|Subject Terms:||Organization theory|
|Classification Codes:||9190: US|
9130: Experimental/theoretical treatment
2500: Organizational behavior
|Copyright Academy of Management Apr 1999|
Early research on creativity centered, to a large extent, on discovering and describing the nature of creative people (e.g., Barron, 1955; MacKinnon, 1965). While noteworthy in its own right, its nearly exclusive focus on the individual level of analysis eclipsed more macro explanations of creativity (Slappendel, 1996). Amabile, working with her colleagues (e.g., Amabile, 1983, 1996; Amabile, Goldfarb, & Brockfield, 1990), enlarged the scope of creativity research from its origins at the individual level to the group or social-psychological level and, eventually, to the organizational level (Amabile, Conti, Coon, Lazenby, & Herron, 1996). In general, scholars in the field have followed this approach, and multilevel models of creativity in organizations are now emerging.
Woodman, Sawyer, & Griffin (1993) proposed one of the first multilevel models; they linked individual-, group-, and organizational-level variables to creative outcomes. Glynn (1996) presented a theoretical model that related intelligence at both the individual and organizational levels to creativity, and Ford (1996) integrated multiple levels of analysis to explain engagement in creative behavior. Oldham and Cummings (1996), in a rare and important empirical test, demonstrated that factors at multiple levels of analysis (i.e., individual, job, and organizational) affect creativity. Drazin and Schoonhoven (1996) urged the development of multilevel models that link a firm's strategic focus to the behavior of its senior managers and, in turn, to individuals' commitment to the creative process.
Given the recent focus on multilevel approaches to creativity, a timely question is one concerning how level-of-analysis (LOA) issues have been modeled. The choice of focal levels of analysis is profound and central to the development of any model; it affects the conceptual framework, research methods, locus of interest, and, consequently, the full measure of a theoretical and empirical approach to a phenomenon. All too often, however, institutionalized, takenfor-granted assumptions about LOA issues become incorporated, unquestioned, into a model. Because these assumptions are the foundation upon which theory rests, they both enable and constrain further theory development. Exposing these LOA assumptions allows for their modification and, thus, for new and alternative ways of theorizing about organizational creativity.
Drawing from the works of multilevel theorists (DiMaggio, 1991; Giddens, 1994; House, Rousseau, & Thomas-Hunt, 1995; Klein, Dansereau, & Hall, 1994; Rousseau, 1985), we examine how the creativity literature addresses LOA issues. We review relevant creativity research to reveal theoretical assumptions about the analytical levels to which generalizations are made; aggregation and composition issues across levels of analysis; treatment of subunits as heterogeneous, homogeneous, or independent of higherlevel influences; and dynamic change over time. Following this, we explore the consequences of altering LOA assumptions and propose an illustrative model. In our review of the literature, we find that a particular set of assumptions dominates the current literature on creativity. These assumptions center on a definition of creativity as an outcome and generalization primarily to the small group or project level of analysis. Our goal is to address the gap in multilevel theorizing that such assumptions may have inadvertently created.
In contrast to existing models, we define creativity as the process of engagement in creative acts, regardless of whether the resultant outcomes are novel, useful, or creative (Amabile, 1988, 1996; Ford, 1996). This process orientation focuses our inquiry on how individuals attempt to orient themselves to, and take creative action in, situations or events that are complex, ambiguous, and ill defined. In other words, this is an issue of how individuals engage in sensemaking in organizations (Greenberg, 1995; Louis, 1980; Volkema, Farquhar, & Bergmann, 1996; Weick, 1995).
A sensemaking approach to creativity affords a fresh perspective on LOA issues. Traditionally, creativity research has depicted the key levels of analysis as being individual, group, and organizational, with creativity at higher levels typically being an aggregation of creative output at lower levels (e.g., Glynn, 1996; Woodman et al., 1993). A sensemaking perspective enriches this LOA perspective by pointing to cross-level, systemic, and embedded effects that may arise from idiosyncratic and/or communal interpretations of what it means to be creative. This implies that conflict, political influence, and negotiated order may operate at more macro-organizational levels (Walsh & Fahey, 1986; Weick, 1995) and over time in organizations to influence creative processes.
A process-orientation and sensemaking perspective leads us to contexts that are expansive enough to allow a full exploration of multiple and different levels of analysis. We examine how creative thought and action unfold in an exemplar setting: an interdependent, complex, large-scale, long-duration organizational project. These projects present an "extreme situation" in which the "processes of interest are transparently observable" (Eisenhardt, 1989: 537). In addition to its theoretical significance, this setting has practical importance as well. Projects marked by interdependence, long duration, and large scale have become increasingly common in practice (e.g., Clark & Fujimoto,1988; DeMaio, Verganti, & Corso, 1994; Hoffman, 1997; Quinn & Pacquette, 1988; Sabbagh, 1996) but, to date, have been relatively understudied in creativity research.
Thus, a model of the creative processes in such a setting can coexist as a companion to the individual and small group models that have dominated the literature. Our setting allows us to make a set of logical arguments that are internally consistent with respect to this exemplar (Doty & Glick, 1994). Other contexts may vary systematically from the one we describe, but we believe that similar processes may operate in other cells of a typology that describe related creative processes in organizations.
We begin by reviewing the literature on creativity to answer the question "How have creativity researchers approached the LOA issue in developing their models?" We analyze the current literature to extract underlying themes and assumptions about levels of analysis. From the outset, we recognize that not all creativity studies are of the same bent; our objective is to identify the dominant approaches to LOA issues that have framed the study of creativity, to uncover their limitations, and to suggest alternatives that might extend multilevel theorizing.
LOA ISSUES IN THE CREATIVITY LITERATURE
Creative behavior often is modeled as the result of individual characteristics and propensities (Amabile, 1996; Mumford & Gustafson, 1988), including personality factors (Barron & Harrington, 1981; Singh, 1986) and cognitive skills, such as linguistic ability, expressive fluency, convergent and divergent thinking, and intelligence (Barron & Harrington, 1981; Basadur & Finkbeiner, 1985; Basadur, Graen, & Green, 1982; Gardner, 1993; Glynn, 1996; Helson, Roberts, & Agronick, 1995; Sternberg, 1988). Scholars have found individual creativity to be highest when individuals are motivated by intrinsic engagement; challenge; task satisfaction; and goaloriented, self-regulatory mechanisms (Amabile, 1988; Amabile, Hill, Hennessey, & Tighe, 1994; Glynn & Webster, 1993; Kanfer, 1990; Kanfer & Ackerman, 1989).
From this individual level of inquiry, creativity researchers have extended their perspective to include contextual variables. They have found that settings that provide opportunities, absence of constraints (Amabile, 1988; Amabile & Gryskiewicz, 1987; Oldham & Cummings, 1996), and rewards (Mumford & Gustafson, 1988) foster creativity. A number of multilevel studies have revealed that leader style, cohesiveness, group tenure, and degree of cooperation are antecedents to creativity (King & Anderson, 1990) and research team effectiveness (Payne, 1990). Further, Andrews (1979) has found that the composition of groups is a determinant of recognition and effectiveness, as well as of publications, for research and development teams.
This work notwithstanding, organizationallevel variables have been incorporated into relatively few models of creativity. Some studies demonstrate that organizational policies, structures, climate (Burkhardt & Brass, 1990; Tushman & Nelson, 1990), and training (Basadur, Graen, & Scandura, 1986; Wheatley, Anthony, & Maddox, 1991) affect creative output. To date, Woodman et al. (1993) offer the most comprehensive theoretical model, linking culture, resources, technology, strategy, and rewards to organizational creativity. We concur with Oldham and Cummings' statement: "Unfortunately, little is known about the conditions that promote the creative performance of individual employees in organizations" (1996: 607; emphasis added).
From this overview, we observe that the levels of generalization and construct definition employed by researchers are the most consequential factors governing the composition of creativity theories. The level of generalization is important because it specifies the focal unit to which the theoretical and empirical statements of the research apply (Rousseau, 1985). Traditionally, creativity researchers have concentrated on the small group (or independent project team) as their focal level of analysis and have emphasized project-level outcomes that combine the talents of team members into project-level efforts. With some exceptions (Glynn, 1996; Woodman et al., 1993), little has been done to extend research beyond the level of the small-group project.
Similarly, we see construct definition as consequential for theory building. In general, scholars have defined creativity as an important outcome to a system, with independent variables treated as factors to be manipulated in order to improve these outcomes. Rousseau (1985) terms this approach to multilevel theorizing functionalist-reductionist. Functionalist and reductionist arguments allow researchers to model the functional contribution of units at lower hierarchical levels to outcomes at higher levels.
This functionalist perspective dominates both the creativity and innovation literature (Drazin, 1990). For instance, Amabile (1988) defines creativity as the production of novel and useful ideas. Similarly, Oldham and Cummings (1996) define creativity as products, ideas, or procedures that are (1) novel or original and (2) relevant and useful. Ford (1996) also adopts an output orientation and views creative outcomes as those that are novel and valuable. Woodman et al. (1993) define creativity as the generation of a valuable, useful new product, service, idea, procedure, or process by individuals working together in a complex social system. Following the functional tradition, the research question posed explicitly or implicitly in most creativity studies is "How do you increase creative outputs in organizations?"
The level of generalization (to small groups) and the definition of creativity (as a novel and useful outcome) govern other critical LOA assumptions in creativity research. Rousseau (1985) offers a framework that categorizes theories into three general types of multilevel models: (1) compositional, (2) cross-level, and (3) multilevel. Compositional models are concerned with specifying the similarity of a process across multiple levels (Rousseau, 1985) or determining whether macrolevel processes are consistent with underlying microlevel processes (DiMaggio, 1991). In the creativity literature, there seems to be an implicit assumption that the creative process is alike at all levels of analysis-that is, for individuals, groups, and organizational systems (Amabile, 1988; Woodman et al., 1993). Creativity researchers readily acknowledge that there are differences among individuals that can impact creativity, but they tend to adopt a set of assumptions that allow them to treat cross-level effects (between individual and organizational levels, for instance) as homogeneous in nature (Klein et al., 1994).
For example, Amabile assumes that "major corporations select individuals who exhibit relatively high levels of these personal qualities, [and that] the variance above this baseline may well be accounted for primarily by factors in the work environment" (1988: 128). Amabile later affirms this view in her statement that "whatever an individual's talents ... the conditions under which he or she works ... can significantly increase or decrease the level of creativity produced" (1996: 17). Support for this notion can be found in Oldham and Cumming's (1996) study. These authors hypothesized a person-by-situation interaction effect on creativity and found statistical support for an ordinal relationship. That is, individuals whose dispositions make them more likely to be affected by a favorable work environment are more creative, but the creativity of all individuals is raised by a supportive environment. Thus, one assumption in creativity research, made explicitly or implicitly, is the homogeneity of higher-level (or situational) effects on individuals.
The degree of inclusion of individuals in a hierarchy of levels is an important construct in cross-level research (House et al., 1995; Rousseau, 1985). Total inclusion implies that only the group in which an individual has formal membership is influential (e.g., a functional department or an assigned project team). Partial inclusion means that an individual occupies multiple organizational roles and is influenced by membership in all of them; situational attributes can cue or make salient membership in a particular group to the exclusion of other groups (Ashforth & Mael, 1996). As Fine observes, "The assumption of a dominant [occupational] identity overly limits people's choices in constructing their work relations" (1996: 92). Although individuals may be members of an organizational work unit and a cross-functional team, they may also situate themselves in terms of an occupational identity (Trice, 1993; Van Maanen & Barley, 1984). Partial inclusiveness complicates cross-level research because effects can no longer be attributed to membership in a single hierarchical group; researchers need to account for multiple, and often competing, influences that cause individuals to situate themselves in accord with this complexity (Ashforth & Mael, 1996). The assumptions adopted by most creativity researchers seem to imply total inclusion; in many ways, this may have been a natural outgrowth of the contexts in which they have studied creativity.
Much of the research on individual creativity has been conducted in behavioral laboratories (Oldham & Cummings, 1996)-occasionally with children as subjects (Koestner, Ryan, Bernieri, & Holt, 1984; Stahl & Koser, 1978). Although some models include variables such as job design (Amabile, 1988; Hackman & Oldham, 1980; Kanter, 1988), supervisory style (Stahl & Koser, 1978; West, 1989), or performance feedback (Carson & Carson, 1993), most tend to assume total inclusiveness of an individual in a small group.
Consistent with this LOA focus, scholars tend to model creativity as a discrete task, conducted by individuals or small groups who are isolated from broader organizational and occupational pressures. This perspective applies to some organizational situations, particularly for "skunkworks" types of small, team-based new-product development that can be found in such companies as 3M, Sony, or Thermos (Dumaine, 1993; Peters, 1988). However, large, complex initiatives, characterized by long time horizons and a project management structure composed of multifunctional, interdependent teams, are increasingly more common venues for creative action (DeMaio et al., 1994). Examples include the development of new aircraft by Boeing (Sabbagh, 1996), new automobiles (Clark & Fujimoto, 1988; Quinn & Pacquette, 1988), space projects at NASA (Hoffman, 1997; Sayles & Chandler, 1971), and defense contracting (Scudder, Schroeder, Van De Ven, Seiler, & Wiseman, 1989). In all of these cases, creativity was critical to the success of the effort, but the organizational context was not that of an isolated individual or a small group unit-rather, a large-scale, long-term project. The assumption of total inclusiveness of individuals in these large, complex organizational systems seems questionable.
Finally, we note that much of the creativity literature has ignored the dimension of time. This is likely the result of a tendency of researchers to define creativity as an outcome, rather than a process (see Ford, 1996, for an important exception). This definition leads researchers toward static models that emphasize explaining variance in the dependent variable (creative outcomes), rather than examining how the dynamic process of creativity unfolds over time (Mohr, 1982).
The most explicit consideration of the role of time in understanding micro-macro relationships is contained in Giddens' (1994) theory of structuration. Central to his model is the concept of the duality of structure. His thesis is that human agency and structure cannot be understood apart from each other.
Knowledgeable individuals understand the workings of their society; in turn, they both react to and participate in their reproduction and redirection. Because individuals can intervene in events and take action to influence events, time is accorded a central role in understanding the interplay between individual and structure (Peterson, 1998). In the organizational literature, Klein et al. (1994) suggest that individuals may react homogeneously to the influence of macrolevel factors within one period of time, but they react heterogeneously across multiple periods of time.
Where Giddens differs from Klein and colleagues is in allowing for the possibility that individuals will use time to influence and change structure. This suggests that creativity may not proceed in linear, hierarchical paths but, rather, in uneven, chaotic bursts that are responses to problems that erupt over time (Kazanjian, 1988; Peterson, 1998; Quinn, 1985). As such, a model of creativity in these contexts would acknowledge the interplay between individuals and structure over time (Giddens, 1994). Uncovering such processes and understanding how they shape patterns of creative effort in large-scale projects is the impetus for our proposing a multilevel, sensemaking perspective on creative processes in organizations.
CREATIVITY AS A SENSEMAKING PROCESS
Our review reveals that two significant assumptions have governed existing multilevel approaches to creativity. First, creativity has been defined as an outcome judged by others to be novel and useful. Second, creativity theories have been primarily built at one level of generalization-that of the small project group or team-with the assumption of individual inclusiveness in a hierarchy of work units embedded in an organization. Individual and small group creativity have been assumed to be similar in composition (Amabile, 1988, 1996), and organizational creativity typically has been considered to be the accumulation of individual or small group creativity (Glynn, 1996). These assumptions have driven a prototypical treatment of other LOA issues. In this section we propose an alternative multilevel model of creativity that builds on a different set of assumptions regarding construct definition and level of generalization.
Construct Definition: Creativity As es Process
We define creativity as a process, rather than an outcome. This distinction is not unique to us (Mohr, 1982); Amabile (1988) has modeled creativity as an individual-level cognitive process consisting of multiple stages. To Torrance (1988), individual creativity is a process of sensing problems, making guesses, formulating hypotheses, communicating ideas to others, and contradicting conformity or "what is expected."
At the individual level, we define creativity as the engagement of an individual in a creative act (Ford, 1996; Torrance, 1988). Creative engagement is a process in which an individual behaviorally, cognitively, and emotionally attempts to produce creative outcomes (Kahn, 1990). For example, engineers working on a project may attempt to design an apparatus that is creative; they may collect data, consult past solutions, contemplate alternatives, propose inventive ideas, and become emotionally invested in their work. Their ideas may or may not be considered by others as creative, but the process of generating those ideas logically can be called "creativity;" in effect, creativity as a process is a necessary, but not sufficient, condition for creative outcomes. In the creativity literature, scholars generally assume, but do not explicitly model, the occurrence of such a creative process. Extant models tend to adopt a variance analysis framework (Mohr, 1982) that seeks to explain the relationship between context and outcomes, presuming that a creative process underlies the generation of creative outcomes.
Creativity is a choice made by an individual to engage in producing novel ideas; the level of creative engagement can vary from person to person and from situation to situation. An individual may choose minimal engagement, proposing simple solutions that may not be novel or useful-a behavior Ford (1996) refers to as "habitual action." Alternatively, an individual may choose to engage in a full manner, using all of his or her abilities in an effort to produce creative outcomes. To Kahn (1990), such processes of engagement (and disengagement) vary over time, ebbing and flowing from moment to moment and from day to day.
Creativity also can be defined as a grouplevel process. The complex, creative projects taken on by large organizations require the concerted engagement of many individuals, rather than just one or a few. Amabile (1988, 1996) considers group and individual processes of creativity to be of similar composition, because both involve cognitive processes of idea generation and idea testing. In settings where the object of creativity is complex and requires skills from multiple bases of expertise, it may be difficult to separate out individual- from grouplevel contributions. For example, in their study of brain storming at a product design firm, Sutton and Hargadon (1996) found that when groups address complex problems, many individuals contribute to the process, and it is difficult to assign credit to any one individual for the design outcome. Groups also go through stages that mirror the processes of individuals-that is, developing criteria, generating alternatives, modifying those alternatives, and amplifying and extending original ideas.
Individuals and groups participate in creative processes in an iterative fashion. Individuals develop ideas, present them to the group, learn from the group, work out issues in solitude, and then return to the group to further modify and enhance their ideas. The iterative, interactive nature of group creativity requires that individuals first choose to engage in individual-level creativity. We make the assumption that individuals act homogeneously within groups as they engage in creative behavior (Klein et al., 1994).
Creative processes at the organizational level may not simply aggregate from individual or group efforts; rather, they may emerge from a process of negotiating multiple and potentially competing interests between different communities or groups within the organization (e.g., those technically responsible for creative activity and those managing the creative process). At the organizational level, creativity can be described in terms of a process that maps when creative behavior occurs and who engages in creative behavior. Such issues are pursued to best advantage in contexts in which there are communities of actors engaged interdependently in large-scale creative pursuits over time. Thus, situations that allow us to relax the assumption of total inclusiveness become our focal level of analysis and the context in which we examine creative processes.
Level of Generalization: Large-Scale, LongDuration Creative Projects
Creativity increasingly occurs in the context of large-scale, long-term projects that employ multiple, interdependent teams. Sabbagh's (1996) account of the development of the Boeing 777 aircraft provides a detailed example. The project was large and of long duration: the 777 ultimately required the development of over four million discrete components and was priced at $100 million per aircraft; it took 512 years to complete.
The organization design used to complete the Boeing 777 was equally complex. A hierarchy of teams, embedded in a complex project management structure, designed and built the aircraft. Each of the major systems of the plane, such as the wings, empennage, and fuselage, had a dedicated "design-build" system team. These teams were further broken down into smaller teams responsible for each subsystem, resulting in 250 separate but highly interdependent designbuild subteams.l
We use large, long-term, complex organizational projects as an exemplar setting for several reasons. First, they represent a frequently occurring, real-world phenomenon, operating at a level of analysis in between that of the project team and the entire organization (House et al., 1995). Creativity at this level consists of the ebb and flow of creative engagement among different communities of individuals involved in the project. Second, because organizations are moving away from traditional hierarchical structures, multilevel research needs to examine organizational forms that exhibit more complex types of interdependence (House et al., 1995). In these settings, communication and coordination are handled directly between communities of individuals assigned to project teams. Third, large-scale, long-time-duration projects may be characterized by considerable situational ambiguity; the ability of a new product or service to perform as anticipated may not be known well in advance, and there may be considerable change to product specifications as the project progresses and reveals untoward obstacles. Further, the horizontal and highly interactive nature of project work may break down individuals' inclusiveness in their home functional departments, in response to numerous subproject assignments over the long life of a large project.
Situations characterized by high levels of ambiguity and low levels of inclusiveness provide individuals little in the way of guidance concerning what is correct behavior. According to House et al., such psychologically weak situations "stimulate groups to engage in collective sense making and construct their own version of reality" (1995: 94). Different communities come to a project team with different professional frames; in turn, these influence team members' interpretations of events that occur during the process of project work. In the face of situational ambiguity, individuals within different communities will consult one another to develop an interpretation of events in lieu of crossing communal boundaries.
A Sensemaking Perspective on Creativity
The goal of theory building in the interpretive or sensemaking perspective is to describe organizational life. The focus is less on understanding how to manipulate a system (so as to increase the level of creativity) than it is on understanding the processes through which individuals and organizations develop systems of meaning about creative action. With its focus on the development of meanings and how they motivate engagement and action, a sensemaking perspective is well suited to our focus on creativity as a process. Thus, our central research question becomes "How does the process of creativity unfold in organizations?"
Although functionalist perspectives dominate organizational research (Gioia & Pitre, 1990), a sensemaking perspective has made significant headway (Burrell & Morgan, 1979). A sensemaking approach has been invoked to explain a diversity of topics, including issue and agenda formation (Dutton & Dukerich, 1991; Dutton & Jackson, 1987), strategy formation in top management teams (Porac, Thomas, & Baden-Fuller, 1989), change management (Gioia & Chittipeddi, 1991; Poole, Gioia, & Gray, 1989), and technology diffusion (Barley, 1986). Researchers have studied the general innovation process using a sensemaking framework (Dougherty, 1992; Hill & Levenhagen, 1995; Ring & Rands, 1989), but we know of only limited work that applies an interpretive framework specifically to the study of creativity (Ford, 1996; Ford & Gioia, 1995).
Those with a sensemaking perspective approach LOA issues somewhat differently from those with traditional perspectives. First, although researchers recognize that individuals are the center of organizational life, those individuals are accorded a different role: they create meanings about their social setting through interactions with others (Weick, 1979). But individuals also have agency and take actions that shape their environments (Giddens, 1994; Gioia & Pitre, 1990); thus, the interpretation process is inherently dynamic.
Because individuals develop and maintain subjective interpretations of their roles in organizations, different levels of analysis emerge (Weick, 1995):
1. an intrasubjective (or individual) level;
2. an intersubjective level, between two or more individuals, that represents shared frames of reference (which may transcend formal groups or subunits); and
3. a collective level that represents the unfolding of change across intersubjective levels.
We examine how each of these levels, independently and interactively, affects creative processes over time in large-scale organizational projects. In particular, we allow for the possibility that events, which are viewed as crises, occur and how these, in turn, reframe meanings so as to shift the balance of power to favor creative activity by certain groups over others.
The Intrasubjective Level: The Development of Sensemaking Frames
When individuals are confronted with an equivocal set of events, they struggle to make sense of them (Dutton & Jackson, 1987; House et al., 1995; Peterson, 1998; Weick, 1979). According to Goffman (1974), individuals pose to themselves the question "What is it that is going on here?" Their answer determines how they will engage in that situation (Kahn, 1990). Meaning-or sense-develops about the situation, which allows the individual to act in some rational fashion; thus, meaning-or sensemaking-is a primary generator of individual action.
Scholars have labeled the meanings that individuals hold as frames (Bateson, 1972; Goffman, 1974), enactments (Weick, 1979), schemata (Poole et al., 1989), and cognitive maps (Porac et al., 1989). What is central to all of these definitions is that an individual (1) develops an intrasubjective cause-and-effect map of events, actions, and consequences; (2) places himself or herself in this map; and (3) takes action according to this map as events unfold. Frames organize meaning, motivation, and subsequent involvement and action; during any experience of work activity, an individual not only develops a sense of what is going on but also a sense of how to engage.
According to Goffman (1974), all primary frames play an important role in governing the extent to which an individual becomes involved in the activity organized by the frame; frames link events to meanings and meanings, in turn, to ensuing action. Involvement can range from complete boredom and abstention to full engagement in the activities at hand. Thus, frames, and the process of framing, can impact significantly the extent to which an individual engages in the process of creativity. For example, testing the performance of a product component could be construed to mean different things to an engineer. Depending upon his or her frame of reference, the test might be regarded as a critical step in developing an important prototype and, thus, worthy of creative engagement; conversely, it might be viewed as "make work" imposed by management and, thus, less inviting of creative engagement.
Individuals "bracket" the unpunctuated experiences of the moment using their best available frames; however, events, actions, and frames are linked together in an iterative cycle of perception (Weick, 1979). Frames are subject to modification and renewal in the face of actions taken by an individual. Accordingly, they mediate between the stream of behavior that confronts an individual and his or her actions; as mediators, they determine how an individual responds to a context. Frames include beliefs about how situations and individual actions are tied together to achieve goals (Ford, 1996). These goals can include intrinsic satisfaction with the task (Amabile, 1996) or extrinsic rewards associated with creative outcomes, including recognition. Individuals respond to situations with intentionality-that is, they predict how their actions will affect them. If the situation and desired outcomes come together in a meaningful way, a person will respond by engaging in creative acts (Ford, 1996; Kahn, 1990); otherwise, the person will limit or refrain from engaging.in creative behavior. Thus, we posit the following:
Proposition 1: An individual forms an intrasubjective frame of reference for creativity; this frame mediates between events and engagement (or disengagement) in creative acts.
The Intersubjective Level: The Construction of Shared Meaning Within Communities
An individual's development of a creative frame of reference does not take place in social isolation; it is shaped by interactions with others who are engaged directly or indirectly in similar endeavors. When faced with an equivocal situation, an individual reduces equivocality by seeking out the interpretations of others (Volkema et al., 1996). It is through these interactions that schemata, scripts, and categorizations diffuse throughout communities (Poole et al., 1989; Poole, Gray, & Gioia, 1990), occupational groups (Trice, 1993), third parties in related work, or networks of weaker ties (Granovettor, 1973). When interdependence is high, a collective mind can emerge (Weick & Roberts, 1993) and, with it, a communal sense of what makes sense. The result is a shift from the self-referential "I" to the more inclusive "we" (Weick, 1995).
In the organizational literature, authors capture intersubjective frames in such constructs as organizational memory (Walsh & Ungson, 1991), organization mind (Sandelands & Stablein, 1987), collective mind (Weick & Roberts, 1993), and organizational intelligence (Glynn, 1996). In recognizing that mental maps are shared and belief systems converge, scholars assume an intersubjective level of analysis that differs from traditional conceptualizations of the individual, group, or organizational levels (Daft & Weick, 1984).
However, although shared frames of reference may develop within communities, they can differ across communities. As Goffman (1974) notes, even though two actors share a similar set of experiences, their frames of reference may differ based on their positions with respect to that activity. As a result, sensemaking may not be neat, tidy, and polite but, rather, may be marked by divergent and sometimes antagonistic frames of reference, as different communities argue their respective beliefs (Trice, 1993; Walsh & Fahey, 1986; Walsh, Henderson, & Deighton, 1988; Weick, 1995). Fine explains:
By the placement of an occupation within an organizational field, workers provisionally create occupational meanings, given the real constraints under which they work and in light of the evaluations of other actors who impinge on their claimed expertise (1996: 111)
Thus, even in a single organization, a multitude of diverse frames can exist, arising from and characterizing different job categories, occupations, positions, status, ideologies, and paradigms (Trice, 1993; Weick, 1995). Creative activity, undertaken in large-scale organizations, is a particularly ripe context for the development of divergent frames of reference. Creativity typically involves tension between innovation and control (Weick, 1995) and, thus, can engender a natural dialectic between different, and perhaps opposing, intraorganizational communities (Huff, 1988). Dougherty (1992) and Trice (1993) are instructive to understanding how different interpretive frameworks embed and affect creativity.
Dougherty (1992) notes that innovation requires the insights of multiple "thought worlds"-that is, the interpretive schemas of different communities of specialists who literally think differently from each other. She identifies four competing thought worlds-technical, field, manufacturing, and planning-and argues that, at best, each thought world views the other as esoteric and, at worst, meaningless. The creative setting that serves as our exemplar (complex, interdependent, large-scale projects) creates a context in which there is likely to be conflict between two of the thought worlds Dougherty (1992) describes: (1) technical specialists who create product design (e.g., engineers, scientists, marketers, and manufacturing personnel) and (2) administrative specialists who plan, manage, and implement the design (i.e., those responsible for budgeting, for scheduling, and for interfacing with clients). These two communities-technical and managerial-likely have different views of what makes sense and, therefore, different views of what constitutes creativity.
Similarly, Trice (1993) views organizations as collections of diverse subcultures that subscribe to different systems of meanings. For him, there are two dominant subcultures-administrative and occupational-that compete over the control of work, access to power and resources, and credit; their inevitable clashes are resolved through adaptation and negotiation.
Daft and Becker (1978) demonstrate explicitly the link between different communities of meaning and creative endeavors. In their study of high schools, they found that 71 percent of the administrative innovation proposals (e.g., changes in community/school relationships, registration systems, human resource management systems, and systems to control costs) originated from administrators and, conversely, 70 percent of the teaching innovation proposals (e.g., curriculum changes, new teaching techniques, and equipment to support instruction) were initiated by teachers. Daft's (1978) finding that different groups promulgated different types of innovations led him to conclude that there was a dual-core nature to the innovation process
Differences in occupational frames are likely to lead to differences in the creative solutions proposed by each community. Technical staffs are likely to emphasize technical creativity as critical to the success of the project. Technical creativity involves the proposal of solutions that have technical foundations; it also involves the use and extension of the knowledge base held by technical personnel. For example, a materials engineer might propose the use of a new composite material, and a software engineer might propose the development of a computerized system to automate tasks previously done by system operators. Alternatively, creative project administrators might propose novel systems in any of a series of project managerial domains, including managing risk, satisfying customers, building political skills, staffing project teams, estimating costs, measuring project performance, and developing project feedback and evaluation systems (Frame, 1994).
In large-scale organizational projects, there is an inherent contest between the sensemaking frames of the administrative and technical communities, as each invokes their own referent framework to make sense of tasks or events (Thomas & McDaniel, 1990; Trice, 1993). Technical staff members, because they deal with tasks that have both heavy mental and material components (Barley, 1996), have a vested interest in developing reputations of creativity in science and engineering. In their study of a productdesign consulting firm, Sutton and Hargadon (1996) give a vivid example of how technicians seek reputational capital in brainstorming sessions. To engineers, brainstorms are "status auctions" in which they bid for prestige based on their ideas; as one industrial designer put it, "You are probably going to ask me about how brainstorms lead to creative products, but what strikes me is that those engineers treat it as a competition. It's a competition!" (Sutton & Hargadon, 1996: 706).
Conversely, project managers compete for a different sort of reputational capital. In his study of occupational "rhetorics," Fine notes that the ideal of "business" is linked to "images of rationality. . particularly as linked to control over the work process, long- and short-term planning, and an awareness of the economic placement of the workplace" (1996: 104). Project management is responsible for meeting preestablished goals of cost, schedule, and functionality imposed by senior managers; their creativity takes the form of finding ways to impress these executives and develop reputations of meeting goals successfully.
Finally, because the technical and managerial communities have higher within-group than between-group contact, their referent frames are supported by behavioral interaction patterns that reinforce one belief system over the other. Thus, we propose that intersubjective frames will likely be homogeneous within a given community but heterogeneous across different communities.
Proposition 2: Technical staffs and project management will differ in the frames of reference they use to understand and engage in a large-scale creative project.
Proposition 2a: The frames of technical staff will emphasize the need to experiment and the centrality of technical creativity to the success of the project.
Proposition 2b: The frames of project managers will emphasize the need to satisfy senior management and clients and the centrality of administrative creativity to the success of the project.
The Collective Level: Negotiated Order Among Differing Communities
Because the referent frames of the managerial and technical communities can differ from each other, discord can ensue. Ambiguity over the relative authority of managers and technicians in a single creative project is a manifestation of deeper conflict between the two occupational subcultures (Barley, 1996; Trice, 1993). The issue is fundamentally one of the separation of position from expertise (Barley, 1996). Managers may accept technicians (begrudgingly) because they solve problems useful to the manager's goals; conversely, technicians may accept managers (begrudgingly) because they can provide the resources needed to create technical designs.
Because they face different goals and tasks and value different forms of creative engagement, project managers and technical staff are likely, at least on occasion, to have opposing views of creativity. These views clash in organizational settings, with each group representing its opinions to the other in the form of an argument that stands for its unique model of what makes sense. These arguments reveal to the other parties the sensemaking perspective each holds. Although one outcome might be mutual accommodation (Poole et al., 1989), another might be that each party sees the other as not making any sense (Dougherty, 1992).
How conflicts are adjudicated between the technical and managerial staffs determines the process and direction of project creativity. Disputes are resolved politically (Walsh & Fahey, 1986), and whoever has the power to resolve a dispute owns the belief system that carries the day (Hickson, Hinings, Lee, Schneck, & Pennings, 1971; Pfeffer & Salancik, 1978). The cumulative resolutions of many disputes, in turn, become events for interpretation themselves. As managerial and technical staffs witness a dispute resolved in favor of one party over another, interpretations develop about how the organization's political system affects creativity. These interpretations guide the extent to which each group sets creative goals for itself and the degree to which the groups become engaged in their own form of creative process.
A negotiated order is likely to emerge from the opposing frames of managers and technicians (Walsh & Fahey, 1986), with occupational specialists doing "whatever is necessary within their ability to achieve the ends of the organization" (Fine, 1996: 111). Consistent with Weick (1995), we use the term collective structure to reflect the simultaneous cognitive and sociopolitical underpinnings of the process. Sensemaking may not be completely shared, but political compromise between opposing groups will nonetheless guide creative behavior. We find it useful to regard collective structure as a sliding scale that shifts to reflect the relative balance of power between the project management and technical staffs. For a particular point in time, the scale indicates which interpretive scheme controls the process of creativity. Following this line of logic, we propose the following:
Proposition 3: At any given point in the history of a project, a balance of power will exist between project managers and the technical staff. This balance will determine the extent to which each community engages in creative behavior.
Crises and the Possibility of Temporary Reframings
Sensemaking frames are fragile and subject to change. Sensemaking requires individuals and communities to enact their environment-a process that is reinforced through interactions with others in an attempt to reduce uncertainty. But even the solidarity of intersubjectivity may not be enough to sustain a frame in the face of inconsistent information. When individuals are involved in a stream of behavior (such as creativity), they always sustain some check upon their total involvement (Goffman, 1974). At the cognitive level, there is always a trace of hesitancy about past interpretation and a willingness to consider the need to reframe. When a disruptive event occurs, a frame may be broken and what once made sense no longer does; the individual experiences temporary disorganization (Goffman, 1974; Poole et al., 1989). If, because of new information, enough individuals experience interpretative disorganization, then an entire intersubjective frame may shift. Thus, there is the potential for a new set of frames to emerge, at least temporarily.
When a project experiences an event significant enough to be called a "crisis" (Dutton & Jackson, 1987; Peterson, 1998; Tyre & Orlikowski, 1994), the negotiated order between managerial and technical staff may shift to establish a new collective order to resolve the crisis. We draw upon Habermas (1975) to develop our definition of crisis; to us, a crisis occurs when the structure of a social system allows for fewer possibilities for problem solving than are necessary for the continued existence of the system. Crises arise from exogenous environmental changes but also in "structurally inherent system-imperatives that are incompatible and cannot be integrated" (Habermas, 1975: 2). A crisis occurs when the negotiated order of a collective system does not allow a problem to be resolved. Accordingly, we posit this:
Proposition 4: During a crisis, members of the technical and managerial staffs of a creative project will feel cognitive disorganization in their respective sensemaking frameworks for creativity.
Crises have sociopsychological effects; Poole et al. (1989: 273) term these critical incidents organizational breakdowns. When a crisis occurs, it causes project members to suspend their existing frames of reference and look at the world differently; Goffman (1974) terms this set of events re-keying. Whereas keying involves a transformation or transcription of experience from one meaning into another, re-keying means to reorganize meaning parenthetically by modifying (and not totally transforming) a current frame. Opposing frames of reference may become united through the common experience of a crisis. Each side, usually carrying its own set of meanings, can alter or drop these meanings in recognition of the need to resolve the crisis.
In the life of a project, crises can occur that can shift the negotiated order between project management and technical staff to allow for crisis resolution. The types of crisis are defined by the major goals of the project: functionality, cost, and schedule. A crisis of functionality occurs when a major design goal of the project is in jeopardy. Design goals, developed to meet strategic objectives or the specific needs of a client, spell out the performance of all or part of the product or service being developed. Examples of functionality crises are a plane that is overweight or fails to meet efficiency goals, a car that does not pass a crash test, or an operating system that is unstable. If a project falls short in meeting a design criterion that is essential for the project's success, a crisis ensues.
If the crisis is framed as one of functionality, the effect will be to shift power to those capable of solving the crisis, because those subunits most capable of dealing with an organization's critical problems are the ones that acquire power (Hickson et al., 1971; Pfeffer & Salancik, 1978). The ability to solve a problem and, consequently, to help the organization cope with uncertainty creates dependence on the subunit. In the case of a functionality crisis, power accrues to the technical staff, and the result is a shift in the negotiated order of the collective structure so that the creative frame of the technical staff becomes primary. This results in management's willingness to search for and accept technically creative solutions that solve the design problem.
Alternatively, crises can arise in response to cost or schedule problems. These crises will shift power toward the project management staff, and, as a result, that staff's sensemaking frame will dominate during the period of crisis. Project managers may refocus the organization's attention on "the conservation of resources and tight control mechanisms" and the "maintenance of the status quo" (Thomas, Clark, & Gioia, 1993: 244). The effect on creative behavior will be to favor simpler cost-effective solutions and to be biased against elaborate searches for creative technical alternatives.
Such a classification may be oversimplified, for a crisis might involve collaboration between administrative and technical staffs (Daft, 1978). A cost problem might be solved by alternative technologies and a technical problem by a project manager's creative interface with a client. Further, depending on the nature of the crisis, project managers and technical staff might simultaneously engage in creative behavior. Our overall point is that crises shift structures of power and meaning. Although we focus on competing models, we recognize that the specifics of the setting will govern the direction of the shift. Thus:
Proposition 5: The type of crisis a project encounters will shift the negotiated order of the collective structure to favor the referent frames of those capable of resolving the crisis.
Proposition 5a: When a crisis in functionality occurs, the collective structure of the project organization will shift to favor the referent frames of the technical staff. The technical staff will engage more in creative behavior.
Proposition 5b: When a crisis in cost or schedule occurs, the collective structure of the project organization will shift to favor the referent frames of the project managers. The project management staff will engage more in creative behavior.
Following Goffman (1974) and Trice (1993), we expect that crisis-engendered reframings are reversible and temporary. They are likely to be bounded in time by recognition of the crisis and its subsequent resolution. The collective structure may revert to its previous balance of power and meaning (Trice, 1993), or a new collective structure may emerge that rewards the side that resolved the crisis by according it relatively greater status. Lasting change may occur when repeated crises leave organizational members with altered interpretive schemes (Poole et al., 1989). The effects may be cumulative over the course of the life of a project, and the histories that develop may spill onto future projects. Based on this logic, we propose the following:
Proposition 6a: After a crisis has passed, a project will revert to its previous negotiated order and the balance of creative engagement (administrative or technical) that order implies.
Proposition 6b: Repeated crises of one type will shift the negotiated order of the project permanently in favor of the affected group, and that group will engage more in creative behavior.
Crises may occur naturally, but they also may be staged intentionally to suit the purposes of an individual or group (Goffman, 1974; Poole et al., 1989). An actor may manipulate the organization by reframing as crises events and issues that occur during the life of a project. Several studies in the sensemaking literature demonstrate that how an organization labels an issue determines how it responds to that issue (Dutton & Jackson, 1987; Staw, Sandelands, & Dutton, 1981; Thomas & McDaniel, 1990). Influential representatives of the project or technical staff may use periodic episodes of sensemaking that occur in a project (e.g., project milestones, interim project test dates, and meetings with stakeholders) to create a condition of crisis. For example, during periodic project testing, the technical staff may seek to push its group's agenda by demonstrating that a key part of the project is failing to meet its design criteria. If the reframing is effective and a crisis is declared, the technical staff may be free to engage in the creative activities it wishes to pursue.
A second, less Machiavellian, possibility exists for using crises to manage a community's goals. It is likely that members of both the technical and project staffs recognize and understand the structure of meaning in which they operate (Giddens, 1994). It may be that the current negotiated order would prevent them from having an idea funded or accepted. Knowing that crises occur periodically, these parties would time the introduction of their idea to an occurring crisis that would favor its acceptance.
The research of Burgelman and Sayles (1986) on internal corporate venturing provides some support for this possibility. In their case study, the authors note that a research scientist had been working on new types of insulation for years but was having trouble getting the work supported. With the coming of oil embargoes in the 1970s and the dramatic increase in the price of energy, the scientist's company faced a crisis in developing new products. The scientist then used these events as an opportunity to introduce a proposal for added research on insulation. The proposal was accepted.
It is likely that individuals engage in staging crises, or in timing the introduction of ideas to periodic crises, when their efforts to introduce creative solutions have been thwarted by their organizations' dominant political order. Our model of periodic crises fits the work of Tyre and Orlikowoski (1994). Based on three intensive case studies, these authors conclude that effective firms and managers take advantage of the "lumpiness" of organizational change. They exploit natural opportunities for change that appear in periodic, discontinuous bursts to introduce ideas and keep the process of innovation fresh. We summarize our expectations as follows:
Proposition 7a: Individuals or communities may stage crises to shift the negotiated order in favor of the type of creative engagement (administrative or technical) that order implies.
Proposition 7b: Individuals or communities may time the introduction of creative solutions to match the timing of periodic crises.
A Process Model of Creativity over Time
Our framework on creativity is composed of four interrelated concepts: (1) individual sensemaking, (2) intersubjectively shared frames of reference, (3) a collective structure that represents a negotiated belief structure between parties that have different frames of reference, and (4) a shift in the negotiated belief structure that results from crises. With this multilevel sensemaking model of creativity, we seek to address the question of how creativity unfolds over time. We summarize our model graphically in Figure 1 as a series of potential paths a creative project could take. We emphasize that these paths are potential, because each project organization would be subject to the influences of its own particular history of crises. The challenge to creativity researchers is to apply this model in studies of individual organizations.
The figure depicts the history of one potential project as it unfolds over time (shown on the x-axis). The history is broken down into different time episodes, labeled A through E. The y-axis represents a sliding scale depicting the relative dominance of the technical or project management staff in the negotiated order. Greater creativity on the part of the technical staff is represented in the upper half of the figure, and greater creativity on the part of the project management staff is represented in the lower half of the figure.
The outset of the project is marked by ambiguity and negotiation between competing belief systems (time period A). As individuals come together for the first time, they face great uncertainty and seek to reduce that uncertainty by making sense of the project and its creative demands (Huff, 1988). When the members of the project experience sufficient time together, intersubjective frames of reference begin to emerge.
As a result of this period of vacillation and uncertainty, a negotiated belief structure emerges in time period B. This time period is marked by relative stability, which will persist until a crisis appears. The relative balance between the technical and managerial belief structures determines the extent to which each side engages in creativity. There is a parallel between the collective and individual levels that determines the energy project members put into creative action or habitual action (Ford, 1996).
Time periods C and D represent crises of functionality and time/budget, respectively. In both cases, when a crisis occurs, the organization experiences a discontinuous and sudden shift to a different balance in the negotiated order. This shift persists until the crisis is resolved, at which point the organization reverts quickly to its prior negotiated order.
During a period of functional crisis (period C), the negotiated belief structure abruptly switches in favor of the technical staff. At this time management is more solicitous of creativity from the technical staff. Technical solutions that previously might have been considered too risky, expensive, or aggressive management now favors, and the technical staff engages in creative behavior. When the crisis is resolved, presumably owing to the generation of creative technical solutions, the negotiated order reverts back to its prior level.
Similarly, in time period D the organization experiences a crisis of budget or schedule. This results in a shift of the negotiated belief structure temporarily in favor of project managers. Project managers engage in creative behavior in their search for solutions to project-related problems that may be solved by securing more resources or extending deadlines, or by developing administrative systems that help the organization regain control over its targets. When this crisis is resolved, the organization once again reverts to its prior negotiated order.
Time period E represents a series of repeated and frequent crises of budget/schedule. The project experiences a series of reframings in quick succession, which gradually results in a new belief structure that permanently favors the belief systems of project managers. Should such a permanent shift occur, the organization is likely to experience a pronounced reduction in creative engagement on the part of the technical staff and an accompanying increase in creativity on the part of the project manager staff. We note that, at first glance, this model might be interpreted as a variant of a punctuated equilibrium approach to change (Gersick, 1991). Our model shares with Gersick's work the notion that a crisis can have a critical impact on the unfolding nature of a project. However, her work focuses on crises that occur as a result of the passage of time; more specifically, she found that crises occur at roughly the midpoint of projects and are generated by anxiety over project completion.[Chart]
Our model differs in that we propose that crises can occur periodically throughout the length of a project and are initiated by events both internal and external to the project. Gersick suggests that all members of a project are mobilized by the midpoint milestone, whereas we focus on how crises shift power, mobilize communities, and affect creativity throughout the life of a project. Although we recognize that crises can result in permanent change, as depicted in time period E of the figure, our model allows and even expects that the history of a project will be marked by multiple crises of varying types, which result in only temporary reframings of a stable negotiated order. Thus, our model has more in common with switching structure models of change (Duncan, 1976) than with punctuated equilibrium models.
Socrates reportedly said at his trial that an unexamined life was not worth living. We believe that the same holds true for theories. All theories contain assumptions; assumptions are necessary because they allow researchers to proceed without constant introspection and justification. But, at least occasionally, an examination of the dominant assumptions of a theoretical domain is warranted. It has been over 15 years since Amabile (1983) introduced her book on the role of context in generating creativity. Since then, multilevel theories on organizational creativity have appeared.
Our goals in this article have been twofold: (1) to examine assumptions about levels of analysis embedded in this literature and (2) to examine how relaxing or altering some of those assumptions leads to alternative perspectives on multilevel theorizing. In doing so, our aim is not to criticize or supplant existing creativity theory but, rather, to illustrate the possibility of other, perhaps complementary, theoretical approaches. As such, our ideas are generalizable beyond the domain of creativity. In the following we describe three theoretical realms to which our multilevel model might apply: (1) assumptions about inclusiveness, (2) process models and the dimension of time, and (3) a sensemaking perspective on defining levels of analysis.
Assumptions About Inclusiveness
In our review of the creativity literature, we found that most researchers adopt a common set of LOA assumptions concerning two key choices: (1) the definition of the creativity construct and (2) the level about which generalizations are made. We have shown that most creativity researchers define creativity as an outcome that is judged by others as novel and useful and focus on the level of the individual or work unit/ small project team. Taken together, these choices lead researchers to make further assumptions about the relationships of individuals to project teams.
In general, we have discovered that creativity researchers tend to adopt a functionalistreductionist approach (Rousseau, 1985), whereby individuals are modeled as entities that contribute to project or work group outcomes, which, in turn, ultimately define the overall organizational outcome. What each lower level of analysis contributes to the next higher level is creativity. Thus, individuals' creativity aggregates to become work group creativity, and work groups' creativity aggregates to become organizational creativity. The result is that organizational creativity is defined in terms of a part-whole relationship so that smaller or microlevel entities contribute to the functioning of the whole macrolevel system (Rousseau, 1985).
A logical extension of the functionalistreductionist approach is to model individuals as an inclusive part of the groups for which they work. That is, individuals are assumed to be influenced solely by the characteristics of the formal hierarchy to which they report. Because the hierarchy has a captive influence on individuals' creative behavior, researchers implicitly assume homogeneous effects of higher-level units on individuals. Although individual differences are acknowledged, group or contextual effects are assumed to override those differences.
By focusing on more complex organizations, characterized by high levels of horizontal interdependence, we have been able to relax the prevailing assumption of inclusiveness. Instead, we have proposed that individuals are influenced significantly by their occupational subculture. By using large-scale projects as an exemplar, we can assume significant numbers of both administrators and technicians to allow for meaningful occupational subcultures to develop. According to multilevel theorists (House et al., 1995; Rousseau, 1985), this amounts to an assumption of partial inclusiveness. Partial inclusion complicates multilevel theories; it suggests that individuals are subject to influence from a multiplicity of groups in which they hold membership (e.g., occupational, organizational, or task). Specifically, we have proposed that organizational-level creativity can be described as an order negotiated between administrators and technicians that occurred at the level of the entire project.
One potential avenue for future research on creativity might be to investigate the degree of inclusiveness present in different organizational or project contexts (House et al., 1995). We chose to focus our theory building on a setting where membership in occupational groupings would have a substantial impact on the frames held by individuals. It is possible that occupational membership would also have a similar impact on smaller project teams or work groups. However, in these settings the influence of formal organizational membership may be stronger because cross-functional and cross-team interdependence and interaction would be minimized. Such small-group settings might be characterized by a higher degree of inclusiveness. Our model also suggests that multilevel theories should incorporate sociopolitical processes. We maintain that various subcultures in organizations compete with each other and that this competition has significant effects on an organization's behavior. The principle of partial inclusion allows for the influence of overlapping role membership and occupational identities (DiMaggio, 1991; Fine, 1996). Our discussion of differences between technical and administrative staff could easily be extended to other forms of role and/or social identity overlap. These might include work versus family roles, functional versus project team roles, and multiple occupational and professional roles and identities.
Organizations are complicated entities. An assumption of inclusiveness has allowed creativity researchers to reduce complexity. This is particularly useful when the goals of researchers have normative implications; inclusiveness implies influence and control over valued outcomes. However, if the goal of the researcher is to describe complex organizational behavior, then relaxing the assumption of inclusiveness seems warranted.
Process Models and Time We have focused on creativity as a process, rather than as an outcome. At the intrasubjective (or individual) level, we have defined creativity as a person's psychological engagement in creative activity. At the level of an entire organizational project, we have shown that creativity involves not only individual engagement but the emerging structuration of who engages and when they engage. At the project level, we have illustrated that creativity consists of the ebb and flow of creative engagement among different occupational subcultures-that is, managerial versus technical staffs. We have posited that the political dynamics of such creative engagement leads to the emergence of a negotiated order, with shifts in the balance of power over the history of the project. Thus, given this dynamic perspective, time and timing are what we have sought to explain.
One of the most salient ways in which we have captured time is in noting how the organization responds to crises, particularly those of functionality, cost, and schedule. When an event occurs that an organization (or its membership) interprets as a crisis, the negotiated order of creative engagement can shift from one occupational subculture to another. This shift occurs through multilevel dynamics that are systemic and sociopolitical in nature. We have argued that the classification of an event as a crisis causes subcultures to reframe-at least temporarily-their shared, intersubjective causal maps. The crisis parenthetically alters the primary frames of subcultures (and member individuals) in a profound way that shifts the balance and determines which subculture (or individual) engages in creative behavior. Crises are time-bound events. Without explicitly including time as a variable in our model, crises would have no meaning.
By explicitly considering the role of time, our model extends previous work on levels of analysis, specifically that of Klein et al. (1994). We suggest that members of occupational subcultures (administrative and technical) are relatively homogeneous in their frames of reference at any given point in time. However, when a crisis occurs, it has the effect of reframing these primary frames of reference and causing them to shift. Within each occupational group, the resultant reframing is likely to be relatively homogeneous. Thus, the interpretative schemas an occupational culture maintains are homogeneous and internally consistent within each time period, but they may be considered heterogeneous across time periods.
Time also plays a role in giving individuals agency. Giddens (1994) proposes that an integration of macrolevel variables (structure) and microlevel variables (agency) can occur only in a dynamic framework. Our model allows individuals to understand the constraints under which they operate and to take action to change those constraints. This requires that time explicitly be incorporated into a model to allow for interpretation, action, and the emergence of a new or modified interpretation. Incorporating time into multilevel models increases the ability of a theorist to propose alternative, and possibly more robust, models.
Sensemaking and Levels of Analysis
In constructing our multilevel model of creativity, we adopted Weick's (1995) framework of intrasubjective, intersubjective, and collective levels of analysis. Although this parallels the traditional tripartite classification of individual, group, and organization levels, it differs in the important respect of focusing on the processes of sensemaking and interpretation in organizations. This focus on cognitive frameworks of sensemaking has allowed us to model the reframing that occurs during crisis, and to model how reframing shifts the negotiated order between administrative and technical frames to engender political contests over how the organization defines and solves problems creatively.
Thus, a primary contribution of a sensemaking perspective is that it alerts us to alternative ways of modeling multilevel influences in organizations. A principal theme we uncovered in the creativity literature was the use of an aggregation model to describe how individual actions can produce patterns of behavior at the organizational level. An alternative view, emerging from a sensemaking perspective, is that individuals may influence and, in turn, be influenced by cross-level effects (Rousseau, 1985) to affect the intersubjectively derived sensemaking frames that subcultures can share. For example, an individual may impact creativity at the collective level by actively seeking to frame events as crises. If successful, a reframing may result, and the negotiated order may shift to motivate subgroups, as well as individuals in those subgroups, to engage in creative acts.
At more macro levels, then, organizational creativity can result not only from activities born of a single individual but, rather, from dynamic and changing processes of sensemaking that emerge and establish themselves as a negotiated order for a point in time. Interpretive processes thus occur simultaneously at multiple levels of analyses in organizations, with effects at one level interacting with, and embedding, other levels of organizational analysis and sensemaking.
Implications for Research
Empirically testing the multilevel theory on creativity we advance here calls for methods that go beyond static or cross-sectional inquiry. Researchers interested in studying sensemaking processes in organizations need to adopt "a set of methodological tactics that enables them to deal with meanings rather than frequency counts" (Weick, 1995: 173). As a starting point for future research, we offer ideas on operationalizing three key features of our model: (1) assumptions about individual inclusiveness in organizations; (2) the importance of frames of reference and their shifts over time; and (3) the measurement of engagement in creative behavior at the individual and collective levels. In all of these arenas, incorporating dynamic models of change, evolution, and responsiveness to organizational events and crises are important techniques.
Researchers could examine assumptions about individuals' inclusiveness in occupational, organizational, or other communities by adopting methods from the burgeoning literature on organizational identification, which Mael and Ashforth (1992) define as the "perceived oneness with or belongingness to" the organization of which one is a member. One might posit that the degree to which an individual identifies with a collective might be an indicator of the degree to which that person feels included in, or influenced by, the collective. Thus, the degree of individuals' inclusiveness might be gauged by the strength of their identification. Organizational identification can be measured with the Mael and Ashforth (1988) scale, which has been adapted across different types of membership contexts (e.g., Bhattacharya, Rao, & Glynn, 1995). Shifts in individuals' identification over time could be detected by assessing members' identification at different points in the life of a project team, in individuals' tenure with the organization, and/or in terms of crises or other punctuation points. Researchers, thus, could test whether individuals who strongly identify with their occupational community might correspondingly be more responsive to situational factors designed to affect their creative engagement.
A central feature of our multilevel model of creativity is that cognitive frames shift over time in response to crises. The successful testing of our propositions, therefore, requires that these referent frames be measurable. One demonstrated technique for examining individuals' causal maps and the extent to which they may be shared within a collective is that of cognitive mapping (Barr, Stimpert, & Huff, 1992; Fiol & Huff, 1992). This approach has been applied to process research on organizational innovation (Swan, 1995). Cognitive maps represent individuals' schema or interpretations about concepts and cause-and-effect relationships in their information environment (Fiol & Huff, 1992). Data for cognitive maps can be generated through interviews (Bougon, 1983); causal grids (Swan & Newell, 1994); or the content analysis of text, interview transcriptions, or statements (see Huff, 1990, for an overview of techniques).
Cognitive mapping techniques have been used to study changes in managers' mental models over time in response to key environmental events or shifts (Barr et al., 1992), as well as managers' beliefs about technological innovation (Swan & Newell, 1994). Building on work in organizational decision making (Ford & Hegarty, 1984) and strategy (Barr et al., 1992; Daniels, deChernatony, & Johnson, 1995), those using cognitive mapping techniques might test our propositions on the development of and change in referent frames for creativity in organizations. Of necessity, such study would benefit from a field-based case study of a single organization and the unfolding of processes over time; Dutton and Dukerich's (1991) seminal study of sensemaking at the New York Port Authority is an instructive model.
To measure the creative engagement of individuals and groups, researchers might employ procedures used by Kahn (1990) in his study of psychological engagement and disengagement. Adopting qualitative methodologies, Kahn used multiple strategies, including participant observation, interviews, and content analysis of archival documentation. Such qualitative methods might be employed, or paired with nonqualitative methods (e.g., survey analysis), to study creative engagement at work.
Another strategy for assessing engagement might be the adaptation of the Experience Sampling Method (ESM): a "quasi-naturalistic method that involves signaling research subjects at random times during the day, frequently for a week or longer, and obtaining a report of the nature and quality of their experience" (Kubey, Larson, & Csikszentmihalyi, 1996: 99). Used extensively to assess whether a person is in "flow"-a condition of high challenges and skills (Csikszentmihalyi & LeFevre, 1989)-it is a technique that can be harnessed to examine whether an individual is in the process of creative engagement. The ESM strategy yields data that can be used in hierarchical linear modeling of experience (Moneta & Csikszentmihalyi, 1996) and can illuminate how and when individuals engage in creativity, as well as the factors that govern the process and its outcomes. Such qualitative, in-depth analyses of how creative processes unfold and change over time, and in response to crises, can begin to explicate the ecology of creative engagement in organizations and, thus, test some of the propositions advanced here.
Multilevel theories are emerging as powerful models for researchers to employ in mapping organizational phenomena. Because they simultaneously and interactively examine how agency at one level of analysis can interact with, and influence, that at other levels, they afford a means of describing the ever-morecomplex and ever-changing organizational landscape. We have focused on how assumptions about levels of analysis, embedded in models as institutionalized norms, can be changed so as to refocus and illuminate organizational processes differently. We have proposed an illustrative model, to complement those in the extant literature, that emphasizes how individuals, communities, and organizational systems can create meanings that impact the direction and flow of creativity in organizations. We now invite future organizational scholars to continue the Socratic tradition of leaving no assumptions unquestioned-even ours; it is such crises of meaning that propel the creative processes, for organizational practice and theorizing.[Footnote]