TI

THEORY AND REVIEW
CAPTURING BOTTOM-UP INFORMATION TECHNOLOGY USE PROCESSES: A COMPLEX ADAPTIVE SYSTEMS MODEL1
Ning Nan
Michael F. Price College of Business, University of Oklahoma, 307 W. Brooks, Room 305D, Norman, OK 73019 U.S.A. {nnan@ou.edu}
Although information systems researchers have long recognized the possibility for collective- level information technology use patterns and outcomes to emerge from individual-level IT use behaviors, few have explored the key properties and mechanisms involved in this bottom-up IT use process. This paper seeks to build a theoretical framework drawing on the concepts and the analytical tool of complex adaptive systems (CAS) theory. The paper presents a CAS model of IT use that encodes a bottom-up IT use process into three interrelated elements: agents that consist of the basic entities of actions in an IT use process, interactions that refer to the mutually adaptive behaviors of agents, and an environment that represents the social organizational contexts of IT use. Agent-based modeling is introduced as the analytical tool for computationally representing and examining the CAS model of IT use. The operationability of the CAS model and the analytical tool are demonstrated through a theory-building exercise translating an interpretive case study of IT use to a specific version of the CAS model. While Orlikowski (1996) raised questions regarding the impacts of employee learning, IT flexibility, and workplace rigidity on IT-based organization transformation, the CAS model indicates that these factors in individual-level actions do not have a direct causal linkage with organizational- level IT use patterns and outcomes. This theory-building exercise manifests the intriguing nature of the bottom-up IT use process: collective-level IT use patterns and outcomes are the logical and yet often unintended or unforeseeable consequences of individual-level behaviors. The CAS model of IT use offers opportunities for expanding the theoretical and methodological scope of the IT use literature. Keywords: Bottom-up IT use, complex adaptive systems, agent-based modeling, individual-level IT use, collective-level IT use
Introduction1
How information technology is utilized has been a central question for the Information Systems research community (Benbasat and Zmud 2003). This question becomes particularly crucial in organizational settings, as IT use processes have been recognized as an important pathway for organiza1
Lynne Markus was the accepting senior editor for this paper. Andrew Burton-Jones served as the associate editor. The appendices for this paper are located in the Online Supplements section of the MIS Quarterlys website (http://www.misq.org).
tions to harvest returned value from IT investments (Jasperson et al. 2005). In recent years, more and more researchers have come to the realization that the uses and consequences of information technology are often enacted through selforchestrated interactions among users, technologies, and institutional properties rather than dictated by organizational policies or managerial intentions (Barki et al. 2007; DeSanctis and Poole 1994; Orlikowski 1996). Accordingly, IS scholars have gradually moved away from the traditional static, discrete view of IT use to theoretical models highlighting the dynamic process where individual-level IT use behaviors and interactions collaboratively create collective-level IT use patterns and outcomes (i.e., bottom-up IT use processes) (e.g.,
MIS Quarterly Vol. 35 No. 2 pp. 505-532/June 2011
505
Nan/Capturing BottomUp IT Use Processes
DeSanctis and Poole 1994; Markus and Robey 1988; Orlikowski 1992). Given the significance of bottom-up IT use processes in organizations, one would naturally expect that IS research would have accumulated rich, robust empirical findings regarding the properties and underlying mechanisms of these processes. An examination of the IS literature indicates otherwise. Previous research has generally focused on discrete elements of IT use (e.g., individual cognition and technology characteristics) or episodic events during IT use processes. Neither approach affords a holistic and precise theoretical logic concerning the bottom-up linkage from individual to collective-level IT use. This lack of empirical research is caused by the inherent conflict between the traditions of IS research and the nature of bottom-up processes. While the former attempts to assume away interconnections among elements in order to gain analytical tractability, the latter is produced by interconnections among elements in a whole. In order to capture bottom-up IT use processes, researchers must look beyond traditional IS research perspectives. This study attempts to contribute to IS research by proposing a framework specifically suited to the examination of bottomup IT use processes. In particular, it seeks to extend the tenets and the instrument of complex adaptive systems (CAS) theory to the IS literature. CAS theory allows researchers to capture interactions among basic entities of actions and relationships between these entities and an environment, and analyze their contributions to macroscopic observations. By mapping the tenets of CAS theory to concepts of IT use, a novel view of IT use processes is presented. This CAS model of IT use is intended to serve as an ideal approach for encoding the complexity of bottom-up IT use processes into a formal model. The CAS model enables an analytically precise and consistent method for studying the properties and mechanisms of bottom-up IT use processes. The structure of this article is as follows: First, the conceptual boundaries are set and the focal concept of this study is established. Then a critical examination of the literature on IT use is conducted and bottom-up IT use processes as an underresearched area are identified. A CAS model of IT use is advanced to overcome this problem. Since the various lines of literature invoked in this theoretical development have not been collated and synthesized before, a detailed review and analysis of the theoretical development process is offered. Following the CAS model development, the application of the CAS model is illustrated by translating a widely cited case study of an IT-based organizational transformation process to a CAS model. Finally, the implications of this research are offered and conclusions drawn.
Conceptualization of IT Use
The theoretical development of this study began with an attempt to document and summarize the extant concepts of the IT use construct. Through the process of collating and analyzing the current understandings of IT use, an attempt was made to precisely specify the meaning of this core construct and explicate its key attributes. Moreover, an inclusive conceptual view allows evaluation of the development of the IT use literature, exposure of missing pieces of the IT use puzzle, and positioning of the CAS model among an array of theoretical perspectives of IT use. The term IT use is commonly used to refer to an activity involving three key elements: users as subjects utilizing the IT system, IT features as building blocks or components of IT artifacts, and tasks as functions being performed. Drawing on prior research (Burton-Jones and Straub 2006; Jasperson et al. 2005), IT use is defined as the interplay between users, IT artifacts, and work activities. Interplay can be embodied by users adoption of IT features, constraints imposed by IT features on user behaviors, or conformity/disconformity between IT use behaviors and work requirements. A wide rather than narrow understanding of IT use is adopted so that the subject matter of this article is unrestricted. The IT use construct has a long history of examination via snapshots of its discrete elements. As indicated in recent reviews of the system usage literature (Burton-Jones and Gallivan 2007; Burton-Jones and Straub 2006), IS researchers tend to employ variance-based models that focus selectively on one or two elements (i.e., users, IT features, or tasks) of an IT use activity. A number of researchers have become dissatisfied with these fragmented understandings of IT use and sought to articulate more comprehensive conceptualizations of the IT use construct. For example, Markus and Robey (1988) presented the option of modeling the uses and consequences of information technology as an emergent process where outcomes arise unpredictably from complex social interactions. Orlikowski (1992), and DeSanctis and Poole (1994) developed structurational models of technology to account for the mutually shaping yet indeterministic interactions between IT features, human actions, and institutional properties in addition to the time-dependent nature of the structuration process. Several IS scholars have recently begun to tap into another important aspect of IT use: its multilevel nature (Burton-Jones and Gallivan 2007; Burton-Jones and Straub 2006; Jasperson et al. 2005). They revealed the existence of two fundamental and complementary linkages between the individual and collective levels of IT use, namely the top-down and bottom-
506
MIS Quarterly Vol. 35 No. 2/June 2011
IT Collective level (2) Bottom-up (6) User (3) Top-down (5)
Task
User (3)
IT Individual level (1) t0 tn
Task
Legend: Potential research areas: 1. Individual-level IT use 2. Collective-level IT use 3. Interactions among the key elements of IT use 4. Dynamic patterns of IT use 5. The top-down process between the individual and collective levels of IT use 6. The bottom-up process between the individual and collective levels of IT use
Time (4)
Figure 1. The IT Use Construct
up processes (Kozlowski and Klein 2000). The top-down process refers to situational influences, or how the properties of IT use on a collective level shape cognition and uses on the individual level. In contrast, the bottom-up process captures how the attributes and behaviors on the individual level give rise to collective-level IT use patterns and outcomes. Together these extant conceptualizations of IT use have led to an inclusive view encompassing four essential attributes of the IT use construct. First, IT use involves users, IT artifacts, and work activities as its key elements (e.g., Burton-Jones and Gallivan 2007). Second, IT use behaviors and consequences are shaped by recurrent interactions within and between different elements of IT use (e.g., Orlikowski and Robey 1991). Third, IT use processes operate at multiple levels linked by bottom-up or top-down processes (e.g., BurtonJones and Gallivan 2007). Finally, time is a significant factor in IT use patterns and outcomes (e.g., Jasperson et al. 2005). Based on these attributes, Figure 1 is developed to depict the conceptual boundaries of the IT use construct in this study. Six potential research areas within the boundaries of the IT use construct, depicted as 1 through 6 in Figure 1, are identified.
ferent areas of the conceptual model of IT use as depicted in Figure 1 will be evaluated. Using the definition and conceptual view of IT use presented earlier, a large body of relevant empirical studies from the premium journals that regularly publish scholarly research on IT use were identified and reviewed (see Appendix A). Assessing the progress made to date, the identified studies were recategorized according to both their research perspectives and research areas as described in Figure 1. The result of this literature assessment is presented in Appendix B. While an examination of the literature showed that substantial progress has been made toward understanding various aspects of the IT use construct, it also revealed the bottom-up linkage between the individual and collective levels of IT use as a remarkably under-researched area. As shown in Appendix B, the number of studies focusing on bottom-up processes is noticeably less than studies focusing on other aspects of IT use (only 5 out of the 48 studies found). This observation is corroborated by recent critiques of the organizational (Kozlowski and Klein 2000) and system usage (Burton-Jones and Gallivan 2007) literature. Previous studies tended to attribute this research gap to a lack of theoretical understanding of the importance and nature of the bottom-up process, and attempted to direct more effort to this area by explicating the forms of the bottom-up process (e.g., Burton-Jones and Gallivan 2007). Yet the growing body of conceptual views and insights regarding the significance of the bottom-up process has not yet given rise to a flow of empirical research. A bottom-up process (also called emergence in the literature and the remainder of this article) refers to collective phenomena that are collaboratively created by individuals, yet are not
Literature on IT Use
As a handful of studies have thoroughly reviewed the primary streams of the IT use literature (e.g., Jasperson et al. 2005; Jones and Karsten 2008; Venkatesh et al. 2003), this effort will not be repeated here. Instead, the progress being made by prior IT use research on empirically examining the dif-
507
reducible to individual action (Sawyer 2005, p. 6). In exploring how this process unfolds, researchers require a framework allowing precise descriptions of all individuals and their recurrent interactions. The two dominant research approaches in the IT use literature, variance-based models and interpretive case studies, do not offer the necessary precise and holistic view. Variance-based models assume away interactions among individual actors for analytical tractability while case studies do not entail precise measures of individual or collective properties. The tenets of complex adaptive systems (CAS) are employed to propose a theoretical lens and a computational simulation tool particularly suitable for capturing the emergence of IT use (Merali 2004). Instead of reducing a phenomenon to a set of causal variables and an error term, CAS models typically demonstrate how aggregate structures arise from simple schemata and the interactions of microstate events in an environment (McKelvey 1997). The CAS framework allows opportunities for analyzing emergent phenomena without abstracting away interdependencies and interactions (Anderson 1999). Accordingly, it responds well to recent calls for more comprehensive and cross-level research in IT use (Burton-Jones and Gallivan 2007; Jasperson et al. 2005).
processes. This section will extend this third-wave of systems theory (called theory of CAS) to the IT use context.
Theory of Complex Adaptive Systems

As a relatively new stream of thought, CAS theory is still accumulating fragments of insight regarding emergence in specific contexts. Although these fragmented insights are insufficient to establish general statements of causality or prediction (Anderson 1999; Miller and Page 2007), they have given rise to a consistent view of the components and inner workings of CAS. The discussion is, therefore, concentrated on defining a complex adaptive system and, according to the taxonomy of theory (Gregor 2006), this view of CAS theory is analytic in nature. Researchers have employed different formulations in explicating the theory of CAS. For example, Holland (1995) explained the properties and mechanisms of CAS through its seven basics: aggregation, flows, nonlinearity, diversity, tagging, internal models, and building-blocks. Drazin and Sandelands (1992) condensed the inner workings of CAS to three levels of structure: deep structure, elemental structure, and observed structures. Although there is no universally agreed upon paradigm describing CAS (Gell-Mann 1994), three components have been consistently recognized as the core of the theory: agents, interactions, and an environment. The discussion is, therefore, framed around these three components while drawing on insights from previous expressions of CAS theory (see Table 1 for an overview). Agent Agents are individual actors or basic entities of actions in a complex adaptive system. Depending on the phenomena of interest, agents can represent a wide variety of entities such as human beings, organizations, objects, or concepts. Each agent is described by attributes and behavioral rules. Attributes are the internal states of agents. They can be fixed (e.g., race) or modified (e.g., wealth) over time (Epstein and Axtell 1996). Attributes can play several important roles in complex adaptive systems. First, they provide yardsticks for the fitness of agents. Fitness refers here to the ability of an agent to achieve a positive payoff (e.g., gaining resources or fulfilling work requirements) during the course of interactions (Holland 1995). For example, in a complex adaptive system of organizational learning, the knowledge level of individual workers is the indicator of their fitness (March 1991). Second, attributes enable agents to select interaction partners and form
A Complex Adaptive Systems Model of IT Use

Complex adaptive systems are defined as systems composed of interacting agents described in terms of rules. The agents adapt by changing their rules as experience accumulates (Holland 1995, p. 10). Due to the natural interface between CAS and a multitude of physical, biological, social, and organizational phenomena, CAS models have been the subject of an increasing number of studies from diverse academic fields including management and organizations (Amaral and Uzzi 2007). A few researchers have applied the concepts or the analytical tool of CAS to IS related topics such as agile software development (Vidgen and Wang 2009), IT-enabled organizational learning (Kane and Alavi 2007), and ITsupported team or organizational processes (Canessa and Riolo 2006; Curseu 2006; Raghu et al. 2004). Together these studies form the third-wave of systems theory, following the first-wave of structural functionalism and the second-wave of general systems theory and chaos theory (for a review, see Chapter 2 of Sawyer 2005). Although complex adaptive systems may include both top-down and bottom-up processes, the third-wave of systems theory has primarily sought to characterize the properties and mechanisms of bottom-up processes. This is because the analytical superiority of CAS theory and the instrument are most salient in untangling these
508
Table 1. Basic Concepts of the Theory of CAS

Concept Agent Attribute Behavioral rule Interaction Connection Flow Environment Structure Description Individual actors or basic entities of actions Internal states of agents Schemata that govern attributes and behaviors of agents Mutually adaptive behaviors Relational links among agents Movements of resources Medium for agents to operate on and interact with Topography of an environment and its relationship to agents Example People, objects, concepts Expertise, wealth, age Tit for tat, checkers-playing Standing ovation, cooperation Friendship, food web Goods distribution, knowledge sharing Landscape, social context Resource exchange between the environment and agents
relational links. Variation in attributes makes otherwise symmetric agents distinguishable, thereby enabling agents to selectively respond to the behaviors of other agents (Holland 1995). For example, in an agent-based model of neighborhood composition, ethnic identity enabled individual households to choose their neighbors and eventually led to neighborhood segregation (Schelling 1969). Meanwhile, similarity among attributes allows agents to aggregate into classes. For example, based on similarity of production skills, individual workers can form work groups, cartels, and industries (Drazin and Sandelands 1992). Finally, as described below attributes are important parameters in the behavioral rules of agents. Behavioral rules are the schemata governing an agents attributes and behaviors. They can be considered as a set of input/ output statements linking an agents perception of the world to changes in its internal state or actions (Drazin and Sandelands 1992; Epstein and Axtell 1996; Holland 1995). Behavioral rules typically operate through three steps. First, a focal agent inspects the structures of the environment and the behaviors of other agents. The breadth and depth of the inspection are bounded by the focal agents attributes such as vision, effort, and intention (Miller and Page 2007). Second, a set of statements (usually in the IF/THEN syntax) is executed using information collected from the first step as input, which may preserve or change the focal agents attributes or actions. These statements can range from simple-fixed heuristics (e.g., tit for tat in a cooperation game; Axelrod 1997) to elaborate optimization routines that evolve over time (e.g., checkers-playing program; Samuel 1959). Third, output from the second step elicits feedback from other agents and the environment, which in turn can affect the fitness of the focal agent either positively or negatively. The focal agent learns from this experience by recombining the elements of previously successful behavioral rules (Holland 1995). Behavioral rules encapsulate the process for agents to gradually improve their fitness in response to information and
feedback from other agents and the environment. They increase agents likelihood of surviving and thriving in the system (i.e., the selection process; Axelrod and Cohen 1999). Behavioral rules produce another important element of CAS when unfolding in time and space: interactions (Drazin and Sandelands 1992). Interaction Interactions capture the mutually adaptive behaviors of agents and are the most commonly observed structures in a CAS (Drazin and Sandelands 1992). Previous studies have explicated interactions through diverse examples such as spontaneous coordination in standing ovations (Miller and Page 2007), bilateral price negotiation in trading markets (Epstein and Axtell 1996), and cooperation in groups (Arrow et al. 2000). Interactions can be viewed by reflecting the common features of these examples as a function of agents, connections, and flows. Agents are the processors of interactions. As discussed above, behavioral rules contain the logic for converting patterns in the torrent of information received by an agent to changes in the agents attributes and behaviors. Recurrent applications of behavioral rules produce a temporal stream of interaction patterns. Connections are relational links among agents, delimited by the attributes of agents. Recall that one of the roles of attributes is to distinguish otherwise symmetric agents. Accordingly, agents selectively form relational links with some agents but not others. As attributes change during the course of interactions, these connections may also evolve. In a CAS, connections designate possible channels where interactions can take place. Connections acquire this role because the agents in a CAS are endowed with bounded rationality; that is, each agents adaptive behavior is based on local informa-
509
Macroscopic Observation
Environment
Emergence
resource
(structure)
Agent
(attribute, behavioral rule)
resource
resource
Interaction
(connection, flow)
Note: Different shapes represent different agents.
Figure 2. An Overview of the Theory of CAS
tion only, derived from those other agents to which it is connected (Anderson 1999, p. 220). For example, in a trading market, each traders buying or selling price is influenced by those with whom the trader has a business relationship. This web of business relationships is the venue where mutual-influence trading behaviors occur. Flows refer to the movements of resources through the web of agents and connections (Holland 1995). Resources can be physical objects such as food and goods, or virtual factors such as information and knowledge. In a CAS, flows provide energy sustaining the interactions of agents. Without a constant flow of energy, systems would degenerate to a state characterized by maximum disorder (Anderson 1999). In contrast, order arises from the mutually adaptive behaviors of agents when energy is continuously injected into the system. The movement of resources is an effective way for a CAS to capture and retain energy (Holland 1995).
(1996). In this model, agents move toward resource-rich locations in an environment. The agents collect resources from the environment, which in turn triggers the redistribution of resources among different resource-bearing sites. The three basic components of a CAS collaboratively create observations at a macroscopic level (see Figure 2 for an illustration). Macroscopic observations are not another component of CAS theory; instead, they are the aggregate properties of agents, interactions, or the environment. A complex adaptive system is intriguing in that macroscopic observations, while being a logical consequence of agents actions and interactions in a given environment, are often unintended and unforeseeable. For example, in Schellings (1969) models of segregation, an individual household had only a slight preference for neighbors in the same ethnic group; however, recurrent application of this slight preference led to the unintended consequence of neighborhood segregation. CAS theory represents a genuinely new way of examining these bottom-up processes. Rather than assuming away interconnections among entities of actions, it shows how order can arise as individual actors seek to increase their fitness in the adaptive landscape defined by interactions and the environment centered on them. As outlined below, CAS theory affords an exciting new opportunity for IS scholars to study bottom-up IT use processes.
Environment Environment is the medium for agents to operate in and interact with (Epstein and Axtell 1996). It is defined by structures and structures characterize the topography of an environment. For example, in an environment simulating a landscape with renewable resources, the amount of resources available at each resource-bearing site represents the structure of the environment (Epstein and Axtell 1996). Environmental structures provide important conditions for actions and interactions to unfold. Meanwhile, structures can be modified by the ongoing interactions of agents. The mutual-influence relationship between the environment and agents can be best illustrated by the artificial society model of Epstein and Axtell
Extending the Theory of CAS to the IT Use Context

The CAS model of IT use is premised on the multilevel, interactive, and dynamic nature of the IT use construct (see Figure 1 for the depiction). It is developed here to provide a
510
theoretical lens for examining the process whereby individuallevel IT use behaviors collaboratively create collective-level IT use patterns and outcomes (i.e., bottom-up IT use processes). Consistent with prior IT use theories (DeSanctis and Poole 1994; Jasperson et al. 2005; Markus and Silver 2008; Orlikowski 1992), the CAS model of IT use is framed around human actors (e.g., IT users), IT features (e.g., the text editor in an office system), and contextual structures (e.g., task requirements and organizational policies). Rather than capturing the complexities of IT use through constructs and relationships among constructs, the CAS model of IT use seeks to encode the IT use process into agents, the interactions of agents, and the relationships between agents and the environment. In the sections below, the IT use process will be conceptualized as a CAS by mapping the concepts of IT use to the basic concepts of CAS theory. Elements of the CAS model of IT use are discussed in light of the model depicted in Figure 3, and the mapping is shown in Table 2. Agent As defined earlier, agents are the individual actors or basic entities of actions in a complex adaptive system; they can represent people, objects, or concepts. Based on this definition, two general categories of agents have surfaced through an examination of prior research on IT use: human actors and IT features (depicted as employees and computers, respectively, in Figure 3). Human actors are individual users involved in IT use processes. Their active role in IT use processes and outcomes has long been recognized by IS scholars. As pointed out by Orlikowski (1992, p. 408), even the most black box technology has to be apprehended and activated by human agency to be effectual, and in such interaction users shape technology and its effects. The importance of human actors in IT use processes is evidenced in a number of IS research streams. For example, technology acceptance models (e.g., Davis 1989; Davis et al. 1989) show that the cognition of human actors is a significant source of variations in initial IT use behaviors. In addition, the social construction of technology suggests that human actors are key players in IT-induced organizational transformations (e.g., Orlikowski 1993, 1996). Although early IS research tended to treat technology as an objective external force, recent literature has increasingly recognized IT features as another set of active players in IT use trajectories and outcomes. As the products of human actions, IT features have human cognition and intentions inscribed in them (Orlikowski 1992). During the IT use pro-
cesses, the inscribed schemes and rules are activated, shaping the behaviors of human actors (Carley 1995) and participating in the information processing activities of organizations (Burton et al. 2006). The active role of IT features has been amply documented by a number of interpretive case studies, particularly those adopting the structurational view or actornetwork theory (e.g., Majchrzak et al. 2000; Walsham and Sahay 1999). By conceptualizing human actors and IT features as two primary categories of agents in IT use processes, we can apply the concepts of attributes and behavioral rules in specifying the mechanisms underlying mutually adaptive interactions in IT use processes. This in turn illuminates the causal pathways from individual-level characteristics and behaviors to the emergence of collective-level IT use patterns and outcomes. In particular, the attributes of human actors can be embodied by the individual differences of IT users such as demographic backgrounds, cognitive styles, and personality traits (for a review, see Jasperson et al. 2005). The attributes of IT features can be described as technology characteristics such as flexibility (Orlikowski 1996), reliability (Goodhue and Thompson 1995), and richness (e.g., Daft and Lengel 1986). Research has shown that individual differences and technology characteristics are important antecedents or moderators in individual-level IT use cognition or behaviors (e.g., Venkatesh et al. 2003). The CAS framework allows us to see the contribution of individual differences and technology characteristics to a bottom-up process of IT use. Building on the earlier discussion on the general roles of attributes in a CAS, the focus here is on three specific contributions of attributes to the emergence of IT use. First, attributes indicate the fitness of human actors or IT features in IT-enabled work systems; they reflect the effectiveness of human actors or IT features in achieving favorable outcomes during IT use processes. For example, IT-based work performance indicates the ability of users in realizing productivity gains in IT-supported tasks. The usefulness of technology indicates the potential of IT features for improving the efficiency of work activities. Second, individual differences and technology characteristics make otherwise identical agents distinguishable, facilitating selective interactions among human actors and IT features. For example, actor-network analyses of IT implementation processes (Sarker et al. 2006; Scott and Wagner 2003) implied that the political interests of human actors could serve as a foundation for the formation of political alliances. Similarly, according to technology acceptance models, the perceived characteristics of technology such as usefulness and ease-of-use are important antecedents for IT adoption and initial use (e.g., Davis 1989; Davis et al. 1989). Viewed in the CAS framework, these perceived IT characteristics effectively facilitate
511
Figure 3. The CAS Model of IT Use
Table 2. Mapping Between CAS and Extant IT Use Concepts

CAS Agent IT Use Human actors IT features Individual differences Technology characteristics Mental activities IT functionalities User-system interactions Interpersonal interactions User-system links Interpersonal ties Movement of intangible IT resources Environment Social and organizational structures Description The individual users involved in IT use processes including IT novices and IT experts The building blocks or basic components of a system such as the text editor in an office system The internal states of human actors such as demographic backgrounds, cognitive styles, and personality traits The internal states of IT features such as system reliability, flexibility, and richness The cognition and emotions of human actors The set of functions or capabilities delivered by IT features such as information access The mutually adaptive behaviors between human actors and IT features The mutually adaptive behaviors between human actors The direct uses of IT features The relationships between human actors The distribution of knowledge, information, beliefs, and other intangible IT resources among human actors and IT features The social organizational contexts The properties of social organizational contexts such as business strategies, culture, rules, and work requirements
Attribute
Behavioral rule
Interaction
Connection Flow Environment Structure
human actors to selectively interact with some IT features but not others. Third, the attributes of human actors and IT features serve as input for behavioral rules. This leads to a discussion on the behavioral rules of human actors and IT features. The behavioral rules of human actors can be embodied by mental activities associated with cognition and emotions.
Consistent with the definition of behavioral rules, these mental activities have been shown to link human actors perception of the world to conscious or automatic changes in their internal states and behaviors. For example, cognition contents such as effort expectancy (Venkatesh et al. 2003) and self-efficacy (e.g., Compeau and Higgins 1995) have been identified as critical factors in conscious IT use intentions and behaviors, while emotions such as the frustration or pleasure
512
of IT use may override the conscious intention to continue or stop using an IT system (Ortiz de Guinea and Markus 2009). These cognition contents and emotions can be viewed as the schemata governing the attributes and behaviors of IT users through the lens of CAS theory. Defined as the set of functions or capabilities delivered by IT features to support the information processing of organizations (Burton et al. 2006), IT functionalities can be considered the behavioral rules of IT features. IT functionalities acquire the role of behavioral rules, since as encoded human cognition and intentions they determine how IT characteristics and behaviors are preserved or modified during interaction with human actors. For example, group support systems have a common feature for brainstorming. When this feature is encoded with highly restrictive information access functionality, it prevents human actors from modifying the intended behavior of the brainstorming feature (Wheeler and Valacich 1996). Mental activities and IT functionalities play a central role in bottom-up IT use processes, since they produce collectivelevel IT use patterns and outcomes by recursively directing individual agents in adapting to and interacting with each other. Specifically, mental activities and IT functionalities (as inscribed human cognition rather than its physical presence) contain the tacit rules for human actors and IT features to increasingly improve their effectiveness (i.e., fitness) in an ITbased work system. This in turn gives rise to the adaptive behaviors of agents. For example, prior research has suggested that IT use behaviors are shaped by cognitive intention formation involving evaluations and expectations (Ortiz de Guinea and Markus 2009). When strong disconfirmation between expected and actual IT use outcome occurs, users are likely to initiate learning interventions or work system interventions (Jasperson et al. 2005). Meanwhile, IT functionalities (as inscribed human cognition) may seek to improve the effectiveness of IT features in an IT-based work system by restricting or facilitating actions that modify or enhance IT features (Orlikowski et al. 1995; Yates et al. 1999). Viewed through the lens of CAS theory, this process reflects the logic of change for human actors and IT features to adjust their attributes or actions in response to feedback from the IT-based work system. When recursively applied in time and space, they produce mutually adaptive behaviors of users and IT features. In addition, mental activities and IT functionalities elicit the expectations and reactions of a focal agent in reference to other agents actions. They comprise the underlying structure that gives rise to interactions among agents (Drazin and Sandelands 1992). For example, the subjective norm has been identified as an important antecedent of technology adoption intentions and behavior (Venkatesh and
Davis 2000); this suggested the role of the subjective norm in generating the expectations and reactions of a focal user in response to other users attitudes.
Interaction Interactions refer to the mutually adaptive behaviors of agents. In IT use processes, the mutually adaptive behaviors of human actors and IT features can unfold in two ways: between human actors and IT features, or among human actors. Mutually adaptive behaviors between human actors and IT features are often referred to as user-system interactions (Kane and Alavi 2008), and identified as a significant source of variations in the uses and consequences of IT systems. For example, the social construction of technology suggested that subtle ongoing interactions between users and IT systems could give rise to large-scale organizational transformation (e.g., Orlikowski 1996). Mutually adaptive behaviors among human actors are commonly referred to as interpersonal interactions (Kane and Alavi 2008), and recognized as important conduits for the propagation of attitudes and behaviors concerning IT use. For example, Burkhardt (1994) found that individuals attitudes toward and use of a newly installed IT system were significantly influenced by the attitudes and use of others with whom they were in direct communication. Similarly, Fulk et al. (1995) found that employees cognition and behaviors regarding e-mail system use were significantly associated with the attitudes and behaviors of their supervisors and five closest coworkers. Much of the extant literature has been concerned with understanding the relationships between interaction patterns and individual- or collective-level IT use outcomes. While the literature demonstrated the importance of these interactions, we have not yet gained sufficient understanding of the role of interactions in a bottom-up IT use process. By describing interactions via the concepts of agents, connections, and flows, we can specify how interactions arise from individual-level factors such as attributes, behavioral rules, relationships, and resource movements. This view of interactions can lead to the development of models including interactions as an integral part of a bottom-up IT use process. By definition, the agents involved in user-system interactions are human actors and IT features, while the agents of interpersonal interactions are human actors only. As processors of interactions, human actors and IT features encapsulate the fundamental rules enabling interactions to unfold. As discussed in the previous section, mental activities drive human
513
actors to increase their effectiveness by observing and acting on the attributes and behaviors of other IT use agents. Similarly, IT functionalities as encoded human cognition and intentions are intended to improve the effectiveness of IT features in an IT-based work system by restricting or facilitating modifications and enhancements to IT features. When recursively applied in time and space, mental activities and IT functionalities produce a temporal stream of mutually adaptive behaviors. Connections among agents designate the possible channels for interactions to occur (depicted as thick arrows between human actors and IT features in Figure 3). In user-system interactions, connections (referred to as user-system links) can be embodied by the direct use of IT systems (Kane and Alavi 2008). Through each use of IT systems, human actors and IT features are allowed the opportunity to observe or experience each others attributes and behaviors. Information derived from this direct use experience becomes input for the behavioral rules of agents. Mutually adaptive behaviors between human actors and IT features are limited to those agents directly connected through user-system links. In interpersonal interactions, connections are represented by interpersonal ties such as friendship, reporting relationships, advice relationships, and joint membership in clubs, trade associations, boards of directors, or alliances (Monge and Contractor 2003). Similar to user-system links, interpersonal ties enable human actors to observe and act on information derived from other actors to whom they are connected. The behaviors of each human actor are, therefore, dependent on the behaviors or attributes of the subset of agents with whom he or she has interpersonal ties. IT resources can be distributed through user-system interactions or interpersonal interactions. The concept of flows specifies the role of IT resource movements in energizing interactions. Prior research indicated that IT resources included not only physical artifacts such as hardware and software, but also intangible resources such as knowledge, information, commitment, values, and norms (Broadbent et al. 1999; Byrd and Turner 2000; Kumar 2004). These intangible resources are particularly important to IT use behaviors and IT-enabled work performance. In the CAS model of IT use, the concept of flows is mapped as the movements of intangible IT resources. Agents must contribute work continuously to sustain or evolve user-system and interpersonal interactions. As intangible IT resources such as knowledge and beliefs flow through the web of interactions, agents can be reminded to repeat previous behaviors or motivated to engage in the active exploration of IT features. For example, previous research has shown that the frequency and richness of communication increases the amount of common knowl-
edge between IT experts and business users. This in turn leads to more innovative use of IT systems (Lind and Zmud 1991). Through the lens of CAS theory, one can view IT experts and business users as agents and relationships between them as channels for knowledge transfer. When circulated through the web of interactions, this knowledge activates and enables business users exploration of alternative methods for utilizing IT features. Figure 3 depicts the flows of several intangible IT resources (information, belief, and data) via user-system links or interpersonal ties (depicted as thick arrows in Figure 3). In summary, interactions in IT use processes are the sequence of reciprocal actions governed by the behavioral rules of human actors or IT features. They take place via user-system links or interpersonal ties. Flows of intangible IT resources such as information and beliefs serve to inject energy into continued interactions. Environment The environment has long been recognized as a critical factor in IT use behaviors and outcomes in the IS literature. The environment generally refers to the social or organizational contexts of IT use processes (e.g., DeSantics and Poole 1994; Orlikowski 1992). In the CAS model of IT use, the concept of environment can be modeled as the agents social and organizational contexts. These contexts form a medium for agents to interact with and act on. The structures of the environment are represented by the properties of social and organizational contexts (depicted as contextual structures in Figure 3). For example, in technology acceptance models, the structures of environment are indicated by facilitation conditions (Venkatesh et al. 2008). In structurational models of technology, environmental structures are often discussed through the institutional properties of organizations such as business strategies, culture, rules, and work requirements (DeSanctis and Poole 1994; Orlikowski 1992). Although previous studies have amply discussed the relationship between environmental structures and IT use behaviors, a precise account of the mechanism underlying these relationships has not yet emerged. In the CAS model, environmental structures are inherently linked to other elements of IT use processes, which enables researchers to specify how environmental structures alter the actions and interactions of agents and vice versa. For example, researchers can model organizational culture as input for the behavioral rules of IT users in capturing the mutually shaping relationship between organizational culture and individual IT use behaviors. By specifying the formulation of these behavioral rules, researchers can precisely articulate propositions regarding the
514
mechanisms underlying the impact of organizational culture on IT use cognition and behaviors. At the same time, organizational culture can be modeled as a function of the attributes of IT users. As the attributes of IT users change, the organizational culture evolves accordingly. Summary Human actors, IT features, user-system interactions, interpersonal interactions, and the environment jointly create macroscopic observations of IT use. As IT use is typically embedded in the performance of other tasks (Gasser 1986), IT use patterns can evolve along with task performance outcomes. Therefore, the macroscopic observation of IT use is defined as collective-level IT use patterns and outcomes (see Figure 3). While acknowledging the possible relationships between IT use patterns and outcomes, these relationships are not conceptualized here in order to reduce the complexity of introducing the CAS model of IT use. Through the looking glass of complex adaptive systems, researchers attain a novel view of the factors and dynamics involved in a bottom-up IT use process. This CAS model of IT use is intended to characterize the key properties and mechanisms underlying the emergence of collective-level IT use patterns and outcomes. Although elements of the CAS model of IT use have been mapped according to existing IS concepts, they are not simply old wine in a new bottle. The metaphor of complex adaptive systems offers significant analytical advantages in examining the emergence of IT use. It provides the basis for concrete descriptions of how orderly IT use patterns and outcomes on a collective level flow from individual actors self-orchestrated technology sense-making, as well as the way actors are interconnected with each other and the environment. By reducing an IT use process into agents, interactions, and an environment, the CAS model of IT use naturally lends itself to agent-based modeling. This computational approach provides an analytically precise and consistent method for specifying the propositions underlying emergence of IT use. This allows researchers to leverage the computational power of todays information technology with the goal of achieving a dynamic and multilevel view of IT use processes.
modeling is to allow individual actors (i.e., agents) to interact with one another and an environment according to prescribed rules and to observe the macroscopic structures emerging from these interactions (Amaral and Uzzi 2007). This method originated in von Neumanns (1966) work on self-reproducing automata and has since been enriched by elements from various research fields (Epstein and Axtell 1996). Beginning with Schellings (1969) models of segregation, agent-based modeling has been applied to many social science domains such as organizational learning (March 1991) and organizational design (Rivkin and Siggelkow 2003). In principle, an agent-based model can be expressed as a set of mathematical formulas; however, these mathematical equations would be too complex for even the most adept practitioners to penetrate (Epstein 2006). Researchers have typically implemented agent-based models as computer programs coded in an object-oriented programming (OOP) language such as C++ or Java. OOP uses objects (i.e., structures that hold data and procedures) and their interactions to construct computer programs. This provides a natural scheme for representing agents and their interactions in a complex adaptive system. The process of codifying a conceptual model of CAS into an agent-based model is roughly analogous to operationalizing measures for theoretical constructs in other types of research (Davis et al. 2007). It involves specifying the computational parameters and algorithms reflecting the conceptual model, then building the agent-based model with an appropriate programming tool. Consistent with the conceptual models of CAS, an agent-based model encompasses three key elements: agents, interactions, and an environment. Their attributes, properties, and mechanisms are codified into computational parameters and algorithms. The computational research community has developed diverse tools for implementing agent-based models (Tesfatsion 2010). An agent-based model provides opportunities to conduct experiments in a simulated system. In a typical session of agent-based model simulation, a population of agents is released into the environment. Agents interact with one another and the environment according to behavioral rules. A simulation session usually allows multiple iterations of interactions until the model reaches a natural or predefined stopping point. In order to trace time-dependent effects, agent-based models use an internal clock to represent time and the clock ticks according to a predefined function. Computer programming offers researchers significant flexibility and control in visualizing an emergent process and collecting data on the attributes or behaviors of all the elements in a model at any point of time (Harrison et al. 2007). An agent-based framework offers several analytical advantages for a study on the emergence of IT use compared with traditional IT research paradigms such as statistical esti-
Instrument of CAS: Agent-Based Modeling

Agent-based modeling (also called multi-agent modeling, bottom-up modeling, and the artificial society system) is a computational simulation tool that has been widely adopted by CAS research (Epstein 2006; Epstein and Axtell 1996). Grounded in CAS theory, the central idea of agent-based
515
mating, differential equations, and verbal analysis. First, consistent with the emergent nature of IT use, an agent-based model does not assume a centralized control mechanism for the orderly behaviors of a system. Instead, it seeks to explore whether or not decentralized interactions among heterogeneous and adaptive actors lead to system-level regularities. Second, while dynamic patterns are not easily described verbally, an agent-based simulation can visually present the temporal contour of an IT use process step-by-step (Drazin and Sandelands 1992). Third, the computational approach of agent-based modeling gives researchers precision and control in the measurement and manipulation of crucial variables. As pointed out by Conway et al. (1959, p. 105), in physical experimentation stochastic variability is, by definition, that which lies beyond the control of the experimenter; in simulated experimentation, the stochastic variability, like every other feature of the model, is deliberately and explicitly placed there by the constructor. Finally, an agent-based simulation provides a very effective method for answering what if questions (Romme 2004). We can explore a wide range of possible contingencies that are difficult to implement in laboratory experiments or field studies. While traditional tools emphasize either flexibility (e.g., verbal analysis) or precision (e.g., differential equations), agent-based modeling attains an improved balance between the two (Miller and Page 2007).
logical approach for theory building (Davis et al. 2007); see Table 3 for more detail. Agent-based simulation is used here in the direction of theory-building. Running simulations is analogous to thinking through the logic of theories involving only a few constructs and relationships. As it is difficult for one to verbally analyze and mentally think through the dynamic interactions among a large number of agents, the simulation can aid theory development by computationally working through the logic of the CAS model of IT use. Findings from the simulation are not intended for hypothesis testing. Instead, they are logical consequences of a computational articulation of the CAS model for IT use. In the case study selected for this exercise, Orlikowski (1996) employed a situated change perspective that provided a contextual analysis of a new IT systems use in an organization over time. The organization, Zeta, installed an incident tracking support system (ITSS) for tracking customer calls in its customer support department (CSD). In appropriating the ITSS features, Zetas CSD employees enacted a series of adaptations and adjustments. This in turn produced significant changes in organizational practices and structures. Work in the CSD was previously tacit, private, and unstructured. As employees of Zetas CSD gradually incorporated the documentation and searching features of the ITSS into everyday practices, the nature and texture of work in the CSD became articulated, public, and more structured. A key insight surfaced via this case study is that organizational transformation could emerge from the subtle micro-level changes of human actors and IT features. This case study provided an excellent starting point for the development of a CAS model as it offered a sufficient description of the bottom-up process from individual-level IT use behaviors to organizational level IT use patterns and outcomes. Furthermore, at the end of the study Orlikowski raised three future research questions that not only helped define the boundaries of the CAS model, but also allowed the opportunity to demonstrate the advantages of agent-based simulation (see Table 4 for detail). In this exercise, the Zeta case was first conceptualized using the CAS framework. Building on the conceptual model, the agent-based model was developed, then model simulations were conducted in reference to Orlikowskis three research questions. The theoretical conclusions gained through this exercise will be offered after the model is presented.
Using the CAS Model of IT Use

This section seeks to substantiate the conceptualization and analytical tool of the CAS model by performing a theory development exercise. An interpretive case study of IT-based organizational transformation will be translated into a specific version of the CAS model of IT use, thereby demonstrating the operationalizability of the CAS framework. Unlike typical IS theory development, which primarily relies on verbal analysis, tables, and diagrams, this theory development exercise includes two elements commonly invoked by empirical work: case study and an agent-based simulation. Use of the case study is based on the belief that most IS theories, including the CAS model of IT use, require some form of realist ontology, as constructs in theoretical statements can refer to entities in the real world (Gregor 2006, p.631). The case observations provided the empirical grounding for the construction of the CAS model of IT use. As detailed below, the conceptual definitions of the CAS model were distilled from the case descriptions, while the computational representation of the CAS model (i.e., the agent-based model) was validated by case findings. Agent-based simulation is viewed as a third way of doing science (Axelrod 1997, p. 3) by serving as a computational laboratory for empirical testing (Burton 2003) or a methodo-
The Conceptual Model

Due to the complexities of reality, it is not possible to represent every aspect of the Zeta case in the CAS model. We must, therefore, separate what is essential from what is dis-
516
Table 3. Different Purposes of Agent-Based Simulation

General Purpose Empirical testing Theory building Purpose of Hypothesis Motivation for model construction and simulation Uninvolved or as conclusion of simulation Purpose of Agent-Based Model Computational laboratory for data collection Computational representation of theory Purpose of Simulation Finding Empirical evidence for hypothesis testing Logical consequence of theory Example Study Burton and Obel 1980 March 1991; Sastry 1997
Table 4. Orlikowskis Three Questions and the Advantages of Using Agent-Based Simulation
Question 1. What is the impact of employee learning/resistance?
Organizational inertia and resistance to change-often seen in organizations and predicted by a number of change theories-were not apparent within the CSD. Members of the CSD appeared to be open to exploring alternative ways of working, of learning from and changing with the new technology (p. 91).
Advantages of Agent-Based Simulation Allowing researchers to vary the learning rates of employees as an attribute of human actors and test their impact
2. What is the impact of ITSS flexibility?

More research is needed to investigate how the nature of the technology used influences the change process and shapes the possibilities for ongoing organizational change. Had a more rigid, more fixed-function technology been used, the pattern of use and change realized within the CSD would have been different (p. 91).
Allowing researchers to vary the flexibility of ITSS features as an attribute of IT features and test their impacts
3. What is the impact of workplace rigidity?

While the changes in the CSD [Customer Support Department] were relatively effective, one may imagine, for example, that in a more hierarchically organized or more rigidly controlled workplace, the sorts of workarounds, adjustments, and innovations enacted by Zeta actors may not have been tolerated or successful (p. 90).
Allowing researchers to vary workplace rigidity by directing interpersonal interactions toward different patterns and evaluating their impacts
pensable in order to capture in our models a simplified picture of reality, which nevertheless will allow us to make inferences that are important to our goals (Simon 1990, p. 7). The conceptual model developed here is intended to demonstrate the operationalizability of the CAS framework in answering research questions motivated by a traditional research approach. Therefore, a parsimonious approach was taken, with the focus given only to the case details most relevant to the three research questions shown in Table 4. Three steps were taken to replicate the Zeta case within the CAS framework. First, the case study was reviewed line by line with the objective of identifying those lines relevant to the basic elements of a CAS model for IT use. Second, from the case details identified in the first step, those most relevant to the three research questions in Table 4 were selected. Third, the case descriptions identified in the previous steps were analyzed and the conceptual definitions of a CAS model of the Zeta case were distilled. These steps have been proven effective in translating statements from one theoretical framework to another (Sastry 1997). They allowed development of a CAS model mimicking the details of the Zeta case, while
providing an alternative view of the IT-based organizational transformation. For concise presentation, the CAS elements and their conceptual definitions are summarized in the first two columns of Table 5 while the replicated case details are in the third column. Agent In the CAS model of IT use, human actors and IT features are conceptualized as the two primary aggregates of agents in IT use processes. These two agent aggregates can be embodied by the employees of the CSD and features of the ITSS respectively. The CSD employees included customer support specialists, managers, and director as the primary users of the ITSS and primary actors in an IT-based organizational transformation. The Zeta case implicitly and explicitly documented a number of individual differences and mental activities; however, to maintain the simplicity of this example, the focus here was on the learning rate of employees as the most salient attribute for
517
Table 5. The Conceptual Definition of the CAS Model

Element Employees of Zeta Conceptual Definition Customer support specialists and managers Replicated Case Description (Orlikowski 1996) The focus of my study was the Customer Support Department (CSD) which is part of the Technical Services Division headed by a senior vice-president. The CSD is a 53-person department run by a director and two managers (p. 67). Once trained, specialists began to use ITSS to do their support work, and as they responded to the modifications in their work and appropriated the technological features of ITSS, they enacted some of the changes intended by the implementation team. Other changes emerged as specialists and their managers accommodated issues and breakdowns in the use of ITSS, and improvised techniques and norms to effectively utilize the new technology in their changing work practices (p. 71). Members of the CSD appeared to be open to exploring alternative ways of working, of learning from and changing with the new technology (p. 90). The changes involved those specifically intended by the implementation team: electronic recording of all customer calls taken by the CSD; electronic documentation of work done on those calls; electronic reuse of prior call resolutions to avoid duplication of effort; and electronic monitoring of process and performance to facilitate process tracing and resource management (p. 71). In monitoring specialists process documentation through ITSS, managers changed their work practices to reflect the window they now had on specialists ongoing performance, a view that had not been possible beforeIn addition, managers began to evaluate the process documentation itself, not merely using it as an Indicator of actions and strategies (p. 75). With ITSS, specialists now made public most aspects of their research work through their own documentation of their ongoing work in progress (p. 76). The ITSS technology was designed to automatically assign a unique number to each incident entered into the database. This number included a code which identified the particular specialist who had documented the incident. Specialists learned each others identifying codes, and enacted an emergent change in their work practices when they began relying on this identifier to gauge the likely quality of potentially reusable incidents (p. 78). The change in specialists work practices to include electronic searching led, over time, to the unanticipated outcome of technological dependence (p. 79). The CSDhas traditionally had a very cooperative culture, reflecting a collegial management style and a shared interest in solving customer problems (p. 67). While junior specialists did lose direct experience with solving certain problems, they did not give up all opportunities for learning. The technology included a feature that enabled them to be notified whenever any action was taken on a record. Thus, a junior specialist, having assigned a call to a partner, could request that the system send electronic mail each time the partner updated the record. This way, junior specialists could follow the progress of calls and learn vicariously, at least (p. 81). By mid-1992, the Incident Tracking Support System (ITSS) had been developed.By the end of 1992, the decision was made to commit to the use of Notes as the platform for tracking all customer calls.This set the stage for a full roll-out of ITSS to all members of the CSD, and the enactment of the organizational changes which are the focus of this discussion (p. 68). Technologyplayed a critical role in mediating the changes in practices and structures.The conceptualization of technology drawn on here is informed by structurational analyses of technology in organizationsand posits technology not as physical entity or social construction, but as a set of constraints and enablements realized in practice by the appropriation of technological features (p. 69). ITSS was designed to allow this process documentation to be open-ended. The Incident History field in which specialists make their progress updates was unstructured, allowing entry of free-form text (p. 73). Over time, however, a number of informal norms about effective process documentation emerged. These norms, once shared and practiced within the CSD for some time, became reinforced and established as important cultural norms about the representation of work process within electronic documentation (p. 76). Many specialists were acutely aware of their new visibilityand responded by improvising some informal guidelinesabout what they would and would not articulate within the electronic text. In so doing, they began to appropriate the features of ITSS to manufacture a virtual or electronic persona of themselves by consciously engaging in impression management (p .77) .for some specialists, the use of ITSS created a forum in which to showcase their efforts, occasions to manage impressions of themselves as highly productive (p. 78). The ITSS database of calls with its documentation of process and resolutions soon contained enough prior incidents to make searching the database a useful step in researching problems. Specialists expanded their appropriation of ITSS features by beginning to use the powerful search engine available (p. 78).
Individual differences Mental activities
Learning rate Learning from ITSS features
Social learning
ITSS Features
The set of constraints and enablements realized in practice by the appropriation of ITSS
Technology ITSS flexibility characteristics IT functionalities The capability inscribed in the features to adapt to the majority practices of employees
518
Table 5. The Conceptual Definition of the CAS Model

Element User-system interactions Conceptual Definition Mutually adaptive behaviors between employees and ITSS features Direct uses of the ITSS Replicated Case Description (Orlikowski 1996) Information technology in the CSD plays a role similar to that of organizational properties-shaping the production of situated practices, and being shaped by those practices in turn (p. 69).
User-system links
ITSS was designed so that specialists would create an incident record in the ITSS database as each call was received, and then regularly update the incident record with the progress being made on the incidentITSSallows databases stored on a server to be accessed from distributed, networked personal computers, the incident records in the ITSS call database were designed to be accessible by all members of the CSD (p. 71). ITSS was deliberately designed to enable users to record, chronologically, the work being done on each incident, as it was being done (p. 73). The ITSS databasesoon contained enough prior incidents to make searching the database a useful step in researching problems. Specialists expanded their appropriation of ITSS features by beginning to use the powerful search enginespecialists depended increasingly on searching to do their problem solving (p. 78). The increased use of ITSSled specialists to spend considerably more time interacting electronically. Specialists began to use the ITSS technologyto communicate with each other via the electronic mail facility.They sent messages seeking technical advice, distributing departmental announcements, and sharing humor (p. 82). A new role, the partner, had been introduced (p. 80). Shared commitment to customer service had been a strong norm in the department since its inception, and it had recently been reinforced by the structural shift to partners and intermediaries (p. 83). Specialists discovered that with this virtual window into the work load of their peers they could browse through each others calls to locate those they could provide help on. Then, using the technology to send electronic mail or enter comments in a records Incident History, specialists could provide suggestions or solutions to each other. Specialistsboth junior and seniorchanged their work practices so that they routinely engaged in electronic help giving, whether solicited or not (p. 83). The mission of the CSD is to provide technical support via telephone to all users of Zetas products, including clients, consultants, Zeta field service representatives, and other Zeta employees (p. 67). Customer support at Zeta, as is often in the case in technical supportis a complex activity. Customer calls are rarely resolved with a brief answer. They typically require several hours of research and include searches of reference material, review of program code, and attempts to replicate the problem (p. 67). Through such ongoing enactment, specialists reinforced and eventually institutionalized a new set of work practices, substantially mediated by information technology and expanded to include documentation (p. 75). In particular, we saw changes in the following areas: patterns of interaction (from face-to-face and reactive to electronic and proactive)evaluation of performance (from outputfocused to a focus on process and output as documented); forms of accountability (from manual and imprecise to electronic and detailed)mechanisms of coordination (from manual, functional, local, and sporadic to electronic, cross-functional, global, and continuous) (p. 89). An unanticipated consequence of the emergence of proactive collaboration was an increase in the effectiveness with which problems were solved (p. 84). In particular, we saw changes in the following areas: the nature and texture of work (from tacit, private, and unstructured to articulated, public, and more structured)distribution of work (from call-based to expertise-based); nature of knowledge (from tacit, experiential, and local to formulated, procedural, and distributed) (p. 89).
Movement of intangible IT resources
Knowledge codification and reuse
Interpersonal interactions
Mutual-influence behaviors of Zeta employees Interpersonal relationships among employees Knowledge transfer
Interpersonal ties Movement of intangible IT resources
Environment
Organizational context at Zeta
Social and Requirements of the organizational customer support structures work IT-based organi- Assimilation of ITSS zational transinto organizational formation practices
IT-based work performance
addressing the first question on the impact of learning/ resistance of employees (see Table 4). Learning rate refers to an individuals likelihood to change his or her work practices under the influence of either the ITSS or other employees. This allowed testing of the impact of the learning rate or resistance of individual employees on an IT-based organizational transformation.
Employee behavioral rules were embodied by two mental activities particularly relevant to learning or resistance: learning from the ITSSs features and learning from other employees (i.e., social learning). Learning from the ITSSs features referred to the mental activity of understanding, evaluating, and applying the systems features. This enabled employees to change work practices via the use of ITSS fea-
519
tures. For example, specialists learned to electronically document their work processes and problem solutions, and subsequently learned to reuse solutions (see Table 5 for the case details). Social learning was the mental activity of perceiving, evaluating, and adopting the more productive practices of others in the workplace. This enabled employees to learn from their top performing coworkers. For example, junior specialists were allowed the opportunity to learn from more experienced coworkers by partnering with senior specialists (see Table 5). The ITSS features were the primary building blocks and components of the ITSS (e.g., the e-mail facility). Consistent with the case study, a structurational view was adopted and the ITSS features were conceptualized as the set of constraints and enablements realized in practice by the appropriation of the Incident Tracking Support System (Orlikowski 1996, p. 69). In reference to the second research question in Table 4, an important attribute of the ITSS is its flexibility or the ease of modifying its features. The IT functionality associated with this flexibility is the capability inscribed in the ITSS features to adapt to employee majority practices. This functionality represents the norms of ITSS feature appropriation. For example, the process documentation feature of ITSS was adapted to the specialists norms for documentation. Interaction Both user-system interactions and interpersonal interactions occurred in the Zeta case. In particular, user-system interactions were embodied by the mutually adaptive behaviors between employees and ITSS features. These took place during direct use of the ITSS (i.e., user-system links). In CAS terms, user-system interactions arose when the CSD employee behavioral rule was set for learning from the ITSS features, the ITSS behavioral rule was set for adapting to employee majority practices, and both rules were recursively applied during direct use of the ITSS. Customer support work processes and problem solutions were codified in the ITSS and made available through the search feature via usersystem interactions. This knowledge codification and reuse process (i.e., flows) energized the continued interactions between employees and the ITSS, enabling the gradual transformation of IT-based work practices. Interpersonal interactions unfolded through the interpersonal ties between employees, creating mutual-influence employee behaviors. In Zetas customer support department, these interpersonal ties may be embodied by communication relationships, advisory relationships, reporting relationships,
supervisory relationships, or partnerships. Social learning rules were recursively applied through these interpersonal ties, creating mutual-influence employee behaviors. Knowledge of customer support practices was shared and enriched through interpersonal interactions (i.e., flows); this not only energized ongoing interactions among employees but also contributed to the IT-based organizational transformation. The patterns of the interpersonal interactions allowed evaluation of the impact of workplace rigidity and answer the third question in Table 4.
Environment In the Zeta model, the environment was represented by the organizational context. The structure of the organizational context can be characterized by a number of factors such as organizational policies, culture, and the portfolio of tasks. To simplify the CAS model, focus was placed on the most important aspect of the environmental structure: the requirements of the customer support work (also referred to as work requirements). These requirements represented the nature and structure of customer inquiries as an exogenous factor affecting the actions and interactions between employees and the ITSS features. The macroscopic observation emerging from the actions and interactions of agents was IT-based organizational transformation. Since ITSS use was embedded in the organizational activities of customer service work, this organizational transformation is characterized by both organizational-level IT use patterns and performance outcomes. In the case study, Orlikowski provided contextual analysis of the IT-based organizational transformation in different areas of work (see Table 5). In the CAS model, two organizational-level measures were employed, capturing the organizational transformation at Zeta: assimilation of the ITSS and IT-based work performance. Assimilation of the ITSS refers to the extent to which employees have fully incorporated ITSS features into their work practices. This was intended to reflect organizationallevel IT use patterns in such areas as interaction channel (from face-to-face to electronic), evaluation of performance (from output-focused to process-focused), forms of accountability (from manual to electronic), and the mechanisms of coordination (from manual to electronic). IT-based work performance refers to employees ability to fulfill work requirements with the support of the ITSS. This was intended to capture IT-enabled performance outcomes such as better distribution of work among specialists, better formulated and distributed knowledge among employees, and more articulated, visible, and structured work practices.
520
Work Requirements [-1 1 0 . . . 0 0 1] 30-tuple d
53
Employees [-1 -1 1 . . . 0 0 0] [0 0 0 . . . -1 0 -1] . . [1 1 1 . . . 0 0 -1] 30-tuple .
ITSS [0 0 0 . . . 0 0 0] 30-tuple
Legend: a: The employees change their 30tuples by learning from the ITSS. b: The employees change their 30tuples via social learning. c: The ITSS changes its 30-tuple by adapting to the majority practices of the employees. d: The work requirements (constant 30-tuple) play out in social learning by defining top performers.
Figure 4. The Design of the Agent-Based Model
The conceptual model is summarized in Table 5. It presents an alternative view of the emergence of IT-based organizational transformation via subtle, ongoing technology sensemaking at the individual level. By reducing the emergence process to the properties and mechanisms of agents, interactions, and the environment, agent-based modeling was employed to specify the inner workings of this emergence process and explore the causal pathways underlying IT-based organizational transformation.
Model Design The well-established approach described above was adopted here in developing the agent-based model of the Zeta case. Figure 4 facilitates the explanation of the model design. As depicted in Figure 4, the ITSS is represented by a string of 30 digits (i.e., a 30-tuple); each digit in the string represents a feature of the ITSS. Although the actual ITSS may encompass a different number of features and a different string length specification can change the quantitative level of the results, the qualitative insights from the model are generally insensitive to this specification.2 An initial value of 0 is assigned to each dimension of the ITSSs 30-tuple to signify the neutrality of the technology before it becomes involved in mutually shaping interactions with human actors. The values of the ITSSs 30-tuple can remain 0, or change to either -1 or 1 (more detail on these two values below) over time as a consequence of adaptation to the employee majority practices. As with the ITSS, the CSD employees (including the managers and the director) are each represented by a string of 30 digits. Each digit in the string can be interpreted as an ITbased work practice in reference to an IT feature on the same dimension. For example, if the first digit in the ITSSs 30tuple represents the search feature, then the first digit in an employees 30-tuple indicates this employees work practices using the search feature. Consistent with the personnel composition in Zetas CSD, a population of 50 specialists, two
Agent-Based Modeling
Although agent-based modeling involves simulation, it is not aimed at providing an accurate computational version of the real world. The goal of agent-based modeling is to employ simple computational parameters and algorithms to operationalize the CAS model, allowing researchers to gain insights into the real world by observing the results generated by simple algorithms (Axelrod 1997). Reflecting this goal, a well-established approach in the agent-based modeling community is to use strings of symbols (e.g., numbers or letters) to represent agents and the environment of a CAS model (e.g., Axelrod 1997; Holland 1995; March 1991). These strings of symbols allow researchers to use a consistent syntax encoding the key elements of a CAS while abstracting away real-world details. For example, in a prior CAS model called Echo, Holland (1995) employed a string of letters to represent each agent. Each letter in an agents string signified an agents capability. The same syntax was used to represent the environment with each letter indicating an available resource. The letters in these strings could change over time as a consequence of adaptive interactions in the system.
This is indicated by previous agent-based modeling studies (e.g., Holland 1995; March 1991) and by testing of the agent-based model for the Zeta case using different values for the string length (10, 20, 40, and 50).
521
managers, and one director was released into the agent-based model. A total of 53 strings of numeric values were used to represent the human actors as depicted in Figure 4. Each dimension of an employees 30-tuple can take an initial value of -1, 0, or 1 with equal probability. These values are not meant to be interpreted in any strict numeric sense; instead, they provide a simple computational representation of the variations in work practices before the introduction of the ITSS. The values in an employees 30-tuple can remain the same or change over time as a result of learning from either the ITSS or via social learning. Identical values on the same dimension of the 30-tuples of the ITSS and a given employee indicate that the employee has incorporated that ITSS feature into daily work practices. The requirements of customer support work (environmental structure) are characterized by another string of 30 digits. The digits in the string represent the productive work practices (in the sense of satisfying work requirements) of applying the ITSS features to customer support work. For example, if the second digit in the ITSSs 30-tuple represents the documentation feature, then the second digit in the work requirements 30-tuple indicates a more productive work practice utilizing the documentation feature to accomplish customer support work. Consistent with the computational representation of employee work practices, each dimension of the 30-tuple of the requirements takes an initial value of -1, 0, or 1 with equal probability. Viewing the requirements of the customer support work as an exogenous factor, the values in the requirements 30-tuple are defined as constant. Neither the ITSS nor the employees have knowledge of the values in the work requirements 30-tuple. Identical values on the same dimension of the 30-tuples of the work requirements and a given employee indicate that the employees IT-based work practices are productive, providing a method for measuring performance. The employee learning rate attribute is signified by a probability p1. The value of p1 varies between 0 and 1, indicating an employees likelihood of work behavior changes induced by direct uses of the ITSS or via interpersonal interactions. An identical value of p1 is assigned to the 50 specialists, defining the learning rate of the two managers as 1.25 times p1 and the directors as 1.5 times p1 to reflect the varied competencies of employees at different positions of the organizations hierarchy. This assumes that more competent employees are more likely to be promoted. The flexibility attribute of ITSS is represented by another probability, p2; p2 can take a value between 0 and 1. This determines the likelihood of the ITSSs 30 digits changing their values during interaction with human actors.
The behavioral rule for employees learning from the ITSS is implemented as an IF/THEN statement (depicted as arrow a in Figure 4): if in a period an employees 30-tuple is different from that of the ITSS on any given dimension, then the employees value on that dimension changes to that of the ITSS with a probability equal to their learning rate (i.e., p1 for the specialists, 1.25 p1 for the managers, and 1.5 p1 for the director). Using the same approach, the social learning behavioral rule for employees (depicted as arrow b in Figure 4) was implemented as: if a focal employees 30-tuple differs from the top performer in his or her direct contacts on any given dimension, then the value on that dimension changes to that of the top performer with a probability equal to the employees learning rate. Direct contacts refer here to other employees possessing direct interpersonal ties with the focal employee (more detail on the formation of interpersonal ties later); the top performer is the employee whose 30-tuple corresponds with that of the work requirements on more dimensions than all other direct contacts of the focal employee (depicted as arrow d in Figure 4). The behavioral rule for the ITSS adapting to the majority practices of the CSD employees is represented by the IF/THEN statement (depicted as arrow c in Figure 4): if the value on a given dimension of the ITSSs 30-tuple differs from the mode of the employees values on the same dimension, then the value of the ITSS changes to the mode of the employees values with probability p2 (i.e., the ITSS flexibility). User-system interactions occur when employees carry out their behavioral rule for learning from the ITSS and the ITSS executes its behavioral rule for adapting to the majority practices of the CSD employees. According to the case description all employees are allowed direct uses of the ITSS. User-system interactions, therefore, take place between the ITSS and every employee in the model. Changes in employees or the ITSSs 30-tuples during user-system interactions represent knowledge codification and reuse between employees and the ITSS. Interpersonal interactions arise as employees act on their social learning behavioral rules via their interpersonal ties. Here, interpersonal interactions are directed toward two possible patterns in order to represent different levels of workplace rigidity. In a rigid workplace, employees only form interpersonal ties with their superiors and only learn from their superior contacts; specialists only build interpersonal ties with and learn from managers, while managers only form ties with and learn from the director. In contrast, a flexible workplace allows discretionary interpersonal ties and social learning. In both the rigid and the flexible work-
522
Table 6. Summary of the Agent-Based Model Design

Element Employees of Zeta Conceptual Definition Customer support specialists and managers Learning rate Learning from the ITSS features Computational Representation Fifty-three strings of numeric values with 30 values in each string (i.e., 30-tuple). Each value takes an initial value of 0, -1, or 1. The values can change during IT use. p1 determines the likelihood for employees to change values in their 30tuples during IT use. If an employees 30-tuple is different from that of the ITSS on any given dimension, then the employees value on that dimension changes to that of the ITSS with probability p1. A string of 30 numeric values, with each value taking an initial value of 0. The values can remain 0 or change to -1 or 1 during IT use. p2 determines the likelihood for the ITSS to change the values in its 30tuple during IT use. If the value on a given dimension of the ITSSs 30-tuple differs from the mode of the employees values on the same dimension, then the value of the ITSS changes to the mode of the employees values with probability p2. Employees carry out their behavioral rule for learning from the ITSS and the ITSS executes its behavioral rule for adapting to the majority practices of the employees. All employees are allowed direct uses of the ITSS. Changes in the employees or ITSSs 30-tuples during user- system interactions. Employees execute their behavioral rule for social learning via their interpersonal ties. Employees only contact their superiors in a rigid workplace and form discretionary ties in a flexible workplace. Changes in the employees 30-tuples during interpersonal interactions. Customer support work. A string of 30 numeric values, with each dimension taking an initial value of 0, -1, or 1. The values are constant. The percentage of identical values between the 30-tuples of the ITSS and the employees on average. The percentage of identical values between the 30-tuples of the work requirements and employees on average.
Individual differences Mental activities
ITSS features
The set of constraints and enablements realized in practice by the appropriation of the ITSS ITSS flexibility The capability inscribed in the ITSS features to adapt to the majority practices of employees Mutually adaptive behaviors between the employees and ITSS features Direct uses of the ITSS Knowledge codification and reuse Mutual-influence behaviors of Zeta employees Interpersonal relationships among employees Knowledge transfer Organizational context at Zeta Requirements of the customer support work Assimilation of the ITSS into organizational practices IT-based work performance
Technology characteristics IT functionalities
User-system interactions User-system links Movement of intangible IT resources Interpersonal interactions Interpersonal ties Movement of intangible IT resources Environment Structure IT-based organizational transformation
places, each employee can build only one interpersonal tie3 so that the total number of interpersonal ties is constant across different workplaces. As the case study does not provide a clear definition of workplace rigidity, this computational representation is grounded in the organizational literature that characterizes workplace rigidity by a centralized organizational structure and formalized change procedures (Staw et al. 1981). Changes in employees 30-tuples during interpersonal interactions represent knowledge exchange among employees.
The macroscopic observation of IT use is indicated by two values: assimilation of the ITSS and IT-based work performance. In the agent-based model, assimilation of the ITSS is calculated as the percentage of identical values between the 30-tuple of the ITSS and 30-tuples of the employees on average. IT-based work performance is measured as the percentage of identical values between the 30-tuple of the work requirements and 30-tuples of the employees on average. Table 6 presents a summary of the agent-based model design. An internal clock was implemented in the model to trace temporal changes. At each tick of the clock all agents were given the opportunity to execute their respective behavioral
The model was tested with either two or three ties allowed, and the qualitative insights from the model were found to be insensitive to this variation.
523
Table 7. Simulation Experiment Treatment

Experiment Employee Learning ITSS Flexibility Workplace Rigidity Treatment Varying the learning rate p1 from 0.1 to 0.9 using 0.2 increments. Varying the flexibility of the ITSS p2 from 0.1 to 0.9 using 0.2 increments. Varying the interpersonal interaction patterns among employees; a flexible workplace was embodied by discretionary interpersonal ties and social learning among employees, while workplace rigidity was represented by centralized ties and top-down social learning.
rules once. A simulation session included multiple clock ticks, mimicking the flow of time. The agent-based model was implemented using the NetLogo programming tool (Wilensky 1999). Before an agent-based model can be applied and generate insights, a model validation must be performed to ensure its robustness. This step is roughly analogous to manipulation checks in laboratory experiments and correlation matrices in multivariate analysis (Davis et al. 2007). Several strategies have been forwarded to evaluate the validity of computational models (Burton and Obel 1995; Carley 1996; Van Horn 1971). Validation is typically aimed at verifying the reasonableness of the simulation results. The model simulation was performed with a high degree of employee learning (i.e., p1 = 0.9), a high level of ITSS flexibility (i.e., p2 = 0.9), and a low level of workplace rigidity (i.e., discretionary interpersonal ties and social learning among employees). This treatment mimicked the environment at Zetas CSD as observed by Orlikowski. Consistent with the case findings, the simulation produced increases in both the assimilation of the ITSS and improvements in IT-based work performance. Although this result cannot prove that the model is a true representation of reality, it provides grounds for belief that the computational model can serve as a tool for gaining insight regarding the research questions. Following the model validation is the experimentation with the agent-based model. In reference to the three research questions, three sets of simulations were implemented to gain insights regarding the effects of the learning/resistance of employees, the flexibility of the ITSS, and rigidity in the workplace, respectively. The particular treatment of each simulation is detailed in Table 7. This experiment design resulted in a total of: 5 (p1) 5 (p2) 2 (rigidity of workplace) = 50 simulation treatments. Mimicking the two-year window in the IT use process at Zeta, it was necessary to set a predefined stopping point for all simulation sessions. The Zeta case indicated that, at the end of the two years, the IT-based organizational transformation
was evident but not yet completed. Accordingly, a simulation session should allow agents sufficient time to enact noticeable organizational transformations while still leaving room for further changes. After testing several different stopping points (i.e., 8, 10, 12, 14, and 16 clock ticks), 12 ticks were found to best represent the two-year window. Once a simulation session terminated at 12 ticks, the measures of IT-based organizational transformationassimilation of the ITSS and IT-based work performancewere obtained (see Appendix C for the pseudo-code of one simulation session). Simulation Findings in Reference to Orlikowskis Questions By repeating each simulation treatment 30 times (1,500 simulation sessions),4 the agent-based model produced a precise view of the effects of employee learning, ITSS flexibility, and workplace rigidity on ITSS assimilation and IT-based work performance over time (see Figures 5 through 10). Simulation results from the various experimental treatments were compared in answering Orlikowskis questions.5 First, the simulation showed that higher learning rates produced more rapid increases in ITSS assimilation in the short run, but the rates of increase leveled off quickly and converged with those of lower learning rates over time (Figure 5). Different learning rates had similar effects on IT-based work performance (Figure 6). Second, the simulation indicated little effect from variations of ITSS flexibility on ITSS assimilation (Figure 7) or IT-based work performance (Figure 8). Third,
4
Higher numbers of repeated simulations (50, 70, and 100) were tested, with the finding that qualitative insights from the model were insensitive to this variation.
Compared with ITSS assimilation, IT-based work performance is relatively low across experimental treatments (below 0.5 in Figures 5 through 10). This is a natural consequence of the agent-based model design since ITSS features and Zeta employees directly observe and adapt to each other, while Zeta employees indirectly adapt to the work requirements by learning from the top performers.
524
Figure 5. Assimilation of ITSS with Varying Learning Rates
Figure 6. IT-Based Work Performance with Varying Learning Rates
Figure 7. Assimilation of ITSS with Varying ITSS Flexibility
Figure 8. IT-Based Work Performance with Varying ITSS Flexibility
Figure 9. Assimilation of ITSS with Varying Workplace Rigidity
Figure 10. IT-Based Work Performance with Varying Workplace Rigidity
the simulation showed that the rigidity of a workplace had no apparent effect on ITSS assimilation (Figure 9), while the less rigid workplace produced slightly greater work performance gains midway through the IT use process (Figure 10).
Orlikowski discussed employee learning, ITSS flexibility, and workplace rigidity as caveats of her case study. This implies the possible effects of their variations on organizational-level IT use patterns and outcomes. A persons casual intuition
525
might also have led to the expectation of such effects from different experimental treatments in the model. Surprisingly, the simulation showed little effect of the three factors, with the exceptions of the initial impact of employee learning and effect of workplace rigidity over time. After reexamining the agent-based model, several logical yet unintentionally encoded consequences of the learning rate, ITSS flexibility, and workplace rigidity were discovered. According to the employees behavioral rules, higher learning rates would accelerate employee learning from the ITSS and top performers. Therefore, a positive impact of learning rate on ITSS assimilation and IT-based work performance was expected. The simulation results and subsequent tracing of the logic of change built into the model led to a new insight: higher learning rates tend to lock a larger proportion of employees into a learning plateau (Kane and Alavi 2007). In the agent-based model, an employee could only learn from the ITSS or top performers when there were differences between them. Yet learning would eliminate the differences and diminish further learning opportunities, resulting in a learning plateau. By enabling faster learning, higher learning rates served to quickly deplete the differences between employees and their interaction partners. On the one hand, this led to rapid stabilization of the majority practices of employees, allowing the ITSS to adapt to employees and increase the speed of ITSS assimilation (Figure 5). The increasing ITSS assimilation leveled off as ITSS adaptations eventually eliminated all of the differences between the ITSS features and majority practices of employees. On the other hand, as higher learning rates led to a rapid convergence of employee work practices they prevented employees from discovering and acquiring more productive work practices. Although the unconverged employees could still achieve rapid performance gains, their gains were diluted by the stagnant performance of those locked in the learning plateau. Since higher ITSS flexibility allowed the ITSS features to adapt more rapidly to the majority practices of employees, a positive impact of ITSS flexibility on ITSS assimilation and work performance was expected. The surprising simulation findings (Figures 7 and 8) pointed to the fact that, by permitting more frequent changes to ITSS features, higher ITSS flexibility created a moving target for employees (Dennis et al. 2001). Employee learning from the ITSS and top performers was impeded because it was difficult to assimilate frequently changing ITSS features, or use the ITSS as a permanent repository for sharing and reusing productive work practices. The aggregate of the expected gains and unintended losses from higher ITSS flexibility produced similar observations from various ITSS flexibility experimental treatments.
A rigid workplace restricted employees interpersonal ties to their superiors. It limited the diversity of employees knowledge sources and could thereby impede ITSS assimilation and work performance gains. Yet since ones superiors were endowed with higher learning rates, they tended to assimilate ITSS features and acquire productive work practices more efficiently. A rigid workplace may, therefore, facilitate ITSS assimilation and work performance gains by ensuring the quality of employees knowledge sources (Staw et al. 1981). This benefit mitigated the downside of a rigid workplace, leading to similar ITSS assimilation results in workplaces with different degrees of rigidity and only slightly greater work performance gains in a flexible workplace. Summary The simulation findings and the new insights gained are summarized in Table 8. Orlikowskis case study of IT-based organizational transformation has been translated into a specific version of the CAS model of IT use. The case details relevant to Orlikowskis three questions are represented with the basic elements of the CAS model. The agent-based model simulation assisted in thinking through the logic of this specific version of the CAS model of IT use. By unraveling the seemingly counterintuitive simulation findings, we have learned that the deliberately encoded effects of a favorable or unfavorable factor in individual actions tends to be offset by the unanticipated effects stemming from adaptive behaviors in response to these individual actions. Employee learning rates, ITSS flexibility, and workplace rigidity were, therefore, found to have limited impact on organizational-level IT use patterns and outcomes. By showing a lack of direct causal linkage between factors in individual actions and observations on the collective level, the simulation findings in effect manifest the hallmark of bottom-up processes: collective-level phenomena are collaboratively created by individuals, yet are not reducible to individual action (Sawyer 2005, p. 6). This theory building exercise extends Orlikowskis case study by speaking to the generalizability of the IT-based organizational transformation at Zeta. Through the CAS lens, we see that the robustness of this organizational transformation against variations in employee learning rate, IT flexibility, and workplace rigidity is inherent to the bottom-up nature of the IT use process at Zeta. The generalizability of the ITbased organizational transformation is, therefore, more a function of the nature of IT use processes than the discrete organizational or technological conditions. Corroborating the practical implications of Orlikowskis study, this exercise advocates a situated and improvisational approach in implementing and using a new IT system. IT managers are encouraged to
526
Table 8. Summary of Simulation Findings and Insights

Orlikowskis Question 1. What is the impact of employee learning/ resistance? Finding Variations in learning rates have short-term impacts on ITSS assimilation. They have little impact on ITbased work performance. Variations in ITSS flexibility have little impact on ITSS assimilation or IT-based work performance. Insight Higher learning rates may impede work performance gains by locking a larger proportion of employees into a learning plateau. This unintended consequence can erode their benefits on employees adaptive behaviors. Higher ITSS flexibility may impede ITSS assimilation and work performance gains by making ITSS a moving target for employees. This unintended consequence can offset their benefits on the ITSSs adaptive behavior. A rigid workplace may facilitate ITSS assimilation and work performance gains by ensuring the quality of knowledge sources. This unintended benefit can compensate for its limits on the diversity of knowledge sources.
2. What is the impact of ITSS flexibility?
3. What is the impact of workplace rigidity?
Variations in workplace rigidity have a slight impact on IT-based work performance over time. They have little impact on ITSS assimilation.
provide a legitimating context for the reciprocal adaptations and adjustments of IT users and IT features since the recurrent interactions of these IT use agents in an organizational environment are the key mechanisms leading to IT-based organizational transformation. In concluding this exercise, the assertion is that the IT-based organizational transformation is not limited to organizations with high employee learning rates, high IT flexibility, and low workplace rigidity; it can occur in any organization permitting mutually adaptive interactions among human actors, IT features, and environmental structures.
this section, the implications and limitations of the CAS model of IT use are discussed.
Implications for Theory

Several key implications to the advancement of the three dominant theories of IT use offered by the CAS model are seen: the structurational theory of technology, technology acceptance models, and the multilevel perspective of system usage. The structurational theory of technology has been recognized as one of the most influential theories in the IS literature (Jones and Karsten 2008). It has motivated the indeterministic view of IT use processes (Markus and Robey 1988), the situated perspective of IT-based organizational transformation (Orlikowski 1996), and the social construction of group decision support systems (DeSanctis and Poole 1994). Yet in a recent review of structurational research in the IS field, Jones and Karsten (2008) concluded that IS researchers had, in general, failed to take full advantage of structuration theory. One of the ignored aspects of structuration theory they saw was the linkage of individual micro-level action and macro-level institutional processes (p. 150). This study points to the possibility of applying the CAS framework toward closing this research gap. The CAS framework provides researchers with a more precise vocabulary for describing and hypothesizing the causal pathways in an emergent IT use process. While structuration
Discussion
In the last three decades, IT use processes have been increasingly recognized as multilevel, interactive, and dynamic phenomena. A number of IS researchers have urged the research community to develop and apply richer and more sophisticated models to examine variations within and across different levels of analysis, allowing us to unravel the complexity of IT use processes (e.g., Burton-Jones and Gallivan 2007; Jasperson et al. 2005). As an attempt to answer this call, the current study focuses on an important yet underresearched aspect of IT use processes: the bottom-up linkage from individual-level IT use attributes and behaviors to collective-level IT use patterns (i.e., emergence of IT use). By proposing the CAS model of IT use and illustrating its application to IS research, the aim is to introduce a new approach for IS research based on its emergent nature and to generate deeper understandings of the IT use construct. In
527
theory is fundamentally nonpropositional (Orlikowski and Robey 1991), the CAS framework can serve as a middlerange theory supporting the analytical and empirical exploration of specific propositions. By formulating assumptions regarding agents, interactions, and the environment, researchers can provide a precise specification of the properties and mechanisms central to the bottom-up linkage. By varying the key assumptions, researchers can systematically evaluate the contributions of each element to the behavior of the entire system. This article has provided an example of using the CAS model to specify the bottom-up linkage from individual-level IT-based behavioral changes to organizational transformation. The steps for applying the CAS model and agent-based modeling have been demonstrated by translating the verbal analysis of an emergent process of IT-based organizational transformation at Zeta to the conceptual and computational representation of a complex adaptive system. This example also indicated the feasibility of extending the structurational theory of technology with the CAS model of IT use, and achieving deeper understandings of the phenomenon of interest. Technology acceptance models (TAM) represent a long tradition in IS literature. Over the years, a rich body of evidence has been accumulated regarding the association between pre-adoption individual cognition and initial IT use intention or behaviors. Although TAM provides important insights into the path-dependent nature of IT use behaviors, it addresses only a discrete element (i.e., individual cognition and behaviors) of IT use processes. A number of researchers have subsequently called for studies broadening the scope of TAM research to reflect the dynamic, multilevel, and interactive nature of IT use (e.g., Jasperson et al. 2005). The CAS model of IT use affords an opportunity for TAM to be situated in a comprehensive view of the IT use construct. Specifically, the consistent and robust findings on individual cognition and behaviors by TAM provide a solid foundation for the formulation of the attributes and behaviors of human actors in a CAS model. These attributes and behaviors encapsulate the logic of change in an IT use process. It would be interesting to observe what collective-level outcomes will arise from interactions among individual agents endowed with the cognition and behavioral intentions prescribed by TAM models. The CAS model can become a framework for TAM and other research streams collaborating in a logical way. The multilevel perspective of system usage has recently been presented by Burton-Jones and Gallivan (2007), representing the frontier of research on the IT use construct. An important research opportunity exposed by the multilevel perspective is identifying the antecedents of the emergence of collective usage, and understanding the configural relationships between
individual and collective usage patterns. The CAS model lends itself naturally to these research endeavors. The CAS model is inherently multilevel, as collective-level usage is considered an emergent property dependent on the aggregation of individual-level behaviors. The antecedents of the emergence of IT use can be examined via the key properties and mechanisms of agents, interactions, and the environment in the CAS framework. This configural relationship is essentially the bottom-up linkage from individual behaviors to collective-level outcomes, formulated as a function of the key properties and mechanisms of CAS.
Implications for Method

McKelvey (1997, p. 375) pointed out that the success of organization science depends in part on finding more fruitful applications of computational and analytical methods to intraorganizational explanation. The argument in the present study is that the research on IT use processes can similarly benefit from the application of agent-based modeling. As depicted in Figure 11, agent-based modeling is uniquely positioned at the sweet spot between the two dominant analytical approaches in the IT use literature: interpretive case analysis and variance-based analysis. The theoretical insights derived from interpretive case studies can illuminate the modeling of interactions between actors and contexts in an IT use process (arrow a in Figure 11). In return, agentbased modeling contributes to interpretive case studies by providing an analytically precise and consistent method for describing and comparing qualitative observations from the field (arrow b in Figure 11). These complementary relationships between interpretive case studies and agent-based modeling have been illustrated in the theory-building exercise translating Orlikowskis case study to a specific version of the CAS model of IT use. In addition, the simulation results from an agent-based model can inform effective strategies for variance-based analysis in selecting the most promising aspects of a bottom-up process and constructing variancebased research models (arrow c in Figure 11). The empirical findings from variance-based analysis can serve as important benchmarks for calibrating and validating agent-based models (arrow d in Figure 11). The belief is that agentbased modeling in concert with other analytical approaches provides a more complete picture of how positive IT use outcomes may arise from recurrent interactions between technology features, human actors, and institutional properties. In addition, Sandelands and Drazin (1989) pointed out that one consequence of fitting processes into variance-based models was the creation of language that gave the illusion of a process when a process was not being examined. They high-
528
Interpretive case analysis
a b
Agent-based modeling
c d
Variance-based analysis
Figure 11. The Complementary Relationship Between CAS and Other Approaches
lighted the difficulty in achieving internal validity for theories and methods in studies of organizational processes. Agentbased modeling can ensure the internal validity of theoretical propositions via the computational representation of IT use processes. This allows researchers precision in expressing time and time-dependent theoretical propositions in a computer programming language. While verbal description may entail ambiguity or even the illusion of a process, computational representation can clearly depict how a process unfolds and evolves over time. Furthermore, by using the possibility of conducting a what if analysis in an agent-based simulation, researchers are encouraged to implement more sophisticated research programs. IT use processes are inherently complex and dynamic, involving a myriad of human, technical, and organizational factors that may evolve over time. In order to fully understand the IT use processes, researchers must implement a research program that can evaluate the combined as well as individual effects of key factors over a long period of time. Such a program can take a team of researchers several years to accomplish using traditional methods. Simulation in agentbased models allows researchers to take advantage of the computation power of todays technology and evaluate the complexities of IT use processes in a much shorter time frame and with less investment of resources. While providing the benefits discussed above, the computational nature of agent-based modeling has provoked concerns in the research community. First, the results from an agentbased model simulation seem to be built-in. It is true that the results are the logical consequences of the model design as otherwise there is a serious internal validity problem with the model; however, this does not mean that new insights gained could be predicted based on the initial model design (Miller and Page 2007). The value of agent-based modeling is in providing insights from clearly defined mechanisms rather than producing surprises from obscure black-box features. Second, due to the inherent flexibility of computational models, agent-based modeling is often criticized for a lack of discipline and rigor (Miller and Page 2007). This reveals a
deeper concern: how to build grounds of confidence that an inference regarding a simulated process is a correct or valid inference for the actual process (Van Horn 1971). Disciplined approaches for assessing the validity of computational models have been developed and practiced in the modeling community for addressing this concern (Carley 1996; Burton and Obel 1995; Van Horn 1971). Although validation rarely results in proof that an agent-based simulation is a true model of the real world, it improves our faith in the insights gained from the simulation.
Limitations
This study represents the starting point of an exploration into bottom-up IT use processes. At this point, the CAS model of IT use is still analytical in nature. It provides a conceptual view of the inner workings of a bottom-up IT use process, but has not yet surfaced explanatory or predictive statements. This limitation suggests a future research opportunity: researchers can collate and compare bottom-up IT use processes in different organizational contexts, distilling the explanatory or predictive relationships embedded in the observed processes. In addition, the CAS model of IT use only captures usersystem interactions and interpersonal interactions, as the IT use literature has suggested the important role of these two types of interactions in shaping the uses and consequences of information technology (Kane and Alavi 2008). It can be argued that interactions among IT features are another source of variation in IT use patterns and outcomes, since they can also influence the characteristics and functionalities of IT systems. Future research may seek to gain insights regarding the interactions among IT features, expanding the CAS model of IT use by adding this third type of interaction.
Conclusions
In this paper, a complex adaptive systems model that is believed to afford researchers a robust framework for examining the emergence of IT use has been proposed. The CAS
529
model of IT use offers both the concepts and the instrument for researchers to formulate and evaluate propositions regarding causal pathways in bottom-up IT use processes. Researchers will be able to better appreciate the power and mechanism of emergence, and recognize the importance of interactions through the lens of the CAS model. The belief is that the IS research community will benefit significantly from taking advantage of the CAS framework and incorporating it into the fundamental theories and analytical tools of the field. Both the assessment in this paper of the IT use literature and other IS researchers evaluations of the general IS literature led to the concern of a lack of novel theories conceptualizing the multi-faceted nature of technology and its relationship to organizations (Akhlaghpour et al. 2009). The CAS model for IT use represents an attempt to develop a new theory for conceptualizing IT use processes. By theorizing IT artifacts as basic entities of actions (i.e., agents) in an IT use process, the CAS model provides a natural scheme for researchers to specify technologys material properties and inherent malleability as the attributes and behavioral rules of IT agents. This enables IS researchers to embrace the ensemble nature of technology, achieving deeper and more holistic analyses of the active role of IT in both business and social life. It is hoped that this study stimulates greater research attention in the analytical power of the CAS model and opens up new vistas for the advancement of the IS field. Acknowledgments
I am highly indebted to the senior editor, associate editor, and two anonymous reviewers for their exceptionally constructive and encouraging comments during the review process.
References
Akhlaghpour, S., Wu, J., Lapointe, L., and Pinsonneault, A. 2009. Re-examining the Status of IT in IT Research: An Update on Orlikowski and Iacono (2001), in Proceedings of the 15th Americas Conference on Information Systems, San Francisco, CA, August 6-9. Amaral, L. A. N., and Uzzi, B. 2007. Complex Systems: A New Paradigm for the Integrative Study of Management, Physical, and Technological Systems, Management Science (53:7), pp. 1033-1035. Anderson, P. 1999. Complexity Theory and Organization Science, Organization Science (10:3), pp. 216-232. Arrow, H., McGrath, J. E., and Berdahl, J. L. 2000. Small Groups as Complex Systems: Formation, Coordination, Development and Adaptation, Thousand Oaks, CA: Sage Publications. Axelrod, R. 1997. The Complexity of Cooperation, Princeton, NJ: Princeton University Press.
Axelrod, R., and Cohen, M. D. 1999. Harnessing Complexity: Organizational Implications of a Scientific Frontier, New York: The Free Press. Barki, H., Titah, R., and Boffo, C. L. 2007. Information System Use Related Activity: An Expanded Behavioral Conceptualization of Individual-Level Information System Use, Information Systems Research (18:2), pp. 173-192. Benbasat, I., and Zmud, R. W. 20003. The Identity Crisis Within the IS Discipline: Defining and Communicating the Disciplines Core Properties, MIS Quarterly (27:2), pp. 183-194. Broadbent, M., Weill, P., Clair, D. S., and Kearney, A. T. 1999. The Implications of Information Technology Infrastructure for Business Process Redesign, MIS Quarterly (23:2), pp. 159-182. Burkhardt, M. E. 1994. Social Interaction Effects Following a Technological Change: A Longitudinal Investigation, Academy of Management Journal (37:4), pp. 869-898. Burton, R. M. 2003. Computational Laboratories for Organization Science: Questions, Validity and Docking, Computational & Mathematical Organization Theory (9), pp. 91-108. Burton, R. M., DeSanctis, G., and Obel, B. 2006. Organizational Design: A Step-by-Step Approach, New York: Cambridge University Press. Burton, R. M., and Obel, B. 1980. A Computer Simulation Test of the M-Form Hypothesis, Administrative Science Quarterly (25:3), pp. 457-466. Burton, R. M., and Obel, B. 1995. The Validity of Computational Models in Organization Science: From Model Realism to Purpose of the Model, Computational and Mathematical Organization Theory (1:1), pp. 57-71. Burton-Jones, A., and Gallivan, M. J. 2007. Toward a Deeper Understanding of System Usage in Organizations: A Multilevel Perspective, MIS Quarterly (31:4), pp. 657-679. Burton-Jones, A., and Straub, D. W. 2006. Reconceptualizing System Usage: An Approach and Empirical Test, Information Systems Research (17:3), pp. 228-322. Byrd, T. A., and Turner, D. E. 2000. Measuring the Flexibility of Information Technology Infrastructure: Exploratory Analysis of a Construct, Journal of Management Information Systems (17:1), pp. 167-208. Canessa, E., and Riolo, R. L. 2006. An Agent-Based Model of the Impact of Computer-Mediated Communication on Organizational Culture and Performance: An Example of the Application of Complex Systems Analysis Tools to the Study of CIS, Journal of Information Technology (21:4), pp. 272-283. Carley, K. M. 1995. Communication Technologies and Their Effect on Cultural Homogeneity, Consensus, and the Diffusion of New Ideas, Sociological Perspectives (38:4), pp. 547-571. Carley, K. M. 1996. Validating Computational Models, available at http://www2.econ.iastate.edu/tesfatsi/EmpValid.Carley.pdf. Compeau, D. R., and Higgins, C. A. 1995. Computer SelfEfficacy: Development of a Measure and Initial Test, MIS Quarterly (19:2), pp. 189-211. Conway, R. W., Johnson, B. M., and Maxwell, W. L. 1959. Some Problems of Digital Systems Simulation, Management Science (6:1), pp. 92-110. Curseu, P. L. 2006. Emergent States in Virtual Teams: A Complex Adaptive Systems Perspective, Journal of Information Technology (21:4), pp. 249-261.
530
Daft, R. L., and Lengel, R. H. 1986. Organizational Information Requirements, Media Richness and Structural Design, Management Science (32:5), pp. 554-571. Davis, F. D. 1989. Perceived Usefulness, Perceived Ease of Use, and User Acceptance of Information Technology, MIS Quarterly (13:3), pp. 319-339. Davis, F. D., Bagozzi, R. P., and Warshaw, P. R. 1989. User Acceptance of Computer Technology: A Comparison of Two Theoretical Models, Management Science (35:8), pp. 982-1003. Davis, J. P., Eisenhardt, K. M., and Bingham, C. B. 2007. Developing Theory through Simulation Methods, Academy of Management Review (32:2), pp. 480-499. Dennis, A. R., Wixom, B. H., and Vandenberg, R. J. 2001. Understanding Fit and Appropriation Effects in Group Support Systems via Meta-Analysis, MIS Quarterly (25:2), pp. 167-193. DeSanctis, G., and Poole, M. S. 1994. Capturing the Complexity in Advanced Technology Use: Adaptive Structuration Theory, Organization Science (5:2), pp. 121-147. Drazin, R., and Sandelands, L. 1992. Autogenesis: A Perspective on the Process of Organizing, Organization Science (3:2), pp. 230-249. Epstein, J. M. 2006. Generative Social Science: Studies in AgentBased Computational Modeling, Princeton, NJ: Princeton University Press. Epstein, J. M., and Axtell, R. 1996. Growing Artificial Societies: Social Science from the Bottom Up, Washington, DC: Brookings Institution Press. Fulk, J., Schmitz, J., and Ryu, D. 1995. Cognitive Elements in the Social Construction of Communication Technology, Management Communication Quarterly (8), pp. 259-288. Gasser, L. 1986. The Integration of Computing and Routine Work, ACM Transactions on Office Information Systems (4:3), pp. 205-225. Gell-Mann, M. 1994. The Quark and the Jaguar: Adventures in the Simple and the Complex, New York: W. H. Freeman. Goodhue, D. L., and Thompson, R. L. 1995. Task-Technology Fit and Individual Performance, MIS Quarterly (19:2), pp. 213-236. Gregor, S. 2006. The Nature of Theory in Information Systems, MIS Quarterly (30:3), pp. 611-642. Harrison, J. R., Lin, Z., Carroll, G. R., and Carley, K. M. 2007. Simulation Modeling in Organizational and Management Research, Academy of Management Review (32:4), pp. 1229-1245. Holland, J. H. 1995. Hidden Order: How Adaptation Builds Complexity, Reading, MA: Perseus Books. Jasperson, J., Carter, P. E., and Zmud, R. W. 2005. A Comprehensive Conceptualization of Post-Adoptive Behaviors Associated with Information Technology Enabled Work Systems, MIS Quarterly (29:3), pp. 525-557. Jones, M. R., and Karsten, H. 2008. Giddenss Structuration Theory and Information Systems Research, MIS Quarterly (32:1), pp. 127-157. Kane, G. C., and Alavi, M. 2007. Information Technology and Organizational Learning: An Investigation of Exploration and Exploitation Processes, Organization Science (18:5), pp. 796-812.
Kane, G. C., and Alavi, M. 2008. Casting the Net: A Multimodal Network Perspective on User-System Interactions, Information Systems Research (19:3), pp. 253-272. Kozlowski, S. W. J., and Klein, K. J. 2000. A Multilevel Approach to Theory and Research in Organizations, in Multilevel Theory, Research, and Methods in Organizations, K. J. Klein and S. W. J. Kozlowski (eds.), San Francisco: Jossey-Bass, pp. 3-90. Kumar, R. L. 2004. A Framework for Assessing the Business Value of Information Technology Infrastructures, Journal of Management Information Systems (21:2), pp. 11-32. Lind, M. R., and Zmud, R. W. 1991. The Influence of a Convergence in Understanding Between Technology Providers and Users on Information Technology Innovativeness, Organization Science (2:2), pp. 195-217. Majchrzak, A., Rice, R. E., Malhotra, A., King, N., and Ba, S. 2000. Technology Adaptation: The Case of a ComputerSupported Inter-Organizational Virtual Team, MIS Quarterly (24:4), pp. 569-600. March, J. G. 1991. Exploration and Exploitation in Organizational Learning, Organization Science (2:1), pp. 71-87. Markus, M. L., and Robey, D. 1988. Information Technology and Organizational Change: Causal Structure in Theory and Research, Management Science (34:5), pp. 583-598. Markus, M. L., and Silver, M. S. 2008. A Foundation for the Study of IT Effects: A New Look at DeSanctis and Pooles Concepts of Structural Features and Spirit, Journal of the Association for Information Systems (9:10/11), pp. 609-632. McKelvey, B. 1997. Quasi-Natural Organization Science, Organization Science (8:4), pp. 352-380. Merali, Y. 2004. Complexity and Information Systems, in Social Theory and Philosophy for Information Systems, J. Mingers and L. Willcocks (eds.), Chichester, UK: John Wiley & Sons, pp. 407-446. Miller, J. H., and Page, S. E. 2007. Complex Adaptive Systems: An Introduction to Computational Models of Social Life, Princeton, NJ: Princeton University Press. Monge, P. R., and Contractor, N. S. 2003. Theories of Communication Networks, New York: Oxford University Press. Orlikowski, W. J. 1992. The Duality of Technology: Rethinking the Concept of Technology in Organizations, Organization Science (3:3), pp. 398-427. Orlikowski, W. J. 1993. CASE Tools as Organizational Change: Investigating Incremental and Radical Changes in Systems Development, MIS Quarterly (17:3), pp. 309-340. Orlikowski, W. J. 1996. Improvising Organizational Transformation over Time: A Situated Change Perspective, Information Systems Research (7:1), pp. 63-92. Orlikowski, W., and Robey, D. 1991. Information Technology and the Structuring of Organizations, Information Systems Research (2:2), pp. 143-169. Orlikowski, W. J., Yates, J., Okamura, K., and Fujimoto, M. 1995. Shaping Electronic Communication: The Metastructuring of Technology in the Context of Use, Organization Science (6:4), pp. 423-444.
531
Ortiz de Guinea, A., and Markus, M. L. 2009. Why Break the Habit of A Lifetime? Rethinking the Roles of Intention, Habit, and Emotion in Continuing Information Technology Use, MIS Quarterly (33:3), pp. 433-444. Raghu, T. S., Jayaraman, B., and Rao, H. R. 2004. Toward an Integration of Agent-and Activity-Centric Approaches in Organizational Process Modeling: Incorporating Incentive Mechanisms, Information Systems Research (15:4), pp. 316-335. Rivkin, J. W., and Siggelkow, N. 2003. Balancing Search and Stability: Interdependencies Among Elements of Organizational Design, Management Science (49:3), pp. 290-311. Romme, A. G. L. 2004. Unanimity Rule and Organizational Decision Making: A Simulation Model, Organization Science (15:6), pp. 704-718. Samuel, A. 1959. Some Studies in Machine Learning Using the Game of Checkers, IBM Journal of Research and Development (3), pp. 210-229. Sandelands, L. E., and Drazin, R. 1989. On the Language of Organization Theory, Organization Studies (10), pp. 457-478. Sarker, S., Sarker, S., and Sidorova, A. 2006. Understanding Business Process Change Failure: An Actor-Network Perspective, Journal of Management Information Systems (23:1), pp. 51-86. Sastry, M. A. 1997. Problems and Paradoxes in a Model of Punctuated Organizational Change, Administrative Science Quarterly (42:2), pp. 237-275. Sawyer, R. K. 2005. Social Emergence: Societies as Complex Systems, Cambridge, UK: Cambridge University Press. Schelling, T. C. 1969. Models of Segretation, The American Economic Review (59:2), pp. 488-493. Scott, S. V., and Wagner, E. L. 2003. Networks, Negotiations, and New Times: The Implementation of Enterprise Resource Planning into an Academic Administration, Information and Organization (13:4), pp. 285-313. Simon, H. A. 1990. Prediction and Prescription in Systems Modeling, Operations Research (38), pp. 7-14. Staw, B. M., Sandelands, L. E., and Dutton, J. E. 1981. Threat Rigidity Effects in Organizational Behavior: A Multilevel Analysis, Administrative Science Quarterly (26:4), pp. 501-524. Tesfatsion, L. 2010. General Software and Toolkits: Agent-Based Computational Economics and Complex Adaptive Systems, available at http://econ2.econ.iastate.edu/tesfatsi/acecode.htm. Van Horn, R. 1971. Validation of Simulation Results, Management Science (17:5), pp. 247-258.
Venkatesh, V., Brown, S. A., Maruping, L. M., and Bala, H. 2008. Predicting Different Conceptualizations of System Use: The Competing Roles of Behavioral Intention, Facilitating Conditions, and Behavioral Expectation, MIS Quarterly (32:3), pp. 483-502. Venkatesh, V., and Davis, F. D. 2000. A Theoretical Extension of the Technology Acceptance Model: Four Longitudinal Field Studies, Management Science (46:2), pp. 186-204. Venkatesh, V., Morris, M. G., Davis, G. B., and Davis, F. D. 2003. User Acceptance of Information Technology: Toward a Unified View, MIS Quarterly (27:3), pp. 425-478. Vidgen, R., and Wang, X. 2009. Coevolving Systems and the Organization of Agile Software Development, Information Systems Research (20:3), pp. 355-376. von Neumann, J. 1996. Theory of Self-Replicating Automata, Urbana, IL: University of Illinois Press. Walsham, G., and Sahay, S. 1999. GIS for District-Level Administration in India: Problems and Opportunities, MIS Quarterly (23:1), pp. 39-65. Wheeler, B. C., and Valacich, J. S. 1996. Facilitation, GSS, and Training as Sources of Process Restrictiveness and Guidance for Structured Group Decision Making: An Empirical Assessment, Information Systems Research (7:4), pp. 429-450. Wilensky, U. 1999. NetLogo, Center for Connected Learning and Computer-Based Modeling. Northwestern University, Evanston, IL (available at http://ccl.northwestern.edu/netlogo). Yates, J., Orlikowski, W. J., and Okamura, K. 1999. Explicit and Implicit Structuring of Genres in Electronic Communication: Reinforcement and Change of Social Interaction, Organization Science (10:1), pp. 83-103.
About the Author

Ning Nan is an assistant professor in the Management Information Systems Division, Michael F. Price College of Business, University of Oklahoma. Her research interests include individual- and collective-level information technology uses and impacts, distributed collaboration, and software development project management. She holds a Ph.D. from the University of Michigan, an M.A. from the University of Minnesota, and a B.A. from Peking University.
532
THEORY AND REVIEW
CAPTURING BOTTOM-UP INFORMATION TECHNOLOGY USE PROCESSES: A COMPLEX ADAPTIVE SYSTEMS MODEL
Ning Nan
Michael F. Price College of Business, University of Oklahoma, 307 W. Brooks, Room 305D, Norman, OK 73019 U.S.A. {nnan@ou.edu}
Appendix A
Summary of the Literature
The studies are selected from all articles published in nine premium journals that regularly publish scholarly research on IT use: MIS Quarterly, Information Systems Research, Organization Science, Management Science, Administrative Science Quarterly, Decision Sciences, Journal of Management Information Systems, IEEE Transactions on Engineering Management, and Database for Advances in Information Systems.
Research Perspective Technology acceptance
Description This research relies on variance-based models to examine antecedents of initial and continued IT use. The antecedents include individual cognition such as perceived usefulness of a new IT system and organizational factors such as management influence.
Related Studies Bhattacherjee and Sanford 2006; Cooper and Zmud 1990; Davis 1989; Davis et al 1989; Edmondson et al. 2001; Joshi 1991; Joshi et al. 1999; Kim and Malhotra 2005; Kraut et al. 1998; Leonard-Barton and Deschamps 1988; Limayem et al. 2007; Lucas et al. 1988; Robertson 1989; Sabherwal et al. 2006; Taylor and Todd 1992; Tyre and Hauptman 1992; Venkatesh et al. 2003; Venkatesh and Davis 2000; Venkatesh et al. 2008; Zhu and Kraemer 2005 Goodhue 1998; Goodhue and Thompson 1995; Zigurs et al. 1999
Task technology fit
This research examines the correspondence between task requirements, individual abilities, and the functionality of an IT system. It highlights the importance of the alignment between the three aspects in inducing positive IT-enabled task performance. This research seeks to identify the sequence of activities (often referred to as phases) in a typical IT use process and to prescribe the stage models as plans for IT use management.
Planned change
El Sawy 1985; Lassila and Brancheau 1999; Nelson and Cheney 1987; Raho et al. 1987
MIS Quarterly Vol. 35 No. 2Appendices/June 2011
A1
Research Perspective System dynamics
Description These studies employ system dynamics models to examine how the accumulative and marginal effects (i.e., stock and flow) of human cognition, such as learning and commitment to using a new technology, can affect IT use behaviors and organizational performance. This research views IT use as social political processes and employs the actor-network framework to examine how ongoing negotiations among alliance (i.e., actor-networks) with heterogeneous political interests lead to alignment of interests, which eventually enables IT use.
Related Studies Braa et al. 2004; Sarker et al. 2006; Walsham and Sahay 1999
Actor-network analysis
Avgerou and McGrath 2007; Boudreau and Robey 2005; Davidson and Chismar 2007; Garud and Kumaraswamy 2005; Lapointe and Rivard 2005; Leonardi 2007; Lyytinen and Rose 2003; Majchrzak et al. 2000; Malhotra et al. 2001; Orlikowski 1996, 2000; Robey et al. 2002; Robey and Sahay 1996; Tyre and Orlikowski 1994; Volkoff et al. 2007
Social construction of technology
This research assumes that IT use is neither determined by human actors nor technologies, but enacted through interactions between the two without a priori plans. It usually relies on case studies to capture IT use enactment processes.
A2
A3
A4
Appendix C
Pseudo-Code of a Simulation Session
Create 53 employees (50 specialists, 2 managers, and 1 director) Ask each employee { Set the 30-tuple, with each dimension takes a value of -1, 0, or 1 with equal probabilities Set learning rate p1 = the learning rate treatment of the current simulation session If I am a manager [Set learning rate = 1.25 p1] If I am a director [Set learning rate = 1.5 p1] If the current workplace rigidity treatment = rigidity [form a tie with a randomly chosen superior] Else [form a tie with another randomly chosen employee] } Create the ITSS Ask the ITSS { Set the 30-tuple, with each dimension takes a value of 0 Set the flexibility p2 = the ITSS flexibility treatment of the current simulation session } Create the work requirements Ask the work requirements { Set the 30-tuple, with each dimension takes a value of -1, 0, or 1 with equal probabilities } Run one tick of the model clock { Ask the ITSS [adapts to the majority practices of employees] Ask each employee [learn from the ITSS and learn from each other (the order of these two actions is randomly determined)] } Repeat the Run one tick of the model clock procedure 12 times Set assimilation of the ITSS = average (the proportion of identical values between the 30-tuples of the ITSS and the employees) Set IT-based work performance = average (the proportion of identical values between the 30-tuples of the work requirements and the employees)
References
Avgerou, C., and McGrath, K. 2007. Power, Rationality, and the Art of Living through Socio-Technical Change, MIS Quarterly (31:2), pp. 295-315. Bhattacherjee, A., and Sanford, C. 2006. Influence Processes for Information Technology Acceptance: An Elaboration Likelihood Model, MIS Quarterly (30:4), pp. 805-825. Black, L. J., Carlile, P. R., and Repenning, N. P. 2004. A Dynamic Theory of Expertise and Occupational Boundaries in New Technology Implementation: Building on Barleys Study of CT Scanning, Administrative Science Quarterly (49:4), pp. 572-607. Boudreau, M.-C., and Robey, D. 2005. Enacting Integrated Information Technology: A Human Agency Perspective, Organization Science (16:1), pp. 3-18. Braa, J. R., Monteiro, E., and Sahay, S. 2004. Networks of Action: Sustainable Health Information Systems across Developing Countries, MIS Quarterly (28:3), pp. 337-362. Cooper, R. B., and Zmud, R. W. 1990. Information Technology Implementation Research: A Technological Diffusion Approach, Management Science (36:2), pp. 123-139.
A5
Davidson, E. J., and Chismar, W. G. 2007. The Interaction of Institutionally Triggered and Technology-Triggered Social Structure Change: An Investigation of Computerized Physician Order Entry, MIS Quarterly (31:4), pp. 739-758. Davis, F. D. 1989. Perceived Usefulness, Perceived Ease of Use, and User Acceptance of Information Technology, MIS Quarterly (13:3), pp. 319-339. Davis, F. D., Bagozzi, R. P., and Warshaw, P. R. 1989. User Acceptance of Computer Technology: A Comparison of Two Theoretical Models, Management Science (35:8), pp. 982-1003. Edmondson, A. C., Bohmer, R. M., and Pisano, G. P. 2001. Disrupted Routines: Team Learning and New Technology Implementation in Hospitals, Administrative Science Quarterly (46:4), pp. 685-716. El Sawy, O. A. 1985. Implementation by Cultural Infusion: An Approach for Managing the Introduction of Information Technologies, MIS Quarterly (9:2), pp. 131-140. Garud, R., and Kumaraswamy, A. 2005. Vicious and Virtuous Circles in the Management of Knowledge: The Case of Infosys Technologies, MIS Quarterly (29:1), pp. 9-33. Goodhue, D. L. 1998. Development and Measurement Validity of a Task-Technology Fit Instrument for User Evaluations of Information Systems, Decision Sciences (29:1), pp. 105-139. Goodhue, D. L., and Thompson, R. L. 1995. Task-Technology Fit and Individual Performance, MIS Quarterly (19:2), p 213-236. Joshi, K. 1991. A Model of Users Perspective on Change: The Case of Information Systems Technology Implementation, MIS Quarterly (15:2), pp. 229-242. Joshi, K., Joshi, K., and Lauer, T. W. 1999. Transition and Change During the Implementation of A Computer-Based Manufacturing Process Planning System: An Analysis Using the Equity Implementation Model, Engineering Management, IEEE Transactions on (46:4), pp. 407-416. Kim, S. S., and Malhotra, N. K. 2005. A Longitudinal Model of Continued IS Use: An Integrative View of Four Mechanisms Underlying Postadoption Phenomena, Management Science (51:5), pp. 741-755. Kraut, R. E., Rice, R. E., Cool, C., and Fish, R. S. 1998. Varieties of Social Influence: The Role of Utility and Norms in the Success of a New Communication Medium, Organization Science (9:4), p 437-453. Lapointe, L., and Rivard, S. 2005. A Multilevel Model of Resistance to Information Technology Implementation, MIS Quarterly (29:3), pp. 461-491. Lapointe, L., and Rivard, S. 2007. A Triple Take on Information System Implementation, Organization Science (18:1), pp. 89-107. Lassila, K. S., and Brancheau, J. C. 1999. Adoption and Utilization of Commercial Software Packages: Exploring Utilization Equilibria, Transitions, Triggers, and Tracks, Journal of Management Information Systems (16:2), pp. 63-90. Leonard-Barton, D., and Deschamps, I. 1988. Managerial Influence in the Implementation of New Technology, Management Science (34:10), pp. 1252-1265. Leonardi, P. M. 2007. Activating the Informational Capabilities of Information Technology for Organizational Change, Organization Science (18:5), pp. 813-831. Limayem, M., Hirt, S. G., and Cheung, C. M. K. 2007. How Habit Limits the Predictive Power of Intention: The Case of Information Systems Continuance, MIS Quarterly (31:4), pp. 705-737. Lucas, Jr., H. C., Walton, E. J., and Ginzberg, M. J. 1988. Implementing Packaged Software, MIS Quarterly (12:4), pp. 537-549. Lyytinen, K., and Rose, G. M. 2003. The Disruptive Nature of Information Technology Innovations: The Case of Internet Computing in Systems Development Organizations, MIS Quarterly (27:4), pp. 557-595. Majchrzak, A., Rice, R. E., Malhotra, A., King, N., and Ba, S. 2000. Technology Adaptation: The Case of a Computer-Supported InterOrganizational Virtual Team, MIS Quarterly (24:4), pp. 569-600. Malhotra, A., Majchrzak, A., Carman, R., and Lott, V. 2001. Radical Innovation Without Collocation: A Case Study at Boeing-Rocketdyne, MIS Quarterly (25:2), pp. 229-249. Nelson, R. R., and Cheney, P. H. 1987. Training End Users: An Exploratory Study, MIS Quarterly (11:4), pp. 547-559. Orlikowski, W. J. 1996. Improvising Organizational Transformation over Time: A Situated Change Perspective, Information Systems Research (7:1), pp. 63-92. Orlikowski, W. J. 2000. Using Technology and Constituting Structures: A Practice Lens for Studying Technology in Organizations, Organization Science (11:4), pp. 404-428. Raho, L. E., Belohlav, J. A., and Fiedler, K. D. 1987. Assimilating New Technology into the Organization: An Assessment of McFarlan and McKenneys Model, MIS Quarterly (11:1), pp. 47-57. Repenning, N. P. 2002. A Simulation-Based Approach to Understanding the Dynamics of Innovation Implementation, Organization Science (13:2), pp. 109-127. Robertson, D. C. 1989. Social Determinants of Information Systems Use, Journal of Management Information Systems (5:4), pp. 55-71. Robey, D., Ross, J. W., and Boudreau, M. C. 2002. Learning to Implement Enterprise Systems: An Exploratory Study of the Dialectics of Change, Journal of Management Information Systems (19:1), pp. 17-46. Robey, D., and Sahay, S. 1996. Transforming Work Through Information Technology: A Comparative Case Study of Geographic Information Systems in County Government, Information Systems Research (7:1), pp. 93-110.
A6
Sabherwal, R., Jeyaraj, A., and Chowa, C. 2006. Information System Success: Individual and Organizational Determinants, Management Science (52:12), pp. 1849-1864. Sarker, S., Sarker, S., and Sidorova, A. 2006. Understanding Business Process Change Failure: An Actor-Network Perspective, Journal of Management Information Systems (23:1), pp. 51-86. Taylor, S., and Todd, P. 1995. Assessing IT Usage: The Role of Prior Experience, MIS Quarterly (19:4), pp. 561-570. Tyre, M. J., and Hauptman, O. 1992. Effectiveness of Organizational Responses to Technological Change in the Production Process, Organization Science (3:3), pp. 301-320. Tyre, M. J., and Orlikowski, W. J. 1994. Windows of Opportunity: Temporal Patterns of Technological Adaptation in Organizations, Organization Science (5:1), pp. 98-118. Venkatesh, V., Brown, S. A., Maruping, L. M., and Bala, H. 2008. Predicting Different Conceptualizations of System Use: The Competing Roles of Behavioral Intention, Facilitating Conditions, and Behavioral Expectation, MIS Quarterly (32:3), pp. 483-502. Venkatesh, V., and Davis, F. D. 2000. A Theoretical Extension of the Technology Acceptance Model: Four Longitudinal Field Studies, Management Science (46:2), pp. 186-204. Venkatesh, V., Morris, M. G., Davis, G. B., and Davis, F. D. 2003. User Acceptance of Information Technology: Toward a Unified View, MIS Quarterly (27:3), pp. 425-478. Volkoff, O., Strong, D. M., and Elmes, M. B. 2007. Technological Embeddedness and Organizational Change, Organization Science (18:5), pp. 832-848. Walsham, G., and Sahay, S. 1999. GIS for District-Level Administration in India: Problems and Opportunities, MIS Quarterly (23:1), pp. 39-65. Zhu, K., and Kraemer, K. L. 2005. Post-Adoption Variations in Usage and Value of E-Business by Organizations: Cross-Country Evidence from the Retail Industry, Information Systems Research (16:1), pp. 61-84. Zigurs, I., Buckland, B. K., Connolly, J. R., and Wilson, E. V. 1999. A Test of Task-Technology Fit Theory for Group Support Systems, Database for Advances in Information Systems (30:3/4), pp. 34-51.
A7
Copyright of MIS Quarterly is the property of MIS Quarterly & The Society for Information Management and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.

TI

Загружено:

Сведения о документе

Исходное описание:

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

TI

Загружено:

Авторское право:

Доступные форматы

THEORY AND REVIEW

MIS Quarterly Vol. 35 No. 2 pp. 505-532/June 2011

Nan/Capturing BottomUp IT Use Processes

MIS Quarterly Vol. 35 No. 2/June 2011

Nan/Capturing BottomUp IT Use Processes

IT Collective level (2) Bottom-up (6) User (3) Top-down (5)

IT Individual level (1) t0 tn

Figure 1. The IT Use Construct

MIS Quarterly Vol. 35 No. 2/June 2011

Nan/Capturing BottomUp IT Use Processes

Theory of Complex Adaptive Systems

A Complex Adaptive Systems Model of IT Use

MIS Quarterly Vol. 35 No. 2/June 2011

Nan/Capturing BottomUp IT Use Processes

Table 1. Basic Concepts of the Theory of CAS

MIS Quarterly Vol. 35 No. 2/June 2011

Nan/Capturing BottomUp IT Use Processes

Figure 2. An Overview of the Theory of CAS

Extending the Theory of CAS to the IT Use Context

MIS Quarterly Vol. 35 No. 2/June 2011

Nan/Capturing BottomUp IT Use Processes

MIS Quarterly Vol. 35 No. 2/June 2011

Nan/Capturing BottomUp IT Use Processes

Figure 3. The CAS Model of IT Use

Table 2. Mapping Between CAS and Extant IT Use Concepts

Connection Flow Environment Structure

MIS Quarterly Vol. 35 No. 2/June 2011

Nan/Capturing BottomUp IT Use Processes

MIS Quarterly Vol. 35 No. 2/June 2011

Nan/Capturing BottomUp IT Use Processes

MIS Quarterly Vol. 35 No. 2/June 2011

Nan/Capturing BottomUp IT Use Processes

Instrument of CAS: Agent-Based Modeling

MIS Quarterly Vol. 35 No. 2/June 2011

Nan/Capturing BottomUp IT Use Processes

Using the CAS Model of IT Use

The Conceptual Model

MIS Quarterly Vol. 35 No. 2/June 2011

Nan/Capturing BottomUp IT Use Processes

Table 3. Different Purposes of Agent-Based Simulation

2. What is the impact of ITSS flexibility?

3. What is the impact of workplace rigidity?

MIS Quarterly Vol. 35 No. 2/June 2011

Nan/Capturing BottomUp IT Use Processes

Table 5. The Conceptual Definition of the CAS Model

Individual differences Mental activities

Learning rate Learning from ITSS features

MIS Quarterly Vol. 35 No. 2/June 2011

Nan/Capturing BottomUp IT Use Processes

Table 5. The Conceptual Definition of the CAS Model

Movement of intangible IT resources

Knowledge codification and reuse

Interpersonal ties Movement of intangible IT resources

Organizational context at Zeta

IT-based work performance

MIS Quarterly Vol. 35 No. 2/June 2011

Nan/Capturing BottomUp IT Use Processes

MIS Quarterly Vol. 35 No. 2/June 2011

Nan/Capturing BottomUp IT Use Processes

Work Requirements [-1 1 0 . . . 0 0 1] 30-tuple d

Employees [-1 -1 1 . . . 0 0 0] [0 0 0 . . . -1 0 -1] . . [1 1 1 . . . 0 0 -1] 30-tuple .

Figure 4. The Design of the Agent-Based Model

MIS Quarterly Vol. 35 No. 2/June 2011