Content uploaded by Rodolphe Durand
Author content
All content in this area was uploaded by Rodolphe Durand on Nov 22, 2015
Content may be subject to copyright.
Classical Deviation: Organizational and Individual Status as
Antecedents of Conformity
Rodolphe Durand &
Pierre Antoine Kremp
HEC Paris
SnO Research Center
Forthcoming
Academy of Management Journal
Abstract
Beside making organizations look like their peers through the adoption of similar attributes
(which we call alignment), this paper highlights the fact that conformity also enables
organizations to stand out by exhibiting highly salient attributes key to their field or industry
(which we call conventionality). Building on the conformity and status literatures, and using
the case of major U.S. symphony orchestras and the changes in their concert programing
between 1879 and 1969, we hypothesize and find that middle-status organizations are more
aligned, and middle-status individual leaders make more conventional choices than their low-
and high-status peers. In addition, the extent to which middle-status leaders adopt
conventional programming is moderated by the status of the organization and by its level of
alignment. This paper offers a novel theory and operationalization of organizational
conformity, and contributes to the literature on status effects, and more broadly to the
understanding of the key issues of distinctiveness and conformity.
1
Distinctiveness and conformity are core interests for researchers of organization and
management studies. Distinctiveness enables firms to gain sustainable competitive advantages
(Porter, 1996), to disrupt entire industries through innovation (Christensen, 1997), and to
develop new identities and styles (Lounsbury & Glynn, 2001; Sgourev & Althuizen, 2014).
However, conformity is required, as audiences are more likely to accept an organization’s
offering when it exhibits known and reassuring features (Hargadon & Douglas, 2001; Hsu,
Hannan, & Koçak, 2009). Moreover, excessive distinctiveness leads to incommensurability;
that is, when audiences face extreme distinctiveness, the lack of a comparison baseline makes
it difficult to evaluate the offering’s quality (Espeland & Sauder, 2007). Accounts of the
distinctiveness-conformity tradeoff usually put the spotlight on emblematic innovative
organizations (e.g., General Electric, Toyota, Google) or on visionary individuals (e.g.,
Edison, Jobs, Gates). The leading explanation for an organization's tendency to conform or
distinguish itself from its peers is based on status, as low- and high-status organizations have
less to lose from lack of conformity than middle-status organizations (Phillips and
Zuckerman, 2001). Yet, research has tended to focus on the factors that set some actors apart
from their peers, rather than on how actors subtly conform to various imperatives (Durand &
Jourdan, 2012; Durand, 2006). This paper offers a renewed conceptualization of
organizational conformity and of its status antecedents.
Existing approaches in organization and management studies offer an incomplete view
of conformity. They conceive of organizational conformity as a source of legitimacy and
associate conformity with similarity across organizations. In this view, through isomorphism,
organizations tend to look alike and act similarly, which, in turn, legitimizes their actions,
protects them from negative evaluations, and increases their chances of survival (Deephouse,
1999; Oliver, 1991; Phillips & Zuckerman, 2001; DiMaggio & Powell, 1983). Studies of
2
organizational conformity explain precisely why deviating organizations suffer penalties and
convincingly show that evading the prescriptions of existing product or market categories
leads to detrimental consequences (Hannan, Polos, & Carroll, 2007; Hsu et al., 2009).
However, these studies focus essentially on the negative side of normative prescriptions and
overlook their enabling role as guidelines for distinctive behaviors: organizations may
conform selectively by overplaying certain features while also introducing novelty (Durand,
Rao, & Monin, 2007; Kim & Jensen, 2011). Understanding that there may be different forms
of conformity requires revisiting the middle status conformity theory to explain better why
and how some organizations are more likely to conform than others.
Building on prior works, we define two forms of conformity—alignment and
conventionality—and analyze how they are influenced by actors’ status. The first form,
alignment, refers to the proximity of an organization to its peers across a range of attributes.
Consistent with prior research, the motivation for alignment comes from an organization’s
need to look like its peers and thereby avoid penalties. The second form, conventionality,
refers to the selective adoption of highly salient and expected attributes. The organization that
exhibits conventionality stands out relative to its peers by over-emphasizing certain widely
recognizable features in its offering. An organization’s motivation for conventionality relies
on it choosing to present and offer identifiable and comparable features, a form of quality test
that rewards the presence of signature elements. For example, in our empirical setting, when
an orchestra's programming choices are close to its peers', it exhibits alignment; when it
chooses to perform canonical composers more often than its peers, it exhibits conventionality.
Drawing on the main proposition that middle-status actors tend to conform more than
low-status or high-status actors (Phillips & Zuckerman, 2001), we investigate the role of both
organizations’ and individual leaders’ status on the likelihood that an organization will be
3
more aligned and/or conventional. Regarding alignment, we follow the middle-status
conformity hypothesis based on the fear of penalties, and expect that middle-status actors will
conform more than their counterparts because they have more to lose from sanctions.
Regarding conventionality, we hypothesize that middle-status actors are more likely to seek
social comparison to attest to their competence. Conventionality enables them to reap positive
rewards based on quality recognition. In contrast, high-status actors can afford to disregard
quality comparisons, whereas low-status leaders try to avoid them. Furthermore, the fact that
individual leaders are more likely to capture positive rewards and less likely to incur penalties
than organizations (because of audiences’ outcome ascription bias) creates a motivational
asymmetry: we expect that the individual leader’s status has a stronger effect on
conventionality while the organization's status has a stronger effect on alignment. Using the
U.S. symphony orchestras setting (1879–1969), we develop novel empirical
operationalizations of alignment and conventionality, and find as expected that although
orchestra status effectively explains alignment (we find a curvilinear relationship), musical
director status is a better predictor of conventionality (also a curvilinear relationship).
Moreover we show that the role of director status in explaining conventionality increases
when the orchestra status is low and when alignment at the orchestra level is low.
These findings contribute to three streams of research. First, this study advances
research on organizational conformity by distinguishing between two previously conflated
forms of conformity and operationalizing them both. It refines the middle-status conformity
theory and shows that normative pressures primarily orient organizations toward aligned
behaviors and individual leaders toward conventional behaviors. Organizations not only
conform to blend in with their peers but also to seek quality comparisons. Second, we
contribute to status research by studying simultaneously organization and individual status
4
and their joint effects on conformity. Third, our distinction between alignment and
conventionality has implications for the literature on deviance and distinctiveness, as
misaligned behaviors can be conventional. The resulting integrated theory of conformity and
additional empirical measures invite reconsiderations of some established findings in these
important lines of research.
TWO FORMS OF CONFORMITY: ALIGNMENT AND CONVENTIONALITY
Conformity with normative prescriptions is a precondition for organizational
legitimacy (DiMaggio & Powell, 1983). However, legitimacy through conformity can be
obtained in two distinct ways. First, organizations can conform by aligning with all the
characteristics of the group and abiding by the group’s norms (Hannan et al., 2007; Ody-
Brasier & Vermeulen, 2014; Phillips & Zuckerman, 2001; Phillips et al., 2013). Here,
conforming pushes the organization to abide by categorical membership norms, or more
broadly to follow a type, prototype, or code prevalent in its field or industry (Hsu et al., 2009;
Sharkey, 2014). This form of conformity has spurred research emphasizing the penalties
associated with misalignment: when an organization violates membership norms or when its
attributes position it too far out of the mainstream, audiences are more likely to ignore the
organization, question its offering, and discount it as illegitimate (Hsu, 2006; Negro, Koçak,
& Hsu, 2010).
Second, organizations can conform by exhibiting—and prominently displaying—not
all the expected attributes but only those that are highly recognizable and key to their field or
industry, which audiences can then attend to, compare, and assess. For an organization,
emphasizing these most highly distinctive features, expresses a form of conformity that we
call conventionality (see also Kim & Jensen, 2011). Audience members are likely to consider
5
actors that exhibit a limited number of highly recognizable attributes as conforming with their
own normative expectations (Durand & Paolella, 2013; Espeland & Sauder, 2007; Hsu &
Grodal, 2014; Vergne & Wry, 2014). Displaying such characteristics is central to an audience
recognizing a producer’s competence and quality (Sgourev & Althuizen, 2014). For example,
when a restaurant offers dishes emblematic of a revered culinary school or when a mutual
fund exhibits the most typical portfolio characteristics (e.g., Durand et al., 2007; Smith,
2011), they signal their willingness to comply with existing culinary or investment norms,
while opening opportunities to distinguish themselves from their peers on other attributes
(Brewer, 1991; Deephouse, 1999; DiMaggio & Stenberg 1985; Rao et al., 2005).
These two forms of conformity—alignment and conventionality—are analytically
distinct. Whereas aligned organizations blend in by adopting all the attributes of their peers,
conventional organizations stand out by exhibiting only their peers’ most salient features and
increasing the prominence of those features in their offerings. When organizations fall short
of perfect alignment (i.e., fail to look or act perfectly alike), they can emphasize or downplay
a limited number of salient features in an effort to orient their choices in more or less
conventional directions. Reciprocally, for a given display of conventionality, by emphasizing
salient features, organizations may choose to adopt or reject other traits that may lead to their
blending in with or distancing from their peers, such that they establish and present more or
less aligned profiles. However, owing to the finite number of traits or attributes that an
organization can choose, alignment and conformity have a conceptual relationship: the more
aligned an organization, the less it can stand out by appearing particularly committed to a few
key, expected attributes and features.
This distinction between alignment and conventionality helps to further separate the
legitimacy from the quality drivers of conformity – a central distinction for research on the
6
role of categorical membership for organizations (Hsu et al., 2009; Negro & Leung 2013).
Organizations conform not only to signal their attachment to group norms but also to derive
quality recognition. This distinction is critical because, in extant research, misaligned
organizations have been routinely treated as deviant; we argue on the contrary, that some
misaligned organizations may in fact be conventional. For example, Durand et al. (2007)
show that some restaurants both preserve and violate established norms concurrently,
allowing them to accrue rewards for their deviant positioning. Similarly, Smith (2011) shows
that mutual funds can differentiate themselves successfully by selecting attributes that depart
from those featured by their competitors; far from becoming overlooked, non-conformist
funds draw considerable attention from investors and can achieve superior capital inflows.
However, these studies neither disentangle the two forms of conformity, nor do they explain
why some organizations more readily decide to blend in or to stand out.
Two different motivations may drive towards distinct forms of conformity: the fear of
penalties and the quest for rewards.
1
First, because sanctions may be imposed on
organizations that fit poorly into existing categories or deviate excessively from the
mainstream, such penalties may lead to organizations choosing to align as a preventative
response. For instance, misaligned organizations are routinely discarded (Zuckerman, 1999),
face defiance about their offering (Negro et al., 2010), and suffer from reduced prices (Hsu et
al. 2009; Ody-Brasier & Vermeulen, 2014), whereas such penalties are typically avoided by
organizations that are aligned with their peers. Second, the opportunity for actors to gain
recognition as producers of quality suggests that conventionality can bring positive rewards.
Since conventionality implies committing to some salient features, actors thus allocate their
1
Penalties and rewards are not symmetrical: penalties entail more than the absence of rewards, and
rewards entail more than the absence of penalties. As our historical material will show for instance, a penalty
usually takes the form of negative critiques, not just the absence of a reward (e.g. absence of positive critiques).
7
efforts and resources to areas where quality is more readily assessable and comparable; in this
way, conventionality helps actors pass (and surpass) a quality threshold.
Although alignment is primarily driven by penalty avoidance (Durand & Paolella,
2013; Hannan et al., 2007; Hsu et al., 2009; Negro et al., 2010) and conventionality is
primarily driven by positive rewards (Durand et al, 2007; Kim and Jensen, 2011), these two
motivations are not necessarily mutually exclusive when organizations adopt alignment and
conventionality strategies. Still, alignment is not an effective way of obtaining rewards since
by definition, alignment places the focal organization at parity with its peers. By contrast,
choosing conventionality is not an effective way of avoiding penalties, as it makes the
organization stand out, which renders mistakes more egregious. We briefly present our setting
before developing our hypotheses regarding the influence of organizational (orchestra) and
individual (musical director) status on alignment and conventionality.
EMPIRICAL SETTING: THE MAJOR U.S. ORCHESTRAS (1879–1969)
To study the determinants of alignment and conventionality, we analyze the programming
choices of major U.S. symphony orchestras from 1879 to 1969. Permanent symphony
orchestras emerged in the United States in the late nineteenth century, as the established local
elites attempted to both elevate public taste and separate so-called classical music from
popular music in a strategy aimed at social distinction (Levine, 1988). Although popular
concerts sometimes included orchestral ensembles, they typically involved eclectic mixes of
short pieces, overtures, marches, dances, or arias designed primarily to entertain audiences
(Weber, 2001). In contrast, the definition of classical music, which centered on a narrow set
of canonical (mostly Austro-German) composers, was tied to the institutionalization of the
symphony orchestra in the last decades of the nineteenth century, as an organizational form
8
capable of endowing musical ensembles with autonomy from market demands and thereby
promoting highbrow musical tastes.
Wealthy individuals and local urban elites found that the most effective way of ridding
orchestra repertoires of their popular content and increasing performance quality was to
establish orchestras as organizations whose budgets did not rely solely on ticket sales.
"Cultural capitalists" (DiMaggio, 1982) such as Henry Lee Higginson (a Boston-based
stockbroker and philanthropist) and Charles Norman Fay (who headed several Chicago
utilities companies) founded permanent, non-profit symphony orchestras in, respectively,
Boston (1881) and Chicago (1890). In terms of music programming, Higginson's aim was to
"eschew vulgar music, i.e. such trash as is heard in theatres, sentimental or sensational
nonsense" (quoted in Howe, 1914: 29). The Boston and Chicago symphony orchestras
became models copied across the country (Horowitz 2005; Levine 1988). Such orchestras
focused exclusively on a limited repertoire of serious works and composers; moreover, by
offering regular concerts throughout the season and thus providing musicians with permanent
employment (and sometimes prohibiting them from seeking outside engagements), more time
could be spent in rehearsal, which led to the performing of more challenging pieces and
improvements in overall performance quality. For these orchestras and their musical directors
then, choosing the musical programming included responding to a smoldering question: to
what extent did they need to align with the programming choices of their peers in their efforts
to blend in and avoid sanctions? Could they attempt to appear conventional, rather than being
aligned, and stand out relative to their peers?
9
Programming Choices and the Dominant Role of the Musical Director
Over our period of study, the internal structure of U.S. symphony orchestras has remained
remarkably unchanged: the musical director,
2
appointed by and reporting directly to the board
of trustees, is solely responsible for artistic decisions (concert programming, interpreting
works, conducting performances, hiring or firing musicians), and the board is in charge of the
financial health of the organization (Glynn, 2005). In theory, the artistic power of the maestro
could be restricted by the orchestra’s board; but, in practice, musical directors enjoy great
autonomy. Board members generally refuse to let their tastes influence programming, as
Higginson declared:
All the catholicity seems to me good. I do not like Wagner's “music”, and take little
interest in much of the newer composers, but I should not like to bar them out of our
programs. The extent to which Wagner has been played from the very beginning is a
token of the entire freedom with which [musical directors] have made their programs.
(quoted in Howe, 1914: 30)
Far from being specific to Boston, this high level of musical director’s autonomy was
widely shared in the orchestra world: musical directors assumed full control of their programs
and could afford to present programs counter to the tastes of their audiences in their pursuit of
higher artistic aims—thus giving orchestra repertoires the appearance of a supply-driven
market. When asked to include lighter and more popular pieces in his Chicago Symphony
Orchestra programs, Theodore Thomas (the Chicago Symphony’s first musical director)
boldly stated: "Those who cannot enjoy the great music, poor fellows, I do not grudge what
they can enjoy…. I will play for them now and then, but it is not for [the] Tell Overture and
Handel’s Largo that Chicago supports my orchestra. One does not buy a Krupp canon to shoot
sparrows" (Levine, 1988: 118). Similarly, Max Rudolf (musical director of the Cincinnati
Symphony from 1958 to 1970) declared in an interview:
2
We refer to conductors acting in their capacity as musical director, music director, principal conductor, or chief
conductor as "musical directors." The titles differ by orchestra, but the roles and functions are equivalent.
10
My board has never told me what I should or should not play. We have discussed certain
reactions on the part of subscribers and newspapers, but they do not exert the slightest
degree of pressure. (quoted in Swoboda, 1967: 63–64).
Social Comparisons and Deviations from Programming Norms
The musical directors’ full authority and control over programming choices, with relative
autonomy from board pressures, does not imply that orchestras and musical directors are
unaffected by either the reception of their performances or the programming choices made by
their peers and rivals. Although musical directors’ choices might not be coerced, they are
shaped by the directors’ understanding of the field-level rules of the game. The symphony
orchestra field is fraught with social comparisons, leading to orchestras being compared by
critics and audiences, especially when orchestras tour the country in an effort to keep their
musicians employed past the regular season in their home cities. For example, Wilhelm
Gericke (who directed the Boston Symphony from 1884 to 1889) recalls the stakes on
February 14, 1887, when his orchestra first performed in New York:
In those days, all New York music-lovers were great admirers of Theodore Thomas and
the New York Philharmonic Orchestra, who had every reason to be thus admired. So the
standard in New York was a high one.… There is no doubt that this first success in New
York affected greatly the Boston audience; from that moment, the Boston Symphony
Orchestra began to stand on solid ground. The members of the Orchestra began to feel
that they belonged to an artistic corporation of first rank. (quoted in Howe, 1914: 112–
113)
Social comparisons consistently focus on two distinct elements: repertoire breadth and
performance quality. Indeed, a motley repertoire risks being misaligned and dismissed by
critics, whereas a focused and canonical repertoire generates positive expectations triggering
quality comparisons. First, critics and concertgoers evaluate and compare the breadth of
concert offerings of multiple orchestras, and routinely disparage orchestras whose
programming choices are misaligned. Musical directors and orchestras receive criticism for
performing uninteresting pieces that seem unfit for a classical program, such as when, in
11
1894, the Boston Symphony performed a piece from the English composer Arthur Sullivan’s
1878 opera HMS Pinafore, which was deemed "wholly out of place in an entertainment
presumably serious" (New York Times, February 9, 1894). Systematic criticisms of orchestra
programming choices often point specifically to the local idiosyncrasies of one orchestra’s
programs relative to others; in condemning the programs of the New York Philharmonic and
New York Symphony, The New York Times critic Olin Downes asked, "Why cannot the local
New York orchestras, like those of other cities, give programs which have variety, balance
and inclusiveness?" (March 23, 1924). Similarly, a scathing 1956 New York Times review
criticized the Philharmonic's programs for seeming "haphazard"—and although the orchestra
and its director, Dimitri Mitropoulos, were credited for having performed important works of
modern composers during the current season, some of their choices were viewed as difficult
to justify: "Schoenberg, Berg, Shostakovich, Prokofiev have fared well, but so have Morton
Gould and Rolf Liebermann, who do not deserve that much attention" (The New York Times,
April 29, 1956). Through this mechanism of social comparison, misaligned programming
choices are likely to be lambasted, providing reason for orchestras to align.
Second, social comparisons focus not only on repertoire breadth but also on
performance quality, as audiences compare the same compositions as performed by different
musical directors and orchestras. This leads musical directors and orchestras to select highly
conventional musical programs and to over-represent certain canonical works and composers
(Peyser, 2000). This tendency became especially visible with the rise of "star conductors"
after World War I, when some orchestras, unable to hire or retain a permanent musical
director, relied on multiple guest conductors for entire seasons (Lebrecht, 1991). As Philip
Hart observes: "For some conductors, it may be easier to spend their time jetting from one
engagement to another and devoting their study time to performing their interpretation of the
12
surefire repertory" (1973: 458). A conventional repertoire presents advantages for the musical
director and to a lesser degree the orchestra. Musical directors are more likely to be praised
for their commitment to high-quality performances, which enhances their reputation and
draws larger fan audiences—while minimizing their time spent rehearsing with the orchestra
(Arian, 1971: 24–25). Orchestras also benefit, as such programming decisions deliver
financial benefits. As a disillusioned New York Times critic noted: "Two birds can be killed
with one stone: the orchestra which could play the Fifth Symphony in its sleep will not
require extra rehearsals and the orchestra will sell tickets" (March 23, 1924).
Musical directors thus can exercise high discretion in determining which works and
which composers to perform, but need to take into account both their orchestra’s capabilities
and concert-programming norms. Depending on their status, social comparison pressures do
not apply equally to all individual leaders and organizations, which raises questions about
whether and why organizations adopt more or less aligned or conventional behaviors.
HYPOTHESES
Our analysis focuses on status as the main antecedent of organizational conformity. In this
section, Hypotheses 1 and 2 relate status to alignment and conventionality for individual and
organizational actors alike, whereas Hypothesis 3 distinguishes between individual-level and
organizational-level status effects.
With regard to alignment, our baseline hypothesis comes from Phillips and
Zuckerman’s (2001) middle-status conformity theory. Actors conform to norms to blend in
with their peers and avoid any sanction associated with violations of membership norms or,
more broadly, deviation from mainstream attributes. For example, the norm for a financial
analyst not to issue a sell recommendation on a firm's stock is based on the fear of losing
13
future access to information on that firm (Phillips & Zuckerman, 2001); similarly, the norm
for a corporate law firm not to move into family law practice is rooted in the principle that
law firms must be fully dedicated to their corporate clients' interests (Phillips et al., 2013).
Likewise, in the field of symphony orchestras, following concert programming norms when
choosing which works or composers to feature helps minimize negative responses from
audiences. Actors in the field were deeply aware of this alignment imperative. For example,
Frederick Stock, who held the longest appointment as musical director in the history of the
Chicago Symphony Orchestra—from 1905 to 1943, confessed spending "endless time
pondering over the problem of program-making" (Ewen, 1943: 133–134); his programming
choices were notoriously balanced.
However, some actors find it easier to distance themselves from prototypical choices;
the middle-status conformity hypothesis predicts that high-status actors can be misaligned
without fearing sanctions, because of the stability of their position (Phillips & Zuckerman
2001; Sharkey, 2014) and because they benefit from greater control and visibility (Podolny,
2005;; Sauder, Lynn & Podolny, 2012). As a result, these actors can evade the normative
constraints and differentiate themselves. They also typically express self-confidence in their
ability to undertake bold changes, although complacency could negatively affect their
performance (Bothner, Smith & White, 2010; Bothner, Kim & Smith, 2012). For example,
after a performance of Shostakovich's Fifth Symphony in Boston, which critics found "banal,
vulgar, thin," and "unoriginal," "Serge Koussevitzky [one of the most revered conductors of
his time] countered with his usual weapon, stating: 'I will keep repeating the music until the
critics realize its greatness'. And he did" (Johnson, 1950: 154). At the other end of the
hierarchy, low-status actors have insecure positions and low control and visibility. Yet they
too can transgress existing norms and avoid pure alignment: they have little fear of sanctions
14
because they have little to lose. Thus, low-status orchestras may seek to surprise audiences by
adopting misaligned choices because their low status reduces the conformity pressures they
face. In contrast, middle-status players face greater pressures to align. Unlike high-status
actors, they cannot afford to defy expectations; and unlike low-status actors, they have more
to lose from negative audience evaluations. As a result, we expect:
Hypothesis 1. Middle-status actors exhibit higher levels of alignment than high- and low-
status actors.
The middle-status conformity perspective relies on a two-stage model in which the audience
first grants membership to a candidate based on compliance with membership norms, and in a
second step evaluates the quality of the subset of full-fledged members (Phillips &
Zuckerman, 2001). While extremely fruitful, this model only considers negative penalties
(through lack of attention) as drivers of conformity, and fails to consider the quest for positive
rewards as a motivation for conformity. Our approach relaxes this assumption and applies
more broadly to cases in which audiences may also evaluate categorical membership and
quality concurrently. Conformity does not only enable actors to blend in and avoid sanctions;
by conforming selectively, actors also seek relative gains, in the form of quality recognition.
In particular, if audiences can evaluate an organization's offering based on a small number of
highly expected, and emblematic attributes, exhibiting such salient features can lead to
beneficial consequences, such as critical acclaim, celebrity, and profit (Durand et al., 2007;
Hayward, Rindova & Pollock, 2004; Lebrecht, 1991). While alignment provides limited
positive reward prospects (whereby the organization blends in and reaches parity with other
group members at best), conventionality is a more effective way for actors to gain recognition
as quality producers because of its focus on highly salient, comparable, and easily assessable
activities (Sauder, 2008; Espeland & Sauder, 2007). However, conventionality is not risk-free
15
because standing out on a few core features entails the risk of failing the quality test.
When actors conform in an effort to maintain or increase their perceived quality, they
tend to adopt more conventional behaviors, which implies a more systematic focus on the
norms’ most salient elements. For example, orchestras and musical directors could choose to
allocate a greater portion of their repertoires to Mozart’s and Beethoven’s symphonies, rather
than attempting to perform less frequently played music, for two reasons. First, audience
members (both critics and concertgoers) can easily recognize and evaluate the performance
quality of frequently heard masterpieces and then compare them against other orchestras’
performances. Second, concentrating on such concert program staples enables orchestras and
musical directors to spend less time and effort in rehearsal but still reach the desired level of
quality because the musicians are trained in and accustomed to performing these canonical
pieces (Arian, 1971). Still, committing to conventional choices can be a double-edged sword
as imperfections in the quality become more obvious.
As with our alignment prediction, we expect actors to adopt conventional or
unconventional choices as a function of their status. Because high-status actors are unlikely to
achieve significant gains by making conventional choices, they may be more prone to
disregard conventions, as audiences already interpret their status as a signal of competence
(Rao et al., 2005; Podolny, 2005). As their commitment and ability to deliver high quality are
unquestioned, high-status actors tend to display fewer expected attributes and more
unexpected ones (Rao et al., 2005). In turn, they attempt to earn rewards in excess of those
they could garner by continuing to play only the most salient composers. Furthermore, among
all actors, high-status actors pay less attention to financial constraints and are less likely to
limit themselves to common programming choices that are economically more efficient. Their
disregard for convention and their resulting relatively bolder strategies are likely to generate
16
superior rewards in the form of higher visibility and positive evaluations (Smith, 2011;
Waguespack & Sorenson, 2011). For instance, Leopold Stokowski (who joined the
Philadelphia Orchestra in 1912 after a short tenure at the Cincinnati Symphony) did not
hesitate to respond directly to critics—or to confront hissing concertgoers during concerts (as
during the first New York performance of Arnold Schoenberg’s Variations for Orchestra Op.
31; The New York Times, October 23, 1929).
At the other end of the spectrum, low-status actors could benefit from conventional
choices but face prohibitive costs in competing head-to-head against their established peers.
Instead they seek to avoid direct quality comparisons with higher-status actors, both because
they lack the inherent resources and capabilities to perform the classics to the same high
quality and because competing on quality would be excessively costly. We therefore expect
low-status actors to adopt unconventional programming in the same way as high-status actors
do, but for a different reason: comparison avoidance. In his survey of conductors in the United
States, Dictators of the Baton (1943), David Ewen describes the programming choices of
Hans Kindler, the first, little-known musical director of the National Symphony Orchestra in
Washington, DC:
Kindler has instituted in Washington one of the most extensive repertoires known by any
of our younger orchestras: more than 700 [pieces], of which a fair proportion are
novelties rarely to be heard elsewhere. And this extensive and varied repertoire extending
from transcriptions of music by the venerable Frescobaldi to a new work by the young
American, William Schuman played with dignity and an appreciation for style. Kindler
may not rise to those empyrean heights to which some other conductors may soar, but at
the same time it can also be said that he never descends to the depths of cheapness,
superficiality, or sensationalism. (251–252)
Similarly, the Seattle Symphony, a low-status orchestra until the mid-1960s (a touring
British conductor in 1941 described it as located in a "cultural dustbin"), could not compete
against high-status orchestras on performance quality alone. As reviews from this period
indicate, the orchestra lacked the "precision," "reliability," and "richness" found in major
17
orchestras (The New York Times, October 9, 1968). Despite his personal preference for the
nineteenth-century Romantic repertoire, musical director Milton Katims adopted
unconventional programs to grab the attention of music critics, on the criterion of originality
rather than quality (Lehmann, 1992; Seltzer, 1975).
In contrast with both higher- and lower-status actors, middle-status actors should be
more likely to adopt conventional repertoires and represent the classical canon, in their efforts
to signal or prove to audiences their commitment to high quality. Middle-status actors can
focus on conventional choices at acceptable costs, with greater potential gains than would be
available to higher- or lower-status actors. First, middle-status actors have better capabilities
than low-status actors to compete on quality, by focusing on the performance of canonical
works that are guaranteed to generate attention and to be evaluated. Second, compared with
high-status actors, middle-status actors are more likely to be sensitive to the potential
economic benefits of conventional programming. For example, Eugene Ormandy had middle-
level status when he was appointed musical director of the Philadelphia Orchestra in 1936;
though he had enjoyed some success and recognition while directing the Minneapolis
Symphony since 1931, he was never cited as one of the great musical directors of the time,
unlike Stokowski, his predecessor in Philadelphia. Early in his tenure, Ormandy faced both
questions about whether he would be able to maintain the quality of his orchestra (the so-
called Stokowski sound, for which Philadelphia had become famous) and challenges to his
programming choices (Kupferberg, 1970). In response, he decided to focus initially on
conventional pieces, later characterized by the Orchestra Board Association in these terms:
"No stunts, no loud noises, nothing to offend.… The audience was genuinely pleased to hear a
program of beautiful music instead of foundry imitations or noises from the nethermost
depths" (quoted in Arian, 1971: 19). As these examples suggest, middle-status actors adopt
18
conventional programming choices to compete on perceived performance quality, in contrast
with both high-status actors who can afford to disregard convention and low-status actors who
seek to avoid direct quality comparison. We thus hypothesize that:
Hypothesis 2. Middle-status actors exhibit higher levels of conventionality than high- and
low-status actors.
While Hypothesis 1 and Hypothesis 2 apply both to organizational and individual actors,
Hypothesis 3 refines the previous hypotheses by distinguishing the levels of analysis where
status is most likely to affect organizational alignment and conventionality. We base our
reasoning on the bias in outcome ascription that leads to a motivation asymmetry between the
organization and its individual leader.
Audiences ascribe outcomes differently to organizations and individual leaders. In
particular, prior research has shown that attributions of success tend to be individual-based.
That is, CEOs, chefs, and musical directors receive praise for the actions they take at the helm
of their organizations, whereas negative outcomes are more often attributed to organizations
(Hayward et al., 2004; Rao et al., 2005). In particular, Hayward et al. (2004) predict that the
ascription of positive outcomes to individual leaders is particularly likely when audiences
perceive organizational choices as distinctive and consistent. In our setting as in many others,
audiences praise musical directors for successes and blame orchestras for flops (Lebrecht,
1991). For example, Artur Rodzinski was credited for having managed, between 1933 and
1943, to transform the Cleveland Orchestra, an "unimpressive ensemble … of provincial
stature," with "marked defections in various departments" into "one of the great orchestras" in
the United States (Ewen, 1943: 225, 229). In a context in which audiences ascribe outcomes
differently to organizations and individual leaders, positive rewards and negative sanctions do
not affect organizational and individual actors equally. Rewards are highly individualized and
19
centered on musical directors. For example, Time Magazine’s cover first featured a musical
conductor, Arturo Toscanini, in 1926. Star conductors that benefitted from such media
attention could tour worldwide and play with local orchestras to impress a public that tended
to attribute the quality of the performance to the maestro’s genius, while blaming any
weaknesses on the orchestra’s deficiencies (Peyser, 2000). As can be found in other settings,
penalties affect individual leaders comparatively less than they affect organizations; criticism
is not limited to an individual leader or a single product or performance but rather
contaminates the whole organization (Vergne, 2012). This difference in the way in which
audiences ascribe outcomes to organizations and individual leaders influences how penalties
and rewards affect orchestras and musical directors. Figure 1 shows a stylized representation
of individual- and organizational-level payoffs of audiences’ negative and positive
evaluations: orchestras are unable to reap the full benefits of positive rewards but are more
deeply affected by negative evaluations, while musical directors are relatively protected from
negative evaluations (sanctions) but benefit disproportionately from praises.
[INSERT FIGURE 1 ABOUT HERE]
When individual leaders and organizations face conformity pressures, differences in
outcome ascription create a motivation asymmetry, and individuals and organizations respond
to conformity pressures differently. More precisely, alignment tends to smooth the differences
between organizations and decrease penalties (to which organizations are more sensitive),
while conventionality allows quality recognition and rewards (from which individual leaders
are more likely to benefit). Because individuals and organizations experience penalties and
rewards differently, the interest of middle-status individual leaders is to conform by seeking
conventionality, whereas the interest of middle-status organizations is to conform through
alignment. Empirical studies in behavioral economics have demonstrated the existence of
20
motivation asymmetry in such situations, where individuals benefit from positive gains but
are relatively protected from negative sanctions: the individuals’ objectives become
disconnected from the objectives of the organizations to which they belong (Tirole, 1988). As
a consequence, decision makers have incentives to undertake bolder moves than their
organizations would call for (Chevalier & Ellison, 1997, 1999; on the banking industry, see
also McNamara & Bromiley, 1997). Individual leaders are therefore more likely to conform to
normative pressures by focusing on limited and selected actions that will highlight them
relative to their peers, as a way to reap positive rewards based on quality demonstrations
(conforming through conventionality). Conversely, individual leaders are less motivated to
conform by blending in, as a way to avoid sanctions (conforming through alignment).
Drawing on outcome ascription differences and motivation asymmetry, we deduce that
individual leaders’ characteristics have a greater effect on conventionality than organizational
characteristics, and reciprocally for alignment. It then follows that the musical director’s
status mainly influences conformity as conventionality because directors are structurally more
interested in and are more likely to benefit directly from quality comparisons. By contrast, on
average, orchestra status mainly affects conformity as alignment because the penalties
associated with misaligned programming apply more directly to orchestras than to musical
directors, whereas the rewards tend to be more appropriated by individuals than by
collectives. Hence:
Hypothesis 3a. The effect of organizational status on alignment is stronger than that of
individual status.
Hypothesis 3b. The effect of individual status on conventionality is stronger than that of
organizational status.
21
DATA AND METHOD
We test our hypotheses by analyzing the concert programming decisions of symphony
orchestras in the United States. We use comprehensive repertoire data collected from 27 of
the 30 major orchestras
3
, as defined by the American Symphony Orchestra League in 1972
(i.e., with budgets exceeding $1 million; Mueller 1973; CPANDA 2008). For each orchestra,
the data includes the total number of annual performances of each piece by each composer
since its founding (e.g., 1842 for the New York Philharmonic, 1960 for the Milwaukee
Symphony Orchestra) and up to 1969. Data on musical director tenures came from the
Directory of American Symphony Orchestras (Craven 1986).
Although these data focus on a limited number of professional orchestras with access
to significant resources prior to the 1970s, we find significant variance in their visibility and
recognition, with major orchestras directly competing for critics’ and concertgoers’ attention
against other world-class orchestras (New York Philharmonic, Boston Symphony), and more
recently founded orchestras that, at some point of their history, had difficulty reaching more
than local audiences (Milwaukee Symphony, Utah Symphony). Still, over our period of study,
audiences (concert goers and critics) did not contest the definition of the genre and shared
similar expectations: the field of major symphony orchestras was entirely dedicated to the
performance of serious classical music and was largely united despite status differences.
4
Dependent Variables. To determine orchestras’ levels of alignment and conventionality, we
evaluated each orchestra’s repertoire programming decisions relative to the programming of
3
This dataset has been analyzed in previous work on cultural innovation and the transformation of the musical
canon over time (Dowd et al., 2002; Kremp, 2010). Although we use some of the same control variables
(director tenure, number of performances) and fixed effects (decade, orchestra, director), our dependent and
independent variables of interest do not appear in these prior studies.
4
For example, even a low status orchestra such as the Utah Symphony modeled itself on world-class orchestras
like the Boston or Chicago Symphony (Swoboda 1969). Maurice Abravanel (Utah’s musical director in the
1950s and 1960s) had connections with the most renowned conductors of his time, such that Bruno Walter and
Wilhelm Furtwängler had recommended him for a temporary position at the Metropolitan Opera.
22
all other orchestras in the same season. We constructed two distinct measures to
operationalize alignment and conventionality in each season. First, alignment is the closeness
of each orchestra’s repertoire to the repertoires of other orchestras. We computed the sum of
squared distances between (1) the proportion of the orchestra’s total number of performances
devoted to each composer whose works were performed by any orchestra during the season,
and (2) the proportion of other orchestras’ performances of each composer’s works. More
precisely, we calculate this distance di for every season with the following equation:
where nic is the number of performances by orchestra i of works by composer c (including
repeat performances of the same pieces); Ni is the total number of performances during the
season by orchestra i of works by any composer; noc is the total number of performances by
other orchestras of works by composer c; and No is the number of performances by other
orchestras of works by any composer.
To construct repertoire alignment, we scaled this distance measure by dimax (i.e.,
maximum possible distance between an orchestra’s repertoire and other orchestras’
repertoires) and computed the negative of the logged scaled distance:
This measure has a zero lower bound (for orchestras approaching the maximum
possible distance between their repertoires and those of other orchestras) and takes higher
values as the distance di gets smaller and the orchestra’s repertoire becomes more aligned
with the average repertoires of all other orchestras. The log transformation reduces the
influence of alignment values on the lower end of the distribution. Our results remain
unchanged without the log transformation.
23
Second, conventionality captures the extent to which the orchestra’s programming
choices focus on a subset of highly salient composers emblematic of the field. In other words,
conventionality can be defined as a systematic bias toward programming choices that are
already widely shared and accepted. To measure this bias in orchestras’ programming
decisions for each season, we analyzed the extent to which an orchestra’s programming
decisions co-varied with the choices made by its peers. We regressed the composer shares of
each orchestra’s annual performances on the composer shares of performances at other
orchestras:
The slope of the regression line âi can be interpreted as a measure of conventionality:
If an orchestra seems to devote a relatively larger share of its program in a given
season to works of composers that also account for large shares of the performances by other
orchestras, the slope of the regression line exceeds 1, indicating an above-average
commitment to canonical composers. Conversely, a slope less than 1 implies that the
orchestra performs less frequently works by composers who account for large shares of the
repertoires of others; that is, it shows a below-average commitment to canonical composers.
Table 1 shows that, empirically, alignment and conventionality are uncorrelated
measures (r = .02). Independent of orchestras’ alignment, the direction of their repertoire
deviations, as measured by conventionality, is equally likely to be biased toward works by
popular composers (higher conventionality, values exceeding 1) or toward those of rarely
performed composers (below 1).
24
Figure 2, Panels a and b, provide graphical illustrations of our operationalization of
alignment and conventionality. We plot the share of various composers in a focal orchestra's
repertoire, against the share of those composers in the aggregate repertoire of all other
orchestras, in a given season. Panel a compares two equally conventional orchestras with
different levels of alignment, where the alignment measure is a function of the distances
between each composer’s position and the 45-degree line. On average, the Baltimore
Symphony Orchestra's choices were less aligned than the Cleveland Orchestra's during 1944–
1945, such that Baltimore’s programs leaned heavily toward some composers (popular ones
such as Brahms and Bach, but also less popular ones such as Mussorgsky and Henri Duparc)
and discounted others (e.g., Beethoven, Wagner, Strauss). These choices do not reveal any
bias toward more or less frequently performed composers, so Baltimore's choices were neither
conventional nor unconventional, yet they were clearly dissimilar from the choices made by
other orchestras, indicating a low level of alignment. In contrast, Cleveland stuck more
closely to the average programming choices of its peers (we can observe that the vertical lines
representing distances are shorter hence closer to the diagonal line than for Baltimore.) Panel
b instead depicts two repertoires with similar alignment but high and low levels of
conventionality, according to the slope of the regression line. The steeper slope of the
regression line for the San Francisco Symphony Orchestra indicates that its programming
choices were more conventional, in that it performed works by the most popular composers
(Beethoven and Mozart) considerably more often than did other orchestras in 1964–1965. In
contrast, Baltimore gave more space to less frequently performed composers, such as
Mussorgsky, and it performed Beethoven, Mozart, and Brahms less often.
[INSERT FIGURES 2a AND 2b HERE]
25
Independent Variables. To operationalize orchestra status, we constructed a network
measure of centrality that captures the orchestra’s prescription capacity in the field, through
the introduction and adoption of new composers (whose works had not yet been played by
any orchestra). For every season, we created an inter-orchestra diffusion network of newly
introduced composers, such that the nodes are orchestras i and j, and the value of ties vijt is a
function of the total number of composers entering the repertoire of orchestra j after they were
first introduced to the field's repertoire by orchestra i, as well as the speed at which the new
composers introduced by orchestra i were incorporated into orchestra j's repertoire:
where i is the "introducer" orchestra; Ci is the set of composers introduced to the repertoire by
that orchestra; j is the "adopter" orchestra; t is a season; tc is the season during which orchestra
i premiered composer c; and 1cjt equals one if composer c entered the repertoire of orchestra j
during season t, and zero otherwise. Adopting a composer already premiered by an introducer
orchestra increases the tie value, but this increase depends on the timing of the adoption and
declines steadily as the time since the composer was first introduced passes, to the point
where it becomes null if adoption occurs five years or more after the introduction. In other
words, a high vijt value indicates that the introducer orchestra's recent innovative
programming choices were rapidly adopted by the adopter orchestra.
To measure orchestra status, we use Bonacich and Lloyd’s (2001) eigenvector-like
measure of centrality for non-reciprocal ties based on the cumulative diffusion networks of
new composers over the previous 25 years.
5
This measure of centrality captures two
dimensions of status: visibility and social recognition (Podolny, 2005; Bourdieu, 1993) and
5
The Bonacich and Lloyd’s centrality measure extends traditional Bonacich (1987) centrality to cases in which
ties are not reciprocal and some nodes receive no in-degrees.
26
assigns a higher score to nodes tied to central nodes; an orchestra’s status is higher if other
high-status orchestras decide to perform the composers it first introduced.
6
Because status is a
relative measure, we center this variable on its annual mean and standardize it.
We create a similar measure for musical director status, using Bonacich and Lloyd’s
(2001) centrality scores in the inter-director diffusion network of new composers. In this
interpersonal (rather than interorganizational) diffusion network, the nodes are musical
directors in charge of concert programming decisions, and the values of the ties are calculated
as functions of the number of new composers introduced by musical director i and later
performed by musical director j, as well as how soon musical director j adopted those new
composers.
7
In the same way that centrality scores derived from the ordering of positions on
‘tombstone’ advertisements are not a direct measure of the status of investment banks, but
reveal and reinforce their status positions (Podolny 2005), or the joint regional affiliation of
wineries is a cause and a consequence of the status hierarchy between appellations (Benjamin
and Podolny, 1999), the centrality of orchestras and musical directors measured from
diffusion networks of new composers is a proxy for, rather than a direct measure of, their
status. Its validity is well supported by qualitative evidence about the field. Lebrecht (2001:9)
confirms that "one measure by which a great conductor can be assessed is the new music that
he [has] brought into being." Obsessed with the need to display their influence on the field
6
This measure of centrality C verifies C = α RC + 1, where R is the cumulated diffusion matrix (rij is the total
number of composers adopted by orchestra j, introduced by orchestra i), and α is a scalar smaller than the inverse
of l0, the largest eigenvalue of R. Higher values of α give higher centrality scores to nodes with out-degrees to
highly central nodes. We use α = .75/l0, but the results remained unchanged when we used values of α between
.5 and .99.
7
Musical director status is positively but imperfectly correlated with orchestra status (r = .51, Table 1), because
the trajectories of orchestras are not tied solely to the careers of one conductor. Musical directors retire or die (in
which case they disappear from the diffusion network) or else change jobs, within or outside the United States,
and they also accumulate status over multiple tenures. So we analyze the relative importance of two comparable
but analytically and empirically distinct status measures.
27
("Conductors like the glamour of the premiere. It gives them a page in the history book" noted
the composer William Schuman; see Swoboda 1967: 179), orchestras and musical directors
concentrate on premieres (in what a critic referred to as a "first performance itch"; The New
York Times, April 2, 1944) and compete with one another to secure new performances first.
8
The prospect of performing a composer after his or her works had already been premiered was
less appealing to orchestras and musical directors, and involved both a recognition of the
quality of the composer as well as an acknowledgement on their part of the quality of the
programing choices of their peer who had secured the premiere. In turn, being selected by
high-status orchestras or musical directors offers a critical channel through which budding
composers become recognized and have their work performed subsequently by other
orchestras (Kremp 2010).
As control variables, we include the total number of annual performances by each
orchestra, to separate the effect of status from performance capabilities. That is, high-status
organizations and individuals likely have access to significant resources (donations, audience
size), which allow them to schedule more concerts during the season. Finally, we include the
length of the musical director’s tenure to date as a control variable, because the effect of
director status must be distinguished from the effect of experience with a given orchestra.
[INSERT TABLE 1 HERE]
Analyses. We used linear models to predict the alignment and conventionality of each
orchestra in the data set for all seasons from 1879 to 1969, with orchestra and musical director
fixed effects to account for unobserved, time-invariant heterogeneity between organizations
and musical directors. To control for field-level changes in mean alignment and
conventionality over time, we used fixed effects for each 10-year period. Results from the
8
For example, the rivalry between Stokowski in Philadelphia and Toscanini in New York centered on securing
premieres of works by Stravinsky and Schoenberg (Kupfenberg 1969).
28
fixed-effect models can be interpreted as within-orchestra, within-director, and within-decade
estimates. Any variable omitted from the model therefore would need to be a time-varying
orchestra or director characteristic affecting alignment or conventionality, and correlated with
individual or organizational status, for it to affect our results. Our predictors are lagged by one
year, which limits the risk that reverse causality might bias our results. We also introduced a
lagged dependent variable as a robustness check in one model specification.
Although the New York Philharmonic was founded in 1842, no status measure can be
calculated until at least two orchestras were active in the field; the next major orchestra
founding only came with the creation of the New York Symphony in 1878, with others
following soon after, such as the Boston Symphony in 1881 and the Chicago Symphony in
1891. Our analyses thus focus on the years 1879–1969.
RESULTS
Table 2 contains the estimates from the linear models of repertoire alignment using
fixed effects for each orchestra, each musical director, and each decade. Model 1 includes
control variables only, showing both director’s tenure and the number of performances per
season tend to increase alignment. Model 2 adds the linear and quadratic effect of
organizational status, both of which are significant and have expected signs, which supports
Hypothesis 1 - that the relationship between organizational status and alignment has an
inverted U shape. Orchestra repertoire alignment is higher when orchestras are located toward
the middle of the status distribution (about 0.85 of a standard deviation above the mean
9
).
9
While this result could seem at odds with a spontaneous definition of middle status as "mean" status, it is
consistent with prior literature on the antecedents of norm conformity. Phillips and Zuckerman (2001) find that
conformity pressures are highest for law firms whose status is located 1.3 standard deviations above the mean,
and investment banks whose status is 0.9 standard deviations above the mean. In both cases, middle status is
defined not as mean status, but as a middle point in the status hierarchy, that is about equidistant between the
highest- and lowest-status actors in the organizational field.
29
Models 3 and 4 help us refine this finding, in that they indicate no significant quadratic
relationships between musical director status and orchestra repertoire alignment. In Model 4,
the curvilinear relationship between orchestra status and alignment persists when we add
musical director status into the model (although the coefficient on the quadratic term for
orchestra status only differs from 0 at the p = .10 level). Our models are therefore consistent
with Hypothesis 3a: conformity through alignment is more the product of organizational than
of individual responses to the risk of negative sanctions.
[INSERT TABLE 2 HERE]
Table 3 presents the fixed-effect models of orchestra conventionality that test
Hypothesis 2. The results differ sharply from our findings with the alignment model: in
particular, Model 6 fails to show that middle-status orchestras have, on average, any
propensity to adopt more conventional repertoires than other orchestras. In contrast, Model 7
indicates that middle-status musical directors tend to favor conventional programing choices
while high- and low-status musical directors underperform popular composers (there is a
statistically significant negative coefficient on the quadratic director status variable), which
supports H2. Including terms for musical director and orchestra status in the same model
(Model 8) confirms these results, and supports H3b: conventionality as a response to positive
rewards associated with quality signaling is more sensitive to individual than to organizational
status.
These result are consistent with the mechanisms underlying H2: all things equal,
middle-status musical directors seek quality recognition and therefore choose to perform
works of the most popular composers, as audiences can more readily evaluate and compare
the quality of performances. On the other hand, high- and low-status musical directors tend to
downplay these choices, although for different reasons: high-status actors tend to disregard
30
conventions and emphasize rare programing choices, while low-status actors seek to avoid
potentially unflattering comparisons with more renowned peers. If this interpretation is
correct, that low status actors avoid quality comparison more than high status actors, the
repertoires of low-status musical directors should overlap and align significantly less on
average than those of high-status directors with the repertoires of other orchestras. We tested
this prediction by comparing the repertoire choices of musical directors with status scores in
the bottom third and top third of the status distribution against other orchestras’ repertoires.
The former had a 49.8% overlap and 3.9 alignment score with other orchestras' repertoires,
compared with a 55% overlap and 4.17 alignment score for the repertoires of musical
directors in the top third of the status distribution. These differences are statistically
significant at the p=.001 level (Welch two-sample t-test, t= -7.76 for overlap; t=-5.92 for
alignment) which corroborates our interpretations of results from Table 3 and support the idea
that low status directors seek to avoid quality comparisons.
[INSERT TABLE 3 HERE]
Table 4 presents the robustness checks for the alignment and conventionality models.
Model 9 tests the robustness of our findings by adding a lagged dependent variable as a
predictor in the alignment model. The quadratic effect on orchestra status remained
observable in the data. Because random-effect models offer more efficient (though possibly
biased) estimates than fixed-effect models, Model 10 replaces the orchestra and director
dummy variables with a two-way random intercept specification, but the results remained
unchanged in direction and statistical significance. The null finding regarding the effect of
musical director status thus cannot be attributed simply to the efficiency properties of our
fixed-effect models. Finally, because the status measure used in our analysis gives an
advantage to orchestras and directors that premiere high numbers of new composers works,
31
Model 11 tests the performance of the alignment model when including, on the right-hand
side of the equation, quadratic and linear terms on the cumulative numbers of new composers
introduced by orchestras and directors. Although the statistical significance of the coefficient
on the orchestra’s status quadratic term decreases, the model yields a small and statistically
non-significant increase in the likelihood-ratio statistic relative to Model 4 (chi-square statistic
= 7.534, for 4 additional degrees of freedom; p-value = .110), which allows us to retain the
more parsimonious Model 4. Our measure of status is therefore sufficient to account for
differences in alignment.
The findings of the conventionality model remain qualitatively unchanged (in sign and
statistical significance), even when we include a lagged dependent variable in the model
(Model 12) or use a random-effect specification (Model 13). Model 14 adds cumulative
innovation as a predictor to the model, but our results remain unchanged: middle-status
directors appear to be more conventional on average; differences in conventionality among
middle and high- or low-status directors are not simply attributable to differences in their
propensity to premiere new composers.
[INSERT TABLE 4 HERE]
Additional Analyses. As an extension of our test of Hypotheses 3a and 3b, we also checked
for the presence of interaction effects between director and orchestra status in the alignment
and conventionality models in Table 5. Regarding alignment, Phillips and Zuckerman discuss
the possible “mitigation of the insecurity” felt by middle-status individuals working at high-
status organizations (2001: 418). We could therefore predict that high-status orchestras
provide musical directors greater protection against negative evaluations, while low-status
orchestras are unable to buffer these penalties. As a consequence, although models 3 and 4 do
not detect any significant effect of individual status on alignment on average (consistent with
32
Hypotheses 1 and 3a), this result could hide substantial heterogeneity in individual status
effects across orchestras: if high-status orchestras buffer penalties, the effect of individual
status on alignment should be more pronounced in low-status orchestras. Still, we find that the
interaction terms between individual status and linear orchestra status are not statistically
significant: our models do not detect any buffering effect of organizational status on
alignment (Models 15 and 16).
However, according to the conventionality model (Model 18), organizational status
moderates the curvilinear effect of individual status on conventionality.
10
Figure 3 represents
the director status effect graphically in low-, medium- and high-status orchestras. Although
the relationship between director status and conventionality retains its inverted-U shape, we
observe that high-status directors behave significantly more conventionally in high-status
orchestras. This result can be interpreted as showing that although orchestra status does not
buffer penalties incurred by directors in case of misalignment, it could amplify the rewards
that they are able to reap through conventionality. Hence, in high-status orchestras, musical
directors have greater incentives to compete on performance quality as this strategy
guarantees outsized rewards. Table 6 offers more detail on the director’s impact on
conventionality (presented in Figure 3) by showing predicted conventionality levels by
musical director and orchestra status, along with the statistical significance of the difference
in predicted conventionality between low-/high- and middle-status directors. This table shows
that as orchestra status increases, high-status directors and, to a smaller extent, low-status
directors start adopting more conventional programs.
10
Model 17 includes interactions between orchestra status and both linear and quadratic terms of director status.
This model fails to estimate the interaction terms with sufficient precision, and does not increase model fit
significantly relative to Model 18 (Chi-squared statistic = 0.108 with 1 additional degree of freedom, p = .742), a
more parsimonious model, which includes only one interaction between orchestra status and the quadratic term
of director status.
33
[INSERT TABLES 5 & 6 HERE]
[INSERT FIGURE 3 HERE]
Finally, although alignment and conventionality are not empirically correlated (Table
1), they are conceptually linked: as alignment increases, it becomes more difficult for
organizations to exhibit extreme (high or low) levels of conventionality. As a consequence,
we expect a negative moderation of orchestra repertoire alignment in the relationship between
director status and conventionality. Table 5 shows that the coefficient of the interaction is
negative and marginally significant in Model 19. The predicted effect of musical director
status on conventionality along with its 90-percent confidence interval is shown in Figure 4, at
various levels of repertoire alignment (low, medium and high alignment at the 10th percentile,
mean value and 90th percentile, respectively). The effect of a musical director's status on
conventionality is comparatively greater (i.e., has a more pronounced inverted U-shaped
curve) in imperfectly aligned repertoires; it becomes less important for highly aligned
repertoires, which stick closely to average programing choices and thus offer little room for
directors’ personal bias or initiative. This result is confirmed in Table 6 bottom panel: high
and low-status individual leaders adopt noticeably and significantly less conventional
repertoires than middle-status leaders when their organizations are misaligned; but in highly
aligned organizations, our model cannot predict any significant differences in conventionality
between high-, middle-, and low-status leaders.
[INSERT FIGURE 4 HERE]
DISCUSSION AND CONCLUSION
In this paper, we identify the distinct influences of status on the ways in which
organizations become more aligned or conventional -- two forms of conformity that were
34
conflated in prior works. In doing so, we suggest that organizations conform in two ways: by
adopting entire sets of attributes (i.e., traits and behaviors) to look like their peers and blend
in; and by focusing on a limited number of salient attributes critical to the definition of their
field or industry. We relate organizational and individual status to each form of conformity
and disentangle some relationships that have remained confounded in prior studies. By
analyzing major U.S. symphony orchestras and their programming norms over 90 years, we
find that alignment is more likely for middle-status organizations while conventionality is
driven by middle-status leaders. Our distinction between alignment and conventionality
contributes to three literatures.
First, the decomposition of the two forms of conformity helps map out different
mechanisms (penalty avoidance and quality comparison; attribution biases and motivational
asymmetry) and where they are most likely to be at play. We expand the middle-status
conformity theory, relax a constraint of the two-stage model (membership identification
followed by assessment) and provide a model that applies to cases in which audiences
evaluate alignment and conventionality concurrently. We show that the mechanisms leading
to conformity are not activated identically for alignment and conventionality and for high-,
medium- and low-status leaders and organizations. Conformity is not only a response to the
organizational imperative to avoid sanctions; it is also a way for individual leaders to orient
their choices and focus organizational resources on some salient attributes of their field or
industry and reap a large share of the benefits from distinctiveness. Therefore, a possible
answer to the trade-off between conformity and distinctiveness that opened this paper is that
conventionality enables organizations to conform to norms and expectations and leaves room
for misalignment and innovation. Nevertheless, as alignment increases, there are less
opportunities for individual leaders to adopt conventional choices.
35
Second, by studying the influence of individual and organizational status on
organizational outcomes concurrently, we contribute to the status literature. Whereas existing
status studies define and test status mechanisms separately at the individual level (Rossman,
Esparza, and Bonacich, 2010; Battilana, 2011; Bendersky and Hays, 2012; Bendersky and
Shah, 2012), or at the organizational level (Castellucci and Ertug, 2010, Zhao and Zhou,
2011, Cowen 2012), our research responds to the call for more studies to analyze individual
and organizational status effects jointly. Taken together, for instance, our results show that the
programming choices of middle-status organizations exhibit greater alignment, while the
choices of middle-status directors are more likely to be conventional. These findings are
consistent with the idea that programming decisions are guided not only by the individual
motivations of musical directors but also by the organizational constraints faced by orchestras.
In other words, to the extent that musical directors exercise their discretion over orchestra
programming choices, their decisions are a function of their personal interests (to seek
positive rewards) and their responsibility toward the organization (to avoid negative
sanctions).
11
The combination of alignment and conventionality in response to conformity
pressures helps individual decision-makers resolve the tension between their interests and the
interests of their organization. Beyond highlighting the need to distinguish between individual
and organizational-level status mechanisms, our additional findings (Table 5) delineate ways
in which status effects can interact across levels of analysis. In particular, our analyses show
that orchestra status moderates the curvilinear effect of director status on conventionality:
organizational status appears to amplify the rewards reaped by individual leaders. This paper
11
For example, the Chicago Symphony orchestra's middle status in the 1890s called for aligned programming
(Levine, 1988: 119; Otis, 1924). However, the status of its musical director, Theodore Thomas, pushed the
Chicago Symphony’s repertoire toward unconventional choices. Indeed, the recognition that Thomas benefitted
from could allow him to make more unconventional decisions than another musical director, but he still needed
to attend to his organization's fundamental need to be recognized as a major orchestra and therefore emphasized
the breadth of his repertoire as a way to stay aligned with other orchestras.
36
thus offers a first step toward an integrated view of status effects on organizational conformity
or deviance.
Third, existing research primarily approaches conformity by investigating membership
norms and the categorical imperative—that is, the compelling need for organizations to
coalesce around the typical attributes of the category of which they claim to be members
(Hannan et al., 2007; Hsu, 2006; Hsu et al. 2009; Negro et al., 2010; Ody-Brasier &
Vermeulen, 2014). The study of membership norms provides a persuasive answer to the
fundamental question of what makes organizations so similar (DiMaggio & Powell 1983;
Hannan et al., 2007; Phillips et al, 2013), but it should not obscure the fact that, beyond
membership, norms allow for observable variations between organizations and thus
opportunities for distinction within fields and industries. When organizations exhibit low
levels of “grade of membership” in their category, or deviate from their category's prototype,
they could still be striving to achieve conventionality, rather than simply disregarding the
categorical imperative. Therefore, this paper sets the stage for rethinking our understanding of
conformity beyond the "golden cage" of categorical membership norms within which
organizations are held (Durand & Paolella, 2013, Wry, Lounsbury & Jennings 2014).
Hence, our study invites a re-examination of existing theories of the consequences of
conformity and deviance, which appear mis-specified or incomplete when we consider the
combined presence of alignment and conventionality. These theories effectively emphasize
the influence of organizational characteristics on conformity (or deviance) through fear of
sanctions, but fail to consider the extent to which organizations, guided by individual actors,
choose to over- or underrepresent salient attributes (i.e., iconic or recognizable manifestations
of normal behaviors) in their efforts to pursue positive rewards and social recognition. Thus,
the extant literature may have underplayed the importance of the search for social and quality
37
comparison (or its avoidance by low-status actors) in the study of the consequences of
conformity.
Empirically, our operationalization of alignment and conventionality has traction, as
our results show that the role of status varies with the actor (organization or individual) and
the form of conformity. In most prior studies, measures of norm conformity rely on
comparisons of the frequencies of behaviors over time, which provide proxies for alignment.
We demonstrate the need for complementary measures, such as the tendency to adopt highly
salient attributes (i.e., conventionality). In addition, our definitions of alignment and
conventionality remain agnostic about the precise nature of prescribed behaviors or choices;
the resulting measures do not rely on any ex ante definition of what constitutes the musical
canon. Instead, the prescriptive elements of the norm derive from our data, not any a priori
criterion as it is unfortunately common in many evolutionary studies (Durand, 2006: 134-35).
Therefore, this study paves the way for further explorations of other antecedents of
conformity and their consequences (i.e., reputation, performance, and strategic behavior), at
both organizational and individual levels, in a wide range of normative contexts.
Although our findings support our theoretical intuition, future research should
replicate these findings in other fields with supplementary characteristics, such as technical
evolutions or competing institutional logics, which are likely to have a strong influence on
norm evolution. Three avenues seem especially promising. First, different audiences may set
different normative expectations (Pontikes, 2012; Rossman & Schilke, 2014); thus, a
refinement of our model could test for the importance of audience heterogeneity as a
determinant or moderator of organizational alignment and conventionality. Second, individual
or organizational characteristics other than status might affect individual and organizational
responses to negative sanctions and positive rewards, which in turn could influence alignment
38
and conventionality. Third, although the literature has recently recognized the importance of
integrating the levels of analysis when studying norm evolution (Wright & Zammuto, 2013),
our analysis does not attempt to explain norm dynamics and considers only conformity with a
contemporaneous norm. Medium- to fast-paced changes in normative environments lead to
questions regarding the discrepancies between the historical norms and contemporary
imperatives to which organizations must conform; research into these discrepancies could
refine further our understanding of the antecedents and the consequences of norm conformity.
39
REFERENCES
Arian, E. 1971. Bach, Beethoven, and bureaucracy: The case of the Philadelphia Orchestra.
University, AL: University of Alabama Press.
Battilana, J. 2011. The enabling role of social position in diverging from the institutional
status quo: Evidence from the UK national health service. Organization Science, 22(4):
817–834.
Battilana, J., & Casciaro, T. 2012. Change agents, networks, and institutions: A contingency
theory of organizational change. Academy of Management Journal, 55(2): 381–398.
Bendersky, C., & Hays, N. A. 2012. Status conflict in groups. Organization Science, 23(2):
323–340.
Bendersky, C., & Shah, N. P. 2012. The cost of status enhancement: Performance effects of
individuals’ status mobility in task groups. Organization Science, 23(2): 308–322.
Benjamin, B. A. & Podolny, J. M. 1999. Status, quality, and social order in the California
wine industry. Administrative Science Quarterly. 44(3): 563--589.
Bonacich, P. 1987. Power and Centrality: A Family of Measures. American Journal of
Sociology. 92(5): 1170--1182.
Bonacich, P., & Lloyd, P. 2001. Eigenvector-like measures of centrality for asymmetric
relations. Social Networks, 23: 191–201.
Bothner, M. S., Kim, Y.-K., & Smith, E. B. 2012. How does status affect performance? Status
as an asset vs. Status as a liability in the pga and nascar. Organization Science, 23(2):
416–433.
Bothner, M. S., Smith, E. B., & White, H. C. 2010. A model of robust positions in social
networks. American Journal of Sociology, 116(3): 943–92.
Bourdieu, P. 1993. The field of cultural production. New York: Columbia University Press.
Brewer, M. B. 1991. The social self: On being the same and different at the same time.
Personality and Social Psychology Bulletin, 17(5): 475–482.
Castellucci, F., & Ertug, G. 2010. What’s in it for them? Advantages of higher-status partners
in exchange relationships. Academy of Management Journal, 53(1): 149–166.
Chevalier, J. & Ellison, G. 1997. Risk Taking by Mutual Funds as a Response to Incentives.
Journal of Political Economy. 105(6): pp. 1167-1200.
Chevalier, J. & Ellison, G. 1999. Career Concerns of Mutual Fund Managers. The Quarterly
Journal of Economics. 114(2): 389--432.
Christensen, Clayton M. 1997. Innovator's Dilemma. Cambridge, MA: Harvard Business
Review Press.
Cowen, A. P. 2012. An expanded model of status dynamics: The effects of status transfer and
interfirm coordination. Academy of Management Journal, 55(5): 1169–1186.
CPANDA (Cultural Policy and the Arts Data Archive). 2008. American symphony orchestra
repertoires 1842-1970. Princeton University, Princeton NJ:
http://www.cpanda.org/cpanda/getDDIsummary.xq?studyID=a00225.
40
Craven, R. R. 1986. Symphony orchestras of the United States. Westport, CT: Greenwood
Press.
Deephouse, D. L. 1999. To be different, or to be the same? It’sa question (and theory) of
strategic balance. Strategic Management Journal, 20(2): 147–166.
DiMaggio, P. J. 1982. Cultural entrepreneurship in nineteenth-century Boston: The creation of
an organizational base for high culture in america. Media Culture Society, 4(1): 33–50.
DiMaggio, P. & Powell, W. 1983. The iron cage revisited: Institutional isomorphism and
collective rationality in organizational fields. American Sociological Review. 48(2): 147-
-160.
DiMaggio, P. J. & Stenberg, K. 1985. Why do some theaters innovate more than others? An
empirical analysis. Poetics. 14: 107--122.
Durand, R. 2006. Organizational Evolution and Strategic Management. London: Sage
Durand, R. & Jourdan, J. 2012. Jules or Jim: Alternative conformity to minority logics.
Academy of Management Journal. 55(6): 1295--1315.
Durand, R., Rao, H., & Monin, P. 2007. Code and conduct in French cuisine: Impact of code
changes on external evaluations. Strategic Management Journal, 28(5): 455–472.
Durand, R. & Paolella, L. 2013. Category stretching: Reorienting research on categories in
strategy, entrepreneurship, and organization theory. Journal of Management Studies.
50(6): 1100-1123.
Espeland, W. N. & Sauder, M. 2007. Rankings and Reactivity: How Public Measures
Recreate Social Worlds. American Journal of Sociology. 113(1): 1--40.
Ewen, D. 1943. Dictators of the Baton. New York:Alliance Book Corporation.
Glynn, M. A. 2005. Maestro or Manager? Examining the Role of the Music Director in a
Symphony Orchestra: 57-69. In The Business of Culture: Strategic Perspectives on
Entertainment and Media, edited by Lampel, J., Shamsie, J., and Lant, T. Mahwah,
NJ:Erlbaum
Hannan, M. T., Pólos, L., & Carroll, G. R. 2007. Logics of organization theory: Audiences,
codes, and ecologies. :Princeton University Press.
Hargadon, A. B. & Douglas, Y. 2001. When innovations meet institutions: Edison and the
design of the electric light. Administrative Science Quarterly. 46(3): 476--501.
Hart, P. 1973. Orpheus in the new world: The symphony orchestra as an American cultural
institution. New York: Norton.
Hayward, M. L., Rindova, V. P., & Pollock, T. G. 2004. Believing one's own press: The
causes and consequences of CEO celebrity. Strategic Management Journal. 25(7): 637--
653.
Horowitz, J. 2005. Classical music in America : A history of its rise and fall. New York:
Norton.
Howe, M. D. 1914. The Boston Symphony Orchestra: A Historical Sketch. New
York:Houghton Mifflin Company.
41
Hsu, G. 2006. Jacks of all trades and masters of none: Audiences' reactions to spanning genres
in feature film production. Administrative science quarterly. 51(3): 420--450.
Hsu, G., Hannan, M. T., & Koçak, Ö. 2009. Multiple category memberships in markets: An
integrative theory and two empirical tests. American Sociological Review. 74(1): 150--
169.
Hsu, G. & Grodal, S. Forthcoming. Category Taken-for-grantedness as a Strategic
Opportunity: The Case of Light Cigarettes, 1964-1993. American Sociological Review.
Johnson, H. E. 1950. Symphony Hall, Boston. Little:Boston.
Kim, B. K., & Jensen, M. 2011. How product order affects market identity repertoire ordering
in the US opera market. Administrative Science Quarterly, 56(2): 238–256.
Kremp, P.-A. 2010. Innovation and selection: symphony orchestras and the construction of
the musical canon in the United States (1879–1959). Social Forces, 88(3): 1051–1082.
Kupferberg, H. 1970. Those Fabulous Philadelphians: The Life and Times of a Great
Orchestra. :Virgin Books.
Lebrecht, N. 1991. The maestro myth: Great conductors in pursuit of power. New York:
Simon & Schuster.
Lehmann, H. & T. 1992. Out of the cultural dustbin: Sentimental musings on art and music
in Seattle from 1936 to 1992. :Crowley Associates.
Levine, L. W. 1988. Highbrow/lowbrow : The emergence of cultural hierarchy in America.
Cambridge, MA:Harvard University Press.
Lounsbury, M. & Glynn, M. A. 2001. Cultural entrepreneurship: stories, legitimacy, and the
acquisition of resources. Strategic Management Journal. 22(6-7): 545--564.
McNamara, G. & Bromiley, P. 1997. Decision making in an organizational setting: Cognitive
and organizational influences on risk assessment in commercial lending. Academy of
Management Journal. 40(5): 1063--1088.
Mueller, K. H. 1973. Twenty-seven major American symphony orchestras: A history and
analysis of their repertoires. Bloomington, IN: Indiana University Press.
Negro, G., Koçak, Ö., & Hsu, G. 2010. Research on categories in the sociology of
organizations. Research in the Sociology of Organizations. 31(): 3--35.
Negro, G. & Leung, M. D. 2013. ``Actual'' and perceptual effects of category spanning.
Organization Science. 24(3): 684-696.
Ody-Brasier, A. & Vermeulen, F. 2014. The Price You Pay Price-setting as a Response to
Norm Violations in the Market for Champagne Grapes. Administrative Science
Quarterly. 59(1): 109--144.
Oliver, C. 1991. Strategic Responses to Institutional Processes. Academy of Management
Review. 16(1): 145--179.
Otis, P. A. 1924. The Chicago Symphony Orchestra, its organization, growth and
development, 1891-1924. Chicago:Clayton F. Summy.
Peyser, J. 2000. The orchestra: origins and transformations. New York:Scribner.
42
Phillips, D. J., Turco, C. J., & Zuckerman, E. W. 2013. Betrayal as market barrier: Identity-
based limits to diversification among high-status corporate law firms. American Journal
of Sociology. 118(4): 1023-54.
Phillips, D. J., & Zuckerman, E. W. 2001. Middle-status conformity: Theoretical restatement
and empirical demonstration in two markets. American Journal of Sociology, 107: 379–
429.
Podolny, J. M. 2005. Status signals : A sociological study of market competition. Princeton,
NJ: Princeton University Press.
Pontikes, E. G. 2012. Two Sides of the Same Coin: How Ambiguous Classification Affects
Multiple Audiences' Evaluations. Administrative Science Quarterly. 57(1): 81--118.
Porter, M. 1996. What is Strategy? Harvard Business Review. (6): 61-78.
Rao, H., Monin, P., & Durand, R. 2005. Border crossing: Bricolage and the erosion of
categorical boundaries in French gastronomy. American Sociological Review, 70(6):
968–991.
Rossman, G., Esparza, N., & Bonacich, P. 2010. I’d like to thank the Academy, team
spillovers, and network centrality. American Sociological Review, 75(1): 31–51.
Rossman, G. & Schilke, O. 2014. Close, But No Cigar: The Bimodal Rewards to Prize-
Seeking. American Sociological Review. 79(1): 86--108.
Sauder, M. 2008. Interlopers and Field Change: The Entry of U.S. News into the Field of
Legal Education. Administrative Science Quarterly. 53(2): 209-234.
Sauder, M., Lynn, F., & Podolny, J. M. 2012. Status: Insights from organizational sociology.
Annual Review of Sociology, 38: 267–283.
Seltzer 1975. The Professional symphony orchestra in the United States. Metuchen,
NJ:Scarecrow Press.
Sgourev, S. V. & Althuizen, N. 2014. ``Notable'' or ``Not Able'' When Are Acts of
Inconsistency Rewarded? American Sociological Review. 79(2): 282--302.
Sharkey, A. J. 2014. Categories and Organizational Status: The Role of Industry Status in the
Response to Organizational Deviance. American Journal of Sociology. 119(5): 1380--
1433.
Smith, E. B. 2011. Identities as lenses: How organizational identity affects audiences’
evaluation of organizational performance. Administrative Science Quarterly, 56(1): 61–
94.
Swoboda, H. 1967. The American symphony orchestra. New York: Basic Books.
Tirole, J. 1988. The theory of industrial organization. Cambridge, MA: MIT press.
Vergne, J.-P. 2012. Stigmatized Categories and Public Disapproval of Organizations: A
Mixed-Methods Study of the Global Arms Industry, 1996--2007. Academy of
Management Journal. 55(5): 1027--1052.
Vergne, J. & Wry, T. 2014. Categorizing Categorization Research: Review, Integration and
Future Directions. Journal of Management Studies.
43
Waguespack, D. M. & Sorenson, O. 2011. The ratings game: Asymmetry in classification.
Organization Science. 22(3): 541--553.
Weber, W. 2001. From miscellany to homogeneity in concert programming. Poetics. 29(2):
125--134.
Wright, A. L. & Zammuto, R. F. 2013. Wielding the willow: Processes of institutional change
in English county cricket. Academy of Management Journal. 56(1): 308--330.
Wry, T., Lounsbury, M., & Jennings, P. D. 2014. Hybrid vigor: Securing venture capital by
spanning categories in nanotechnology. Academy of Management Journal. 57(5): 1309-
1333.
Zhao, W., & Zhou, X. 2011. Status inconsistency and product valuation in the California wine
market. Organization Science, 22(6): 1435–1448
Zuckerman, E. W. 1999. The categorical imperative: securities analysts and the illegitimacy
discount. American Journal of Sociology. 104(5): 1398--1438.
44
Table 1. Descriptive Statistics and Correlation Matrix
Mean
SD
Min.
Max.
(1)
(2)
(3)
(4)
(5)
(6)
(1) Alignment
4.06
0.50
2.19
5.30
1.00
(2) Conventionality
0.96
0.21
0.32
1.91
0.02
1.00
(3) Orchestra status
0.05
0.97
-1.61
2.78
0.43
-0.14
1.00
(4) Director status
0.09
1.00
-1.19
4.29
0.23
-0.02
0.51
1.00
(5) Director tenure
6.39
7.06
0.00
42.00
0.16
0.00
0.27
0.63
1.00
(6) log(performances)
4.14
0.51
0.00
5.11
0.62
-0.03
0.55
0.38
0.25
1.00
45
Table 2. Linear Models of Alignment with Orchestra, Director and Decade Fixed
Effects
Variable
Model 1
Model 2
Model 3
Model 4
Orchestra status t-1 (standardized) 2
-0.067*
-0.052^
(-2.433)
(-1.809)
Orchestra status t-1 (standardized)
0.114*
0.143**
(2.510)
(3.006)
Director status t-1 (standardized) 2
-0.004
0.003
(-0.374)
(0.279)
Director status t-1 (standardized)
-0.040
-0.072
(-0.955)
(-1.633)
Log(number of performances t-1)
0.220***
0.212***
0.213***
0.204***
(6.287)
(6.035)
(6.050)
(5.817)
Director tenure t (years)
-0.004
-0.004
-0.001
0.000
(-0.993)
(-0.863)
(-0.189)
(0.088)
Orchestra
FE
FE
FE
FE
Musical director
FE
FE
FE
FE
Decade
FE
FE
FE
FE
Likelihood-ratio statistic1 (d.f.)
12.371**
(df=2)
5.387^
(df=2)
18.693***
(df=4)
R2
0.665
0.669
0.666
0.670
Notes: N = 1090 orchestra-years, including 27 orchestras and 154 musical directors. The t-ratios are in
parentheses. FE = fixed effects.
^ p < .10. * p < .05. ** p < .01. *** p < .001.
1 Likelihood-ratio chi-squared statistics use Model 1 as a baseline.
46
Table 3. Linear Models of Conventionality with Orchestra, Director and Decade Fixed
Effects
Variable
Model 5
Model 6
Model 7
Model 8
Orchestra status t-1 (standardized)2
-0.005
-0.001
(-0.321)
(-0.050)
Orchestra status t-1 (standardized)
-0.019
-0.038
(-0.725)
(-1.448)
Director status t-1 (standardized)2
-0.019**
-0.020***
(-3.230)
(-3.358)
Director status t-1 (standardized)
0.069**
0.079**
(2.920)
(3.215)
Log(number of performances t-1)
0.000
0.000
-0.002
-0.003
(0.079)
(0.201)
(-0.961)
(-1.034)
Director tenure t (years)
-0.027
-0.027
-0.027
-0.026
(-1.385)
(-1.392)
(-1.384)
(-1.314)
Orchestra
FE
FE
FE
FE
Musical director
FE
FE
FE
FE
Decade
FE
FE
FE
FE
Likelihood-ratio statistic1 (d.f.)
-
0.893
(df=2)
12.905**
(df=2)
15.510**
(df=4)
R2
0.416
0.417
0.423
0.425
Notes: N = 1090 orchestra-years, including 27 orchestras and 154 musical directors. The t-ratios are in
parentheses. FE = fixed effects.
^ p < .10. * p < .05. ** p < .01. *** p < .001.
1 Likelihood-ratio chi-squared statistics use Model 5 as a baseline.
47
Table 4. Robustness Checks: Alignment and Conventionality Models.
Alignment
Conventionality
Variable
Model 9
Model 10
Model 11
Model
12
Model 13
Model 14
Orchestra status t-1(std.)2
-0.051^
-0.038*
-0.025
-0.001
-0.009
-0.008
(-1.792)
(-2.240)
(-0.756)
(-0.049)
(-0.923)
(-0.421)
Orchestra status t-1 (std.)
0.143**
0.106***
0.152**
-0.038
-0.015
-0.029
(3.007)
(3.497)
(3.071)
(-1.446)
(-1.002)
(-1.043)
Director status t-1(std.)2
0.004
-0.009
0.013
-0.020***
-0.012*
-0.021**
(0.327)
(-0.921)
(0.906)
(-3.349)
(-2.333)
(-2.637)
Director status t-1 (std.)
-0.071
-0.008
-0.094
0.079**
0.033^
0.071*
(-1.616)
(-0.265)
(-1.644)
(3.202)
(1.933)
(2.213)
Alignment t-1
-0.467
(-1.350)
Past orchestra innovation t-1
(std.)2
-0.030
0.024
(-0.635)
(0.896)
Past orchestra innovation t-1 (std.)
-0.198^
-0.071
(-1.945)
(-1.245)
Past director innovation t-1 (std.)2
-0.011
0.003
(-0.706)
(0.283)
Past director innovation t-1 (std.)
0.010
0.019
(0.186)
(0.628)
Conventionality t-1
0.001
(0.023)
Director tenure t (years)
0.000
-0.001
0.005
-0.003
-0.001
-0.003
(0.028)
(-0.480)
-0.025
(-1.302)
(-0.808)
(-1.013)
Log(number of performances t-1)
0.256***
0.281***
0.182***
-0.026
-0.022
-0.032
(4.934)
(9.244)
(4.93)
(-1.313)
(-1.386)
(-1.558)
Orchestra
FE
RE
FE
FE
RE
FE
Musical director
FE
RE
FE
FE
RE
FE
Decade
FE
FE
FE
FE
FE
FE
Likelihood-ratio statistic1 (d.f.)
20.900***
(df=5)
10.156*
(df=4)
26.227***
(df=8)
15.510**
(df=5)
20.701***
(df=4)
20.171***
(df=8)
R2
0.671
-
0.673
0.425
-
0.427
Notes: N = 1090 orchestra-years, including 27 orchestras and 154 musical directors. The t-ratios are in
parentheses. FE = fixed effects, RE = random effects.
^ p < .10. * p < .05. ** p < .01. *** p < .001.
1 Likelihood-ratio chi-squared statistics for all fixed-effect models use Model 1 and 5 as a baseline;
the deviance of Model 10 and 13 are compared to equivalent random-effect models including only
control variables and decade dummies.
48
Table 5. Additional Analyses: Moderation of Director Status Effect in
Alignment and Conventionality Models.
Notes: N = 1090 orchestra-years, including 27 orchestras and 154 musical directors. The t-ratios are in
parentheses. FE = fixed effects; RE = random effects. ^ p < .10. * p < .05. ** p < .01. *** p < .001.
1 Likelihood-ratio chi-squared statistics compare models to equivalent models including only control
variables and the corresponding fixed and/or random effects.
Alignment
Conventionality
Variable
Model 15
Model 16
Model 17
Model 18
Model 19
Orchestra status t-1 (std.) 2
-0.078*
-0.057^
-0.012
-0.009
-0.001
(-2.243)
(-1.941)
(-0.619)
(-0.543)
(-0.049)
Orchestra status t-1 (std.)
0.133**
0.139**
-0.045^
-0.044
-0.022
(2.760)
(2.920)
(-1.657)
(-1.634)
(-0.804)
Director status t-1 (std.) 2
-0.006
-0.007
-0.035***
-0.036***
-0.021***
(-0.317)
(-0.420)
(-3.626)
(-3.670)
(-3.337)
Director status t-1 (std.)
-0.082^
-0.071
0.079**
0.081*
0.070**
(-1.816)
(-1.602)
(3.140)
(3.295)
(2.826)
Dir. status t-12 × Orch. status t-1
-0.002
0.006
0.007
0.009*
(-0.228)
(0.762)
(1.240)
(2.006)
Dir. status t-1 × Orch. status t-1
0.038
0.006
(1.125)
(0.299)
Alignment t (standardized)
-0.027*
(-2.503)
Dir. status t-12 ×Alignmentt
0.008
(1.562)
Dir. statust-1 × Alignmentt
-0.025^
(-1.856)
Log(number of perf. t-1)
0.209***
0.207***
-0.003
-0.003
-0.002
(5.915)
(5.864)
(-1.118)
(-1.116)
(-0.956)
Director tenure t (years)
0.0002
0.0003
-0.021
-0.021
-0.016
(0.047)
(0.057)
(-1.072)
(-1.089)
(-0.786)
Orchestra
FE
FE
FE
FE
FE
Musical director
FE
FE
FE
FE
FE
Decade
FE
FE
FE
FE
FE
Likelihood-ratio statistic1 (chi-
squared)
20.930**
(df=6)
19.397**
(df=5)
20.481**
(df=6)
20.372***
(df=5)
25.372***
(df=7)
R2
0.671
0.671
0.427
0.427
0.430
49
Table 6. Predicted Conventionality by Musical Director Status in Models with Alignment and
Orchestra Status Interactions.
Mean predicted conventionality
Low-status
directors
Middle-status
directors
High-status
directors
Orchestra status (Model 18)
Low
0.860**
0.999
0.629**
Medium
0.867**
1.019
0.739**
High
0.882**
1.061
0.973*
Orchestra alignment (Model 19):
Low
0.839***
1.056
0.870*
Medium
0.883**
1.032
0.915*
High
0.924
1.010
0.958
Predicted conventionality values from Models 18 and 19 for a hypothetical orchestra, taking averaged
fixed effects and control variables. For low and high-status directors, stars (^ p < .10, * p < .05, ** p <
.01, *** p < .001) indicate the probability Pr{predicted difference > 0}, i.e. the probability that low-
or high-status directors are predicted as more conventional than middle-status ones, in Model 18 and
Model 19.
Technical note: Predicted differences between high/low and middle-status directors are obtained by
randomly drawing parameter vectors from the multivariate normal distribution of coefficients on
director status variables (quadratic, linear, and interaction terms) estimated in the model (n=10,000
draws). For each draw, conventionality differences were calculated between low-, middle-, and high-
status musical directors (status = -1, +1.70, and +4 s.d. from the mean) three orchestra status
conditions (orchestra status set respectively at its P10, mean, and P90 value for low, medium and high
status) and under three orchestra alignment conditions (orchestra alignment set respectively at its P10,
mean, and P90 value for low, medium, and high alignment respectively).
Interpretation top panel: Model 18 predicts that in a low-status orchestra, high-status directors choose
repertoires with 0.629 conventionality on average (which differs significantly from the 0.999
conventionality of middle-status directors at the p=.001 level); the predicted conventionality of high-
status directors increases to 0.973 in high-status orchestras (which differs from the 1.061
conventionality predicted for middle-status directors, at the p=.05 level).
Interpretation bottom panel: Model 19 predicts that in orchestras with low alignment, high status
directors choose repertoires with .870 conventionality on average (which differs significantly from the
1.056 conventionality of orchestra with a mean alignment). At high levels of orchestras’ alignment,
the conventionality values for different individual leaders’ status are no more statistically different.
50
Figure 1: Impact of Penalties and Rewards on Orchestras and Musical Directors
51
Figure 2a: Example of Contrasting Repertoires with Low vs. High Alignment Despite Similar Conventionality
0.00 0.02 0.04 0.06 0.08 0.10
0.00 0.02 0.04 0.06 0.08 0.10
Proportion of performances by composer (all other orchestras,1944)
Prop. of performances by composer (Baltimore,1944)
Low Alignment (Baltimore, 1944)
Brahms
Bach
Mussorgsky
Alignment = 3.56
Conventionality = 1
Beethoven
Wagner
Duparc
Stravinsky
Mozart
Tchaikovsky
Strauss (Richard)
0.00 0.02 0.04 0.06 0.08 0.10
0.00 0.02 0.04 0.06 0.08 0.10
Proportion of performances by composer (all other orchestras,1944)
Prop. of performances by composer (Cleveland,1944)
High Alignment (Cleveland, 1944)
Brahms Wagner
Beethoven
Alignment = 4.72
Conventionality = 1.01
Composers
Regression line (slope = conventionality)
Hypothetical 45-degree line (slope = 1)
Distance to other repertoires
Mozart
Tchaikovsky
52
Figure 2b: Example of Contrasting Repertoires with Low vs. High Conventionality Despite Similar Alignment
0.00 0.05 0.10 0.15
0.00 0.05 0.10 0.15
Proportion of performances by composer (all other orchestras,1964)
Prop. of performances by composer (Baltimore,1964)
Low Conventionality (Baltimore, 1964)
Mussorgsky
Alignment = 3.81
Conventionality = 0.62
Mozart
Brahms Beethoven
Strauss (Richard)
0.00 0.05 0.10 0.15
0.00 0.05 0.10 0.15
Proportion of performances by composer (all other orchestras,1964)
Prop. of performances by composer (San Francisco,1964)
High Conventionality (San Francisco, 1964)
Beethoven
Mozart
Alignment = 3.83
Conventionality = 1.48
Composers
Regression line (slope = conventionality)
Hypothetical 45-degree line (slope = 1)
Distance to other repertoires
Brahms
53
Figure 3. Interaction effects of musical director status and orchestra status in the conventionality model
Notes: For each level of alignment, the dark and thick curves represent the predicted effect of director status on conventionality according to the point
estimates of Model 18 (Table 5). Lighter-colored and thinner curves display quadratic curves constructed from parameter vectors randomly drawn (n = 10,000
draws) from the multivariate normal distribution of coefficients estimated in the model, within a 90% confidence interval. The color intensity of each curve is
proportional to the multivariate normal distribution's density estimated at the set of parameters: curves constructed from less likely parameters appear in
lighter colors. Therefore, unlike usual representations of predicted curvilinear effects, which only report the mean estimated curve without showing confidence
intervals, these figures also include a representation of the uncertainty about our estimates.
54
Figure 4. Interaction effects of musical director status and alignment in the conventionality model
Notes: For each level of alignment, the dark and thick curves represent the predicted effect of director status on conventionality according to the point
estimates of Model 19 (Table 5). Lighter-colored and thinner curves display quadratic curves constructed from parameter vectors randomly drawn (n = 10,000
draws) from the multivariate normal distribution of coefficients estimated in the model, within a 90% confidence interval. The color intensity of each curve is
proportional to the multivariate normal distribution's density estimated at the set of parameters: curves constructed from less likely parameters appear in
lighter colors. Therefore, unlike usual representations of predicted curvilinear effects, which only report the mean estimated curve without showing confidence
intervals, these figures also include a representation of the uncertainty about our estimates.
