers compared to professionals in the core. But core professionals do not have higher odds of receiving an accolade from critics than do producers in the periphery. When we allowed for unobserved heterogeneity at film level, we found even stronger support for an asymmetry in audiences’ consecration choices: not only do critics fail to privilege core incumbents, but they actually favor peripheral producers when awarding accolades. These findings yield several implications for future research that we now elaborate.
Contributions to Cultural Sociology
Conditional on prior achievements and status, cultural producers’ location in the social structure of a field may affect the consecration of their work and thus shape their reputation for talent and creativity. This result complements the vast research that treats individual merit as the primary factor in the production of work worthy of esteem (Gardner 1993; Sternberg 1985). In contrast to this prevailing approach, relatively little attention has been devoted to how consecration is shaped by processes of social validation enforced by external evaluators. Our findings are consistent with recent evidence (Allen and Parsons 2006; Rossman et al. 2010) that consecration does not occur in a social void, but is instead embedded in patterns of relationships and shaped by audiences that grant or deny distinction to competing candidates. Crucially, attaining such consecration likely hinges on whether peripheral players can appeal to a homologous social audience whose members share the same or similar dispositions and whose views, beliefs, and tastes are attuned to their own. Lacking the authority of core players—as well as their privileged access to resources, relationships, and other external markers of credibility—peripheral players face significant obstacles as they strive to establish legitimacy and attain consecration from established institutions. In any field of cultural production, the existence of a homologous audience represents a critical enabling condition for peripheral players to marshal credibility and increase their likelihood of success.
Contributions to Sociology of Stratification
In one of the most elaborate conceptualizations about the determinants of cultural legitimacy, Bourdieu (1984) situated taste at the center of a comprehensive theory of the relationship between social inequality and cultural practice, revealing the hidden social forces threaded through aesthetic judgment (see also Shrum 1991). From the choice of books to the consumption of rock music, symbolic forms operate within a system of exchanges and domination central to reproduction of the social structure. By focusing on the socio-structural conditions underlying the consecration process, our study contributes to the literature on determinants of stratification (Merton 1968), which has focused mainly on actors (e.g., individuals or organizations) vying for recognition rather than on the audiences responsible for conferring it (Zuckerman 1999). Research on the social structure of markets, for instance, predominantly focuses on market actors’ attributes and effects of their social position on the opportunities available to them (Benjamin and Podolny 1999; Podolny 1994). By emphasizing homology as a critical meso-level mechanism in the stratification of recognition, our study opens the door for a research agenda that exposes how audience evaluations shape the allocation of rewards among actors occupying different positions in a field’s social structure (Merton 1968).
Contributions to Sociology of Evaluation
In any cultural field, whether art or science, assessment of a given offer reflects subjective evaluation by the field’s relevant audiences. If audiences are diverse, highly divergent evaluation criteria may coexist. This variation allows for more cosmopolitan and liberal cognitive styles, thus raising the chance that work that departs from accepted norms will find a more receptive audience. Indeed, work that fails to garner attention and approval from one audience might still win the “intellectual attention space” (Collins 1998) of another homologous audience, whose motivations are different and whose evaluation criteria may be more attuned to dispositions of particular subsets of cultural producers.
Building on Bourdieu’s intuition on the role of homology variations across audiences in shaping the allocation of symbolic capital between core and periphery, we examined a case in which cultural producers face audiences with different degrees of homology vis-à-vis the producers they evaluate. Specifically, we considered the selective judgments of individual worthiness that are the expression of distinct aesthetic logics embedded in peer- and critic-based audiences. Focusing on these audiences allowed us to consider more carefully the social mechanisms that underlie the allocation of symbolic capital to core and peripheral producers.
Previous research has delved into the question of how cultural products become consecrated and integrated into the canon primarily by using case studies (for a recent comprehensive review, see Lamont 2012). Case studies may provide deeper insight into the underlying mechanisms of the phenomenon of interest, but they cannot unveil more general patterns that might extend to other cases—whether within the same context or across different arenas. Following prior research (e.g., Allen and Lincoln 2004), we adopted a large-sample research design to go beyond the “accumulation of finite case studies to capture general subprocesses at work” (Lamont 2012:206)—thus identifying more precise social-structural conditions that might affect the likelihood of consecration.
Several questions merit further attention. First, assignment of awards in the film industry is single-blind, not double-blind; as our results show, professionals’ reputations and cumulative recognition affect how favorably their work is received (Clemens et al. 1995). Our models controlled for a variety of factors associated with a producer’s reputation and status, but exploring the extent to which the social structural ordering of consecration applies beyond single-blinded contexts would be a significant addition. Also, we focused on contemporaneous consecration, that is, the recognition awarded to cultural producers within a short time after they produced their work. Contemporaneous consecration does not typically impart the same level of cultural legitimacy as retrospective consecration (Allen and Lincoln 2004), which attests that producers and products survived the test of time (Becker 1982; Bourdieu 1993). Yet contemporaneous consecration certainly raises the reputation of cultural producers, thereby increasing their access to resources needed to continue their work (Anand and Watson 2004; Lincoln 2007). What is the relationship between contemporaneous and retrospective consecration of cultural producers? Do the odds of being retrospectively consecrated differ if contemporaneous consecration originates from critics as opposed to peers? These are but some of the many questions that future research could explore in greater depth.
Footnotes
Appendix: Social Network Structure of the Field
To analyze the social structure of the industry, we reconstructed the bipartite affiliation network between professionals and movies. The bipartite network is a triple G = (┬, ┴, E) where ┬ (movies) and ┴ (professionals) are two disjoint sets of nodes—that is, the top and bottom nodes, respectively—and E ⊆ ┬ × ┴ is the set of links of the network. This differs from classical (unipartite) networks in that links exist only between top nodes and bottom nodes. Examples studied in the past include networks of individuals joined together by common participation in social events (Davis, Gardner, and Gardner 1941), CEOs of companies joined by common membership in social clubs (Galaskiewicz and Marsden 1978), collaborations among Broadway artists (Uzzi and Spiro 2005), and co-authorships (Newman 2004). For instance, in the case of co-authoring ┬ would be the set of papers and ┴ the set of authors, each author being linked to the papers she (co-)authored. Because group membership can often be established from membership lists or other sources, studies of these networks do not have to rely on interviews or questionnaires, thereby allowing one to construct much larger and more accurate networks than in traditional social network studies (Newman, Watts, and Strogatz 2002).
Given a bipartite network G = (┬, ┴, E), one can easily obtain its unipartite version defined as GI = (┴, EI) where {u, v} is in EI if u and v are both connected to the same (top) node in G. As illustrated in Figure A1, starting from the bipartite individual-by-movie network one can then recover the unipartite version. In this unipartite version of the network, each top node (movie) induces a complete subnetwork among the bottom nodes (individuals) to which it is connected. Links form between individuals when they work on multiple movies.
In defining a tie, we had to make an assumption about the duration of the relationship between professionals. With no control for relationship decay, professionals’ network connectedness would be highly inflated due to the likely inclusion of ties to inactive artists. Following a common practice in network studies, we made the adjacency matrixes time-limited by using a three-year moving window to control for the duration of each tie. In essence, each year we added nodes and ties resulting from new movies, and deleted nodes and their ties that had been inactive for three years (see also Uzzi and Spiro 2005). We started with professionals who worked in 1995 and used the earlier three-year data to construct the accumulative relational profiles (i.e., the period 1992 to 1994 can be viewed as the time needed to establish the network structure that professionals brought to the period 1995 onward). We used the resulting 10 time-varying matrices to compute all individual-level network measures. Using alternative windows of two, four, and five years produced no appreciable differences in our results.
Acknowledgements
All authors contributed equally. We thank the ASR editors and six anonymous reviewers for insightful comments on previous versions of the manuscript. We acknowledge suggestions from seminar participants at Amsterdam Business School, Ca’ Foscari University of Venice, Cass Business School, EM Lyon, HEC, IESE Business School, Imperial College, INSEAD, Luiss Business School, Medici Summer School, University of Pisa, Robert H. Smith School of Business, Ross School of Business, SKEMA Business School, Sloan School of Management, Stern Business School, and University of Trento. An early version of the paper was presented at the 2008 Academy of Management Conference, the 2009 Israeli Strategy Conference, the 2010 Druid Conference, the 2012 INSEAD Network Conference, and the 2013 Marconi Institute for Creativity Conference. We also thank Jon Morgon for editorial assistance. The authors are responsible for errors and omissions.
Notes
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
