- Email: [email protected]
Social network analysis in accounting information systems research
International Journal of Accounting Information Systems 14 (2013) 127–137
Contents lists available at SciVerse ScienceDirect
International Journal of Accounting Information Systems
Social network analysis in accounting information systems research James Worrell 1, Molly Wasko ⁎, Allen Johnston 2 University of Alabama at Birmingham, 319 Business and Engineering Complex, 1530 3rd Avenue South, Birmingham, AL 35294-4460, United States
a r t i c l e
i n f o
Article history: Received 3 August 2010 Received in revised form 10 May 2011 Accepted 16 June 2011 Keywords: Social network analysis Centrality Methods in accounting research Fraud
a b s t r a c t This paper introduces social network analysis as an alternative research method for conducting accounting information systems related research. With advances in information and communication technologies, transaction data are being recorded in electronic form, resulting in a variety of research opportunities to examine dyadic interactions. A network consists of a set of nodes connected by ties. Social network research focuses on how outcomes are influenced not just by the attributes of the nodes (e.g. individuals), but also by the ties connecting nodes to each other. The nodes are typically conceptualized as actors, such as individuals, teams, or organizations. A unique network structure is created to reflect each different type of tie, such as trust, advice, collocation, or organizational affiliation. Social network analysis can be used for research examining individual, dyadic or network levels of analyses, and is a powerful tool for conducting multi-method research. Given the vast amounts of trace electronic data collected via accounting information systems, this paper reviews how social network analysis not only opens new research avenues for accounting information systems researchers, but identifies opportunities for the field of accounting information systems to inform social network research by identifying new network structures and dynamics leveraging transactional data. © 2011 Elsevier Inc. All rights reserved.
1. Introduction There has been a general shift in management research over the past decade towards more relational theories of organizations that view actions and actors not as independent, autonomous agents, but as embedded within ⁎ Corresponding author. Tel.: + 1 205 934 8806(office); fax: + 1 205 975 4429. E-mail addresses: [email protected] (J. Worrell), [email protected] (M. Wasko), [email protected] (A. Johnston). 1 Tel.: + 1 205 934 8820. 2 Tel.: + 1 205 934 8870. 1467-0895/$ – see front matter © 2011 Elsevier Inc. All rights reserved. doi:10.1016/j.accinf.2011.06.002
128
J. Worrell et al. / International Journal of Accounting Information Systems 14 (2013) 127–137
socio-technical systems. In contrast to theories that examine individuals based on their attributes, such as gender, age, education, or occupation, social network perspectives focus on how the relationships between entities, such as individuals, functional units, or organizations, influence interactions and outcomes. The concept of a “network” is broad and can be applied to a variety of phenomena where a set of relations is ascribed to an identified set of actors. What unites social network perspectives is the focus on the patterns and implications of the ties within a collective (Scott, 1991; Wasserman and Faust, 1994). For example, at the individual level ties facilitate the spread of information among network participants, enable the flow of both tangible and intangible resources among network members, and place constraints on each member's behavior (Burt, 1992). Social network research focuses on the significance of relationships as essential for understanding social action, but varies widely in the attributes that are studied. A network is defined as a set of nodes connected by ties. Nodes are typically “actors”, and can be people, teams, organizations or information systems. Relations, or ties, connect the actors and can vary in content, direction, and relational strength, all of which influence the dynamics of the network (Garton et al., 1999). The content of ties refers to the resource exchanged or common bond, such as information, money, advice, or kinship. The direction of ties indicates an “ego” who gives the resource, and the “alter” who receives it, although ties in some networks are undirected, such as a shared attribute (e.g. gender), or joint membership on a team. The relational strength of ties pertains to the level of activity, such as quantity of communications, or the intensity, such as the social influence exerted by the tie, indicating that ties can be valued or weighted. For instance, the relational strength of ties could indicate the amount of energy, emotional intensity, intimacy, commitment or trust connecting the actors. Relational ties are often studied in management research as important aspects of social influence that can exert control, such as social punishments or ostracism (Ostrom, 1990). Other aspects of social influence foster cohesion and prosocial behavior in the network, such as trust, identification, the diffusion of information and commitment (Coleman, 1990; Nahapiet and Ghoshal, 1998). Each tie defines a different network, and while some ties are often related (a trust network is often correlated to a friendship network), ties are often assumed to function differently. Not all ties are considered to have positive outcomes; for instance, network research is often used to map the flow of disease or terrorist networks. Therefore, some network research focuses on how to improve the flow of the resource through the network, such as adoption of a new accounting information system, or how to disrupt the flow of resources in the network, such as taking out key nodes in a fraud network. Depending upon the theory being applied, some studies examine network variables as independent variables causing consequences, such as adoption of a technology or improved performance, while other studies examine network variables as dependent variables, identifying the causes underlying the pattern of network connections, how networks come to be, and how networks change over time (Borgatti and Foster, 2003).
2. Example of social network analysis This section describes a social network study to provide insights about how to apply social network analysis. The context for the study was a business school consisting of 89 faculty members organized into 7 departments. The purpose of the study was to understand how social networks impact the research performance of individual faculty, and the practical question was how collaboration among faculty members within a business school impacts an individual faculty's research performance (Smatt, 2009). The data for this study were collected via survey using a roster method with the names of all faculty members in the school listed. Each faculty member responded to survey items based upon their relationships with all other faculty members. Social network data can also be collected from archival data, such as using electronic trace data from transaction processing systems, electronic data interchange (EDI) or email communications. Data are recorded in a square matrix to indicate the ego (in the first column), the alter (in the first row) and the tie (weighted, directed and/or dichotomous) in the corresponding ego/alter cells. A separate network is created for each type of tie; for instance, in this study multiple networks were collected, including collocation, same department, hierarchical level (assistant, associate, full, and administrator), how well faculty members knew each other, trusted each other and sought advice from each other.
J. Worrell et al. / International Journal of Accounting Information Systems 14 (2013) 127–137
129
Faculty members were first given the roster of names and asked to indicate “who do you know well”. The data were recorded in a square social network matrix, and used to create the social network graph depicted in Fig. 1, which will be used to illustrate the concept of network centrality. One of the most common variables assessed in network research is centrality, or an actor's position in the overall network of relations (Scott, 1991; Wasserman and Faust, 1994). Freeman (1979) proposed three separate measures of centrality: betweenness, closeness, and degree centrality. Betweenness centrality refers to the extent to which an ego falls between two other actors in the network, revealing the potential point of control for resource flows. Closeness centrality is the extent to which an ego can easily connect with all other actors, or the amount of direct or indirect links that an actor has with others in the network. The measure of closeness reveals the efficiency of the network, indicating how quickly an actor can gain access to resources. Degree centrality refers to the amount of direct links an ego has with others in the network, uncovering how active/connected an ego is within the network, assessed by summing the total number of ties that an ego has with alters (Freeman, 1979). In a directed network, indegree centrality represents the number of alters selecting an ego (ties coming in) and outdegree centrality represents the number alters selected by an ego (ties going out). Fig. 1 is a graphical representation of the faculty network. The tie indicates that the ego faculty knew the alter faculty well, node size is based on indegree centrality (how many alters selected that ego), and the node color indicates membership in the same department. The different types of centralities are identified, demonstrating that even though an ego may have a large indegree centrality (indicating that the faculty is well known), that does not necessarily indicate high closeness or betweenness. For instance, the largest nodes (most well known faculty members) are generally only well known by others within their departments, not across departments. The gray circled area to the right is the accounting department, indicating that faculty members in this department have fewer ties among each other and among faculty members in other departments. It is common to see network data analyzed at the individual level and dyadic level. For example, at the individual level, centrality scores can be assessed for each individual and then combined with other data, such as education and experience, and used in regression analyses to predict performance at work. At the dyadic level, quadratic assignment procedure (QAP) and multiple regression-quadratic assignment procedure (MRQAP) can be used to predict whether one type of dyadic tie predicts another dyadic tie (for
Fig. 1. Network of business school faculty.
130
J. Worrell et al. / International Journal of Accounting Information Systems 14 (2013) 127–137
instance, the extent to which ego trusts alter is influenced by the extent to which ego is collocated with alter) (Baker and Hubert, 1981; Borgatti and Cross, 2003). Network analysis is also useful to identify characteristics of the network as a whole, such as to identify cliques, core/periphery structures, network density and/or centralization. For the faculty study depicted in Fig. 1, indegree centrality in the research advice network was used in structural equation modeling (SEM) to predict research productivity. The results indicated that centrality in the research advice network had a significant, positive impact on research productivity above and beyond tenure, department and other individual-level variables (Network centrality was the only independent variable used to predict research performance as assessed by number of top tier publications. The human capital variables were used as controls for alternative explanations.). This has important implications for encouraging faculty collaboration, especially among the accounting faculty in this study who collectively were less central in the overall research advice network. The purpose of this section was to provide an overview of social network analysis. For researchers wanting more detailed information about how to perform social network analysis, and definitions of the different types of network measures (such as centrality, density and core-periphery), we encourage them to review Wasserman and Faust's (1994) Social Network Analysis, which is widely considered the seminal text in this area. Another good starting reference is Scott's (1991) Social Network Analysis. The leading academic organization related to social network research is INSNA: the International Network for Social Network Analysis (www.insna.org). For individuals interested in learning more about the computer programs used to analyze social network data, INSNA sponsors an annual conference, called the Sunbelt Conference that offers professional development workshops from leading experts in the field (http://www. insna.org/sunbelt/). 3. Social network analysis in AIS research While a rich stream of research utilizing social network analysis exists within the sociology, anthropology, management and information systems disciplines, accounting researchers have been slow to add this method to their toolkits. Traditionally, social network analysis research has investigated networks in three manners: network connections which facilitate flows of resources between nodes, ties between nodes that influence behavior and enforce social norms, and networks themselves as either independent or dependent variables (Borgatti and Foster, 2003). In the following section, we provide a brief overview of accounting information systems (AIS) research to date that has used social network analysis. This is followed by potential research questions where social network analysis might inform existing AIS phenomena. 3.1. Research on network enterprises Accounting researchers have become increasingly interested in examining the dynamics of inter-organizational relations. In response to changing organizational dynamics and calls for research that investigate accounting and management control across organizational boundaries (Hopwood, 1996), researchers have expanded their view of the organization beyond traditional hierarchical perspectives to include network enterprises. A network enterprise is comprised of members with independent goals that cooperate and coordinate in an effort to share costs, gain access to resources, and affect a common end (Mouritsen and Thrane, 2006). Network enterprises vary in size and scope, from smaller consultancies that cooperate and coordinate to perform complex engagements, to a network of regulatory agencies that collectively set accounting and auditing standards. Social network analysis expands the scholar's view of the network enterprise as a collection of traditional organizations that band together to achieve a common goal by enabling a more granular examination of how position within, composition of, and links among the network participants influence organizational outcomes. Coordination and exchange within network enterprises often differs from traditional forms of governance. Rather than managing economic activities as a series of arms-length transactions, exchanges often occur within the fabric of embedded social relationships (Uzzi, 1997; Malone, 2004). Network governance suggests that economic exchanges are characterized by higher levels of trust, fine-grained information transfer between trading partners, and joint problem-solving arrangements (Jones et al.,
J. Worrell et al. / International Journal of Accounting Information Systems 14 (2013) 127–137
131
1997). Network enterprises utilize social controls (such as restricting access to the network, macroculture, collective sanctions, and reputational concerns) to safeguard against opportunism and malfeasance, instead of relying on contractual obligation, formal control mechanisms, or hierarchical structures (Jones et al., 1997; Uzzi, 1997). As a newer form of organizational governance, network enterprises represent a fertile field for accounting research. Social network analysis provides an alternative method with which to investigate the emergence of network enterprises and to describe their relations, coordination, and collaboration. For example, Richardson (2009) used social network analysis to examine the networks of regulatory agencies that promulgate accounting and audit standards, whereas Gulati and Gargiulo (1999) used social network analysis to explain the emergence of network enterprises based on existing ties, position within various networks, and prior experiences. For accounting researchers investigating network enterprise-related phenomena, social network analysis could prove fruitful in addressing the following questions: • In network enterprise arrangements, how effective are social controls at managing economic exchange when compared to traditional mechanisms used in arms-length transactions? • What are the economic consequences for organizations engaging in exchanges characterized by both traditional arms-length transactions as well as those conducted in the context of embedded ties? Is there an “optimal” mix or structure? • How does the strength of ties between trading partners in a network enterprise influence risktaking/risk-averse behavior? • For organizations that participate in network enterprises, is there an optimal level of embeddedness to guard against the risk of over-reliance on/excessive loyalty to network members? 3.2. Research where accounting information systems are nodes Traditionally, actors in a network have been viewed as human entities (individuals, groups, organizations). However, this need not always be the case. Networks may include nonhuman actors such as software, hardware, information systems and infrastructure standards (Walsham, 1997). When viewed in this light, networks of human and nonhuman elements represent stable social structures comprised of actors whose interests have been aligned. The stability of the network and any variability in outcomes is theorized to be a result of how goals are translated, resources are enrolled, and irreversibility is established in proposed courses of action (Callon, 1991; Latour, 1996; Walsham, 1997). Traditionally, qualitative methods such as case studies have been used to investigate and explain the relationships between human and non-human actors in a network (see Bloomfield et al., 1992; Bloomfield and Best, 1992; Preston et al., 1992; Walsham and Sahay, 1999). Scholars are now beginning to apply social network analysis to examine the information system as a node in a network. For example, Kane and Alavi (2008) use social network analysis to investigate how human actors and information systems interact as equals in a multi-modal network. Their findings suggest that the centrality of information systems positively impacts the quality and efficiency of organizational outcomes. While using social network analysis in this manner is in the early stages of application, it certainly holds promise when paired with research on interorganizational systems (IOS) that link trading partners. For AIS researchers who subscribe to the view that information systems (such as accounting systems) represent significant actors within a socially constructed network, social network analysis provides an additional tool in investigating the effects of information systems on organizational outcomes and performance by examining the manner in which information systems interact with other nodes in a network. This approach moves information systems beyond simply being viewed and studied as electronic ties between nodes in a network, but rather as autonomous and powerful nodes within a network. Given this view of information systems as nodes, studies aimed at addressing the following research questions could contribute to the AIS literature: • What is the optimal nature and strength of ties between information systems linking nodes to facilitate the flow of information or to foster innovation?
132
J. Worrell et al. / International Journal of Accounting Information Systems 14 (2013) 127–137
• How does the position of the information system in the network affect competitiveness and strategic advantage? • How can information systems that link nodes help manage risks associated with transactions, improve innovation, and increase collaboration? • Can we predict changes in the usage of transaction processing systems and related internal controls that have occurred, signaling adjustments to business processes and/or practices which have not yet been identified nor documented by either the firm's internal or external auditors? 3.3. Research on connections/relations between human actors By far, the most common use of social network analysis in the accounting literature has been as a tool to identify the relationships among and between individuals and groups. Social network analysis has been used to investigate Board memberships across organizations (Conyon and Muldoon, 2006), accounting scholars' connections and productivity (Wakefield, 2008), composition and connections of advice networks (McDonald and Westphal, 2003), socialization of new staff auditors (Morrison, 2002), and the role of accounting and financial experts in organizational decision making under conditions of uncertainty (Chapman, 1998; Masquefa, 2008). While scholarly inquiry in this vein has certainly been the most prolific, future studies could inform the following questions: • How can communication and coordination structures within work units, divisions, firms and/or professional associations be more efficiently and effectively organized? • How effective are ties and relationships at encouraging/discouraging ethical business practices as compared to more formal and traditional means (codes of conduct, training, threat of sanctions, etc.)? • How can accounting and audit personnel better position themselves within their organizations to increase their effectiveness, perceived value, access to resources, and/or access to critical information? 4. Social network analysis to predict, detect and prevent fraud Recently, scholars have turned to social network analysis as a tool to investigate one of the oldest accounting crimes in human history: fraud. The intersection of social network analysis' relative maturity, electronic data capture, and public interest created a perfect storm of sorts with the widely publicized Enron scandal. The e-mail records subpoenaed by the Department of Justice, and subsequently released as the “Enron corpus”, have allowed researchers to investigate the pattern of communication and coordination among the various actors in this fraud. Early efforts have proven insightful as they suggest how communication and coordination patterns changed as the fraud progressed (Murshed et al., 2007). While early work using social network analysis to investigate fraud has been largely exploratory in nature, there is the potential to combine communication and exchange theories with social network analysis to investigate fraud in ways previously unexplored. Rather than simply examining the relationships between members of a network, it may be possible to examine how fraud is facilitated by different types of ties among members of a social network and to identify characteristics of the social structures within those networks that may have an influence on mitigating or prohibiting opportunities for fraud. AIS and audit researchers can look to several applicable theories from the social psychology and criminology domains to better understand how frauds are conceived and executed through the social networks that exist within an organization, or at the boundary of the organization between employees and clients. For instance, social capital theory (Nahapiet and Ghoshal, 1998) has been applied to reveal a negative relationship between social capital and criminal activities within a community (Katz, 2002). Sampson and Laub (1993), Katz (1999), and Katz (2000) demonstrate that as social capital increases within a community, not only is desistance from crime more likely, but the capacity for empathy among community members also increases. Building upon this understanding and applying it to the context of criminal networks may allow us to understand how community members engage in communication and other activities that enhance the social capital available within the community and, subsequently, establish obstacles for behaviors that encourage fraud opportunities, or are otherwise harmful to the community.
J. Worrell et al. / International Journal of Accounting Information Systems 14 (2013) 127–137
133
Perhaps an even more intriguing theoretical lens from which to examine a fraud network is provided by social disorganization theory, which attributes criminal behavior to the breakdown of communal relationships that traditionally foster mutual benefits among community members. First established by Shaw and McKay (1969), social disorganization theory suggests that as interactions among members of a community become less frequent, less structured, and less beneficial to community members, social controls which serve to influence member actions are weakened, thereby resulting in opportunities for negligent or criminal behaviors. In terms of its direct application to fraud activities, social disorganization theory suggests that social networks that are limited in terms of strong ties between members are more likely to be infested with crime. Members of this weak tie network, if given the opportunity and pressure, are positioned to exploit the lack of social controls present within the network and rationalize their behaviors accordingly. See Fig. 2 for an illustration of this relationship. Another potentially rewarding application of social disorganization theory to the examination of fraudrelated criminal networks concerns the identification and understanding of collusion structures among social network members. Social exchange theory posits that all exchanges, including social exchanges, are a subjective cost–benefit analysis comparing alternatives (Cook, 1991). The main assumptions underlying why individuals engage in social exchange include anticipated reciprocity and expected gain in reputation, influence, or other rewards. While dense network structures facilitate the flow of information about the reputations and actions of actors, these strong network ties are also high in expectations of obligation and reciprocity, possibly including obligations to engage in and/or support criminal activity. Although only indirectly supported by evidence from the Enron email corpus (Murshed et al., 2007), the potential for a curvilinear relationship among social activity and criminal activity structures exists, whereby a lack of social ties results in individual criminal activities within a network, but an over-abundance of social ties leads to collusion among the criminally-inclined. See Fig. 3 for an illustration of this potential relationship. Exploratory research in this area may prove beneficial. While the extant literature involving social disorganization theory is primarily concentrated on explaining criminal behaviors within neighborhoods and social networks that occur in physical spaces, we believe that the application of this theory to fraud investigation and prevention within digital social networks is justifiable. An area ripe for future research is the investigation of fraud and crime in digital social networks, such as Facebook, Twitter and other online communities, and how tie strength differs between electronic and physical community relationships. Another area of research that social network
Fig. 2. Social controls and fraud as explained by social disorganization theory.
134
J. Worrell et al. / International Journal of Accounting Information Systems 14 (2013) 127–137
Fig. 3. Social controls and criminal activity structures.
analysis invites is the inclusion of characteristics shared among actors in a dyad, not simply the pattern of transactions, to help identify fraud. For instance, using electronic trace data from accounting information systems, social network analysis can uncover areas of high density transactions, low density transactions and transactions among nodes that are closely tied versus sparsely connected or share some common attributes that may have the potential to identify patterns of fraudulent activities. 5. Conclusion In Section 3, we provided an overview of current accounting research utilizing social network analysis. From this examination, three distinct literature streams emerged: transactions and control mechanisms in the context of inter-organizational dynamics, networks where accounting systems are viewed as nodes, and connections between human actors in an accounting context. From these streams, we presented a series of research questions which could serve to expand our knowledge in these areas and which social network analysis is uniquely positioned to inform in the near-term. In Section 4 we discussed one of these examples in depth, to provide deeper insights into how social network analysis could be applied in the context of fraud detection and investigation. While social network analysis offers opportunities to expand and contribute to the research streams mentioned above, we would be remiss if we didn't expand our discussion of how social network analysis can be leveraged to inform other areas of interest to AIS researchers. To this end, we looked to the topics typically addressed in the International Journal of Accounting Information Systems for areas where social network analysis might provide distinct benefits over other methods and foster a better understanding of these. Although there are numerous topics where we feel this method might prove fruitful, we focused on three that we feel have the most promise. 5.1. Control and auditability of information systems Since the Sarbanes Oxley Act of 2002, accounting researchers have focused anew on investigating issues around control and accountability of information systems. Of particular interest has been evaluating the nature and effectiveness of the control environment (Bowen et al., 2007; Klamm and Watson, 2009; Bart and Turel, 2010). The control environment represents the foundation for an organization's system of
J. Worrell et al. / International Journal of Accounting Information Systems 14 (2013) 127–137
135
internal controls and includes management's ethical values and integrity, organizational structure, and authority and reporting arrangements (COSO, 1992). Social network analysis could prove useful for research aimed at better understanding the nature and effectiveness of the control environment as well as the overall system of internal controls. For example, it could be used • as a method for examining communication and coordination between key stakeholders important to the identification and control of risk (such as auditors, members of the audit committee, risk owners, and management), • as a method for examining the effectiveness of coordination mechanisms inscribed in organizational structures as well as informal coordination mechanisms, and • as a method for examining the effectiveness of traditional controls (such as segregation of duties, approvals, and restricting access) compared to social controls embedded in the fabric of relationships. 5.2. Electronic dissemination of accounting information Information from accounting and transaction processing systems is a necessary component of planning, budgeting and management control. For this information to be useful, it must be accessible to decision makers in a proper format and timely fashion. However, it is often difficult to identify which actors require specific information, especially in globally dispersed organizations or networked enterprises with diverse stakeholder groups. As we have previously discussed, actors can be individuals, work units, organizations, regulatory bodies or even other information systems. Social network analysis provides a useful methodological approach to examine the interconnections between diverse actors and how these linkages can inform questions around which actors need accounting information and in what format. For example, it could be used • as a method to examine flows of information between linkages in inter-organizational systems connecting networked enterprises, • to identify whether key information is kept confidential or is transferred beyond organizational boundaries, and • as a method to identify key sources of accounting information as well as consumers of this information. 5.3. Organizational/social perspectives on impact of technology on accounting Prior research has demonstrated that informal networks augment or supersede formal hierarchies with respect to knowledge sharing. An extension of research aimed at understanding the dissemination of accounting information would be to more fully explore where and to whom people look for advice and expertise. While Murthy and Taylor's (2009) work on examining knowledge sharing practices on the Accounting Education using Computers and Multimedia (AECM) email list takes a first step, further investigation into networks aimed at connecting expertise, such as the American Accounting Association's AAA Commons initiative, might serve to bridge the gap between formal networks and informal networks. Social network analysis provides a new approach to more fully explore how accounting and business process knowledge is dispersed across organizational or professional boundaries, as well as understand the manner in which it is accessed. For example, it could be used • as a method to identify the location of expertise within and outside traditional hierarchies and organizational structures, • as a method to identify informal expertise networks and provide insight on how people perform their work, • as a method to aid in the design of systems aimed at connecting people with expertise (such as AAA Commons), and • as a method to explain attributes of individuals and groups critical to internal control, dissemination and creation of accounting-related information, and policy and standards. To conclude, the purpose of this article was to present social network analysis as an alternative method that has high potential for expanding AIS research. Social network analysis focuses on the pattern of ties
136
J. Worrell et al. / International Journal of Accounting Information Systems 14 (2013) 127–137
connecting nodes within a network, and emphasizes that both the pattern of relationships and characteristics of the dyads influence the actions of network members. While most prior research using social network analysis has focused on network dynamics in face to face networks, advances in information and communication technologies, especially accounting information systems, open new possibilities for the types of network structures and research questions that can be investigated.
References Baker F, Hubert L. The analysis of social interaction data. Sociological Methods and Research 1981;9:339–931. Bart C, Turel O. IT and the board of directors: an empirical investigation into the “governance questions” Canadian board members ask about IT. J Inf Syst 2010;24:147. Bloomfield B, Best A. Management consultant systems development, power and the translation of problems. Sociol Rev 1992;40: 533–60. Bloomfield B, Coombs R, Cooper D, Rea D. Machines and manoeuvres: responsibility accounting and the construction of hospital information systems. Acc Manage Inf Technol 1992;2:197–219. Borgatti SP, Cross R. A relational view of information seeking and learning in social networks. Manage Sci 2003;49:432–45. Borgatti SP, Foster PC. The network paradigm in organizational research: a review and typology. J Manage 2003;29:991–1013. Bowen P, Cheung M-YD, Rohde F. Enhancing IT governance practices: a model and case study of an organization's efforts. Int J Acc Inf Syst 2007;8:191–221. Burt RS. Structural holes: the social structure of competition. Cambridge, MA: Harvard University Press; 1992. Callon M. Techno-economic networks and irreversibility. In: Law J, editor. A sociology of monsters: essays on power, technology and domination. London: Routledge; 1991. Chapman C. Accountants in organisational networks. Acc Organ Soc 1998;23:737–66. Coleman JS. Foundations of social theory. Cambridge, MA: Belknap Press of Harvard University Press; 1990. Conyon M, Muldoon M. The small world of corporate boards. J Bus Finance Acc 2006;33:1321–43. Cook KS. The microfoundations of social structure: an exchange perspective. In: Huber J, editor. Macro-micro lingakes in sociology. Newbury Park, CA: Sage; 1991. p. 219–34. COSO. Internal control-integrated framework. Jersey City, NJ: American Institute of Certified Public Accountants; 1992. Freeman LC. Centrality in social networks: conceptual clarification. Soc Netw 1979;1:215–39. Garton L, Haythornthwaite C, Wellman B. Studying on-line social networks. In: Jones S, editor. Doing internet research: critical issues and methods for examining the net. Thousand Oaks, CA: Sage; 1999. p. 75–105. Gulati R, Gargiulo M. Where do interorganizational networks come from? Am J Sociol 1999;104:1439–93. Hopwood AG. Looking across rather than up and down: on the need to explore the lateral processing of information. Acc Organ Soc 1996;21:589–90. Jones C, Hesterly W, Borgatti S. A general theory of network governance: exchange conditions and social mechanisms. Acad Manage Rev 1997;22:911–45. Kane G, Alavi M. Casting the net: a multimodal network perspective on user–system interactions. Inf Syst Res 2008;19:253–72. Katz R. Building the foundation for a side-by-side explanatory model: a general theory of crime, the age-graded life course theory and attachment theory. Western Criminology Review. 1:http://wcr.sonoma.edu/v1n2/katz.html1999. Katz R. Explaining girls' and women's crime and desistance in the context of their victimization experiences: a developmental test of revised strain theory and the life course perspective. Violence Against Women 2000;6:633–60. Katz R. Re-examining the integrative social capital theory of crime. West Criminol Rev 2002;4:30–54. Klamm B, Watson MW. SOX 404 reported internal control weaknesses: a test of COSO framework components and information technology. J Inf Syst 2009;23:1. Latour B. On actor-network theory: a few clarifications. Soziale Welt 1996;47:369–81. Malone T. The future of work: how the new order of business will shape your organization, your management style, and your life. Boston, MA: Harvard Business School Press; 2004. Masquefa B. Top management adoption of a locally driven performance measurement and evaluation system: a social network perspective. Manage Acc Res 2008;19:182–207. McDonald ML, Westphal JD. Getting by with the advice of their friends: CEOs' advice networks and firms' strategic responses to poor performance. Adm Sci Q 2003;48:1–32. Morrison E. Newcomers' relationships: the role of social network ties during socialization. Acad Manage J 2002;45:1149–60. Mouritsen J, Thrane S. Accounting, network complimentarities and the development of inter-organisational relations. Acc Organ Soc 2006;31:241–75. Murshed STH, Davis JG, Hossain L. Social network analysis and organizational disintegration: the case of Enron corporation. International Conference on Information Systems. Phoenix, AZ: Association for Information Systems; 2007. Murthy U, Taylor E. Knowledge sharing among accounting academics in an electronic network of practice. Acc Horiz 2009;23:151. Nahapiet J, Ghoshal S. Social capital, intellectual capital, and the organizational advantage. Acad Manage Rev 1998;23:242–66. Ostrom E. Governing the commons: the evolution of institutions for collective action. Cambridge: Cambridge University Press; 1990. Preston A, Cooper D, Coombs R. Fabricating budgets: a study of the production of management budgeting in the National Health Service. Acc Organ Soc 1992;17:571–94. Richardson A. Regulatory networks for accounting and auditing standards: a social network analysis of Canadian and international standard-setting. Acc Organ Soc 2009;34:571–88. Sampson RJ, Laub JH. Crime in the making pathways and turning points through life. Cambridge, Massachusetts: Harvard University Press; 1993. Scott J. Social network analysis. Newbury Park, CA: Sage Publications; 1991. Shaw CR, McKay HD. Juvenile delinquency and urban areas. Chicago, Illinois: University of Chicago Press; 1969.
J. Worrell et al. / International Journal of Accounting Information Systems 14 (2013) 127–137
137
Smatt C. An investigation of the impact of the structure and quality of relationships on knowledge exchange and individual performance. Management Information Systems. Tallahassee: Florida State University; 2009. p. 143. Uzzi B. Social structure and competition in interfirm networks: the paradox of embeddedness. Adm Sci Q 1997;42:35–67. Wakefield R. Networks of accounting research: a citation-based structural and network analysis. Br Acc Rev 2008;40:228–44. Walsham G. Actor-network theory and IS research: current status and future prospects. In: Lee A, Liebenau J, DeGross J, editors. Information systems and qualitative research. London: Chapman and Hall; 1997. Walsham G, Sahay S. GIS for district-level administration in India: problems and opportunities (n1). MIS Q 1999;23:39. Wasserman S, Faust K. Social network analysis. Cambridge: Cambridge University Press; 1994.
Contents lists available at SciVerse ScienceDirect
International Journal of Accounting Information Systems
Social network analysis in accounting information systems research James Worrell 1, Molly Wasko ⁎, Allen Johnston 2 University of Alabama at Birmingham, 319 Business and Engineering Complex, 1530 3rd Avenue South, Birmingham, AL 35294-4460, United States
a r t i c l e
i n f o
Article history: Received 3 August 2010 Received in revised form 10 May 2011 Accepted 16 June 2011 Keywords: Social network analysis Centrality Methods in accounting research Fraud
a b s t r a c t This paper introduces social network analysis as an alternative research method for conducting accounting information systems related research. With advances in information and communication technologies, transaction data are being recorded in electronic form, resulting in a variety of research opportunities to examine dyadic interactions. A network consists of a set of nodes connected by ties. Social network research focuses on how outcomes are influenced not just by the attributes of the nodes (e.g. individuals), but also by the ties connecting nodes to each other. The nodes are typically conceptualized as actors, such as individuals, teams, or organizations. A unique network structure is created to reflect each different type of tie, such as trust, advice, collocation, or organizational affiliation. Social network analysis can be used for research examining individual, dyadic or network levels of analyses, and is a powerful tool for conducting multi-method research. Given the vast amounts of trace electronic data collected via accounting information systems, this paper reviews how social network analysis not only opens new research avenues for accounting information systems researchers, but identifies opportunities for the field of accounting information systems to inform social network research by identifying new network structures and dynamics leveraging transactional data. © 2011 Elsevier Inc. All rights reserved.
1. Introduction There has been a general shift in management research over the past decade towards more relational theories of organizations that view actions and actors not as independent, autonomous agents, but as embedded within ⁎ Corresponding author. Tel.: + 1 205 934 8806(office); fax: + 1 205 975 4429. E-mail addresses: [email protected] (J. Worrell), [email protected] (M. Wasko), [email protected] (A. Johnston). 1 Tel.: + 1 205 934 8820. 2 Tel.: + 1 205 934 8870. 1467-0895/$ – see front matter © 2011 Elsevier Inc. All rights reserved. doi:10.1016/j.accinf.2011.06.002
128
J. Worrell et al. / International Journal of Accounting Information Systems 14 (2013) 127–137
socio-technical systems. In contrast to theories that examine individuals based on their attributes, such as gender, age, education, or occupation, social network perspectives focus on how the relationships between entities, such as individuals, functional units, or organizations, influence interactions and outcomes. The concept of a “network” is broad and can be applied to a variety of phenomena where a set of relations is ascribed to an identified set of actors. What unites social network perspectives is the focus on the patterns and implications of the ties within a collective (Scott, 1991; Wasserman and Faust, 1994). For example, at the individual level ties facilitate the spread of information among network participants, enable the flow of both tangible and intangible resources among network members, and place constraints on each member's behavior (Burt, 1992). Social network research focuses on the significance of relationships as essential for understanding social action, but varies widely in the attributes that are studied. A network is defined as a set of nodes connected by ties. Nodes are typically “actors”, and can be people, teams, organizations or information systems. Relations, or ties, connect the actors and can vary in content, direction, and relational strength, all of which influence the dynamics of the network (Garton et al., 1999). The content of ties refers to the resource exchanged or common bond, such as information, money, advice, or kinship. The direction of ties indicates an “ego” who gives the resource, and the “alter” who receives it, although ties in some networks are undirected, such as a shared attribute (e.g. gender), or joint membership on a team. The relational strength of ties pertains to the level of activity, such as quantity of communications, or the intensity, such as the social influence exerted by the tie, indicating that ties can be valued or weighted. For instance, the relational strength of ties could indicate the amount of energy, emotional intensity, intimacy, commitment or trust connecting the actors. Relational ties are often studied in management research as important aspects of social influence that can exert control, such as social punishments or ostracism (Ostrom, 1990). Other aspects of social influence foster cohesion and prosocial behavior in the network, such as trust, identification, the diffusion of information and commitment (Coleman, 1990; Nahapiet and Ghoshal, 1998). Each tie defines a different network, and while some ties are often related (a trust network is often correlated to a friendship network), ties are often assumed to function differently. Not all ties are considered to have positive outcomes; for instance, network research is often used to map the flow of disease or terrorist networks. Therefore, some network research focuses on how to improve the flow of the resource through the network, such as adoption of a new accounting information system, or how to disrupt the flow of resources in the network, such as taking out key nodes in a fraud network. Depending upon the theory being applied, some studies examine network variables as independent variables causing consequences, such as adoption of a technology or improved performance, while other studies examine network variables as dependent variables, identifying the causes underlying the pattern of network connections, how networks come to be, and how networks change over time (Borgatti and Foster, 2003).
2. Example of social network analysis This section describes a social network study to provide insights about how to apply social network analysis. The context for the study was a business school consisting of 89 faculty members organized into 7 departments. The purpose of the study was to understand how social networks impact the research performance of individual faculty, and the practical question was how collaboration among faculty members within a business school impacts an individual faculty's research performance (Smatt, 2009). The data for this study were collected via survey using a roster method with the names of all faculty members in the school listed. Each faculty member responded to survey items based upon their relationships with all other faculty members. Social network data can also be collected from archival data, such as using electronic trace data from transaction processing systems, electronic data interchange (EDI) or email communications. Data are recorded in a square matrix to indicate the ego (in the first column), the alter (in the first row) and the tie (weighted, directed and/or dichotomous) in the corresponding ego/alter cells. A separate network is created for each type of tie; for instance, in this study multiple networks were collected, including collocation, same department, hierarchical level (assistant, associate, full, and administrator), how well faculty members knew each other, trusted each other and sought advice from each other.
J. Worrell et al. / International Journal of Accounting Information Systems 14 (2013) 127–137
129
Faculty members were first given the roster of names and asked to indicate “who do you know well”. The data were recorded in a square social network matrix, and used to create the social network graph depicted in Fig. 1, which will be used to illustrate the concept of network centrality. One of the most common variables assessed in network research is centrality, or an actor's position in the overall network of relations (Scott, 1991; Wasserman and Faust, 1994). Freeman (1979) proposed three separate measures of centrality: betweenness, closeness, and degree centrality. Betweenness centrality refers to the extent to which an ego falls between two other actors in the network, revealing the potential point of control for resource flows. Closeness centrality is the extent to which an ego can easily connect with all other actors, or the amount of direct or indirect links that an actor has with others in the network. The measure of closeness reveals the efficiency of the network, indicating how quickly an actor can gain access to resources. Degree centrality refers to the amount of direct links an ego has with others in the network, uncovering how active/connected an ego is within the network, assessed by summing the total number of ties that an ego has with alters (Freeman, 1979). In a directed network, indegree centrality represents the number of alters selecting an ego (ties coming in) and outdegree centrality represents the number alters selected by an ego (ties going out). Fig. 1 is a graphical representation of the faculty network. The tie indicates that the ego faculty knew the alter faculty well, node size is based on indegree centrality (how many alters selected that ego), and the node color indicates membership in the same department. The different types of centralities are identified, demonstrating that even though an ego may have a large indegree centrality (indicating that the faculty is well known), that does not necessarily indicate high closeness or betweenness. For instance, the largest nodes (most well known faculty members) are generally only well known by others within their departments, not across departments. The gray circled area to the right is the accounting department, indicating that faculty members in this department have fewer ties among each other and among faculty members in other departments. It is common to see network data analyzed at the individual level and dyadic level. For example, at the individual level, centrality scores can be assessed for each individual and then combined with other data, such as education and experience, and used in regression analyses to predict performance at work. At the dyadic level, quadratic assignment procedure (QAP) and multiple regression-quadratic assignment procedure (MRQAP) can be used to predict whether one type of dyadic tie predicts another dyadic tie (for
Fig. 1. Network of business school faculty.
130
J. Worrell et al. / International Journal of Accounting Information Systems 14 (2013) 127–137
instance, the extent to which ego trusts alter is influenced by the extent to which ego is collocated with alter) (Baker and Hubert, 1981; Borgatti and Cross, 2003). Network analysis is also useful to identify characteristics of the network as a whole, such as to identify cliques, core/periphery structures, network density and/or centralization. For the faculty study depicted in Fig. 1, indegree centrality in the research advice network was used in structural equation modeling (SEM) to predict research productivity. The results indicated that centrality in the research advice network had a significant, positive impact on research productivity above and beyond tenure, department and other individual-level variables (Network centrality was the only independent variable used to predict research performance as assessed by number of top tier publications. The human capital variables were used as controls for alternative explanations.). This has important implications for encouraging faculty collaboration, especially among the accounting faculty in this study who collectively were less central in the overall research advice network. The purpose of this section was to provide an overview of social network analysis. For researchers wanting more detailed information about how to perform social network analysis, and definitions of the different types of network measures (such as centrality, density and core-periphery), we encourage them to review Wasserman and Faust's (1994) Social Network Analysis, which is widely considered the seminal text in this area. Another good starting reference is Scott's (1991) Social Network Analysis. The leading academic organization related to social network research is INSNA: the International Network for Social Network Analysis (www.insna.org). For individuals interested in learning more about the computer programs used to analyze social network data, INSNA sponsors an annual conference, called the Sunbelt Conference that offers professional development workshops from leading experts in the field (http://www. insna.org/sunbelt/). 3. Social network analysis in AIS research While a rich stream of research utilizing social network analysis exists within the sociology, anthropology, management and information systems disciplines, accounting researchers have been slow to add this method to their toolkits. Traditionally, social network analysis research has investigated networks in three manners: network connections which facilitate flows of resources between nodes, ties between nodes that influence behavior and enforce social norms, and networks themselves as either independent or dependent variables (Borgatti and Foster, 2003). In the following section, we provide a brief overview of accounting information systems (AIS) research to date that has used social network analysis. This is followed by potential research questions where social network analysis might inform existing AIS phenomena. 3.1. Research on network enterprises Accounting researchers have become increasingly interested in examining the dynamics of inter-organizational relations. In response to changing organizational dynamics and calls for research that investigate accounting and management control across organizational boundaries (Hopwood, 1996), researchers have expanded their view of the organization beyond traditional hierarchical perspectives to include network enterprises. A network enterprise is comprised of members with independent goals that cooperate and coordinate in an effort to share costs, gain access to resources, and affect a common end (Mouritsen and Thrane, 2006). Network enterprises vary in size and scope, from smaller consultancies that cooperate and coordinate to perform complex engagements, to a network of regulatory agencies that collectively set accounting and auditing standards. Social network analysis expands the scholar's view of the network enterprise as a collection of traditional organizations that band together to achieve a common goal by enabling a more granular examination of how position within, composition of, and links among the network participants influence organizational outcomes. Coordination and exchange within network enterprises often differs from traditional forms of governance. Rather than managing economic activities as a series of arms-length transactions, exchanges often occur within the fabric of embedded social relationships (Uzzi, 1997; Malone, 2004). Network governance suggests that economic exchanges are characterized by higher levels of trust, fine-grained information transfer between trading partners, and joint problem-solving arrangements (Jones et al.,
J. Worrell et al. / International Journal of Accounting Information Systems 14 (2013) 127–137
131
1997). Network enterprises utilize social controls (such as restricting access to the network, macroculture, collective sanctions, and reputational concerns) to safeguard against opportunism and malfeasance, instead of relying on contractual obligation, formal control mechanisms, or hierarchical structures (Jones et al., 1997; Uzzi, 1997). As a newer form of organizational governance, network enterprises represent a fertile field for accounting research. Social network analysis provides an alternative method with which to investigate the emergence of network enterprises and to describe their relations, coordination, and collaboration. For example, Richardson (2009) used social network analysis to examine the networks of regulatory agencies that promulgate accounting and audit standards, whereas Gulati and Gargiulo (1999) used social network analysis to explain the emergence of network enterprises based on existing ties, position within various networks, and prior experiences. For accounting researchers investigating network enterprise-related phenomena, social network analysis could prove fruitful in addressing the following questions: • In network enterprise arrangements, how effective are social controls at managing economic exchange when compared to traditional mechanisms used in arms-length transactions? • What are the economic consequences for organizations engaging in exchanges characterized by both traditional arms-length transactions as well as those conducted in the context of embedded ties? Is there an “optimal” mix or structure? • How does the strength of ties between trading partners in a network enterprise influence risktaking/risk-averse behavior? • For organizations that participate in network enterprises, is there an optimal level of embeddedness to guard against the risk of over-reliance on/excessive loyalty to network members? 3.2. Research where accounting information systems are nodes Traditionally, actors in a network have been viewed as human entities (individuals, groups, organizations). However, this need not always be the case. Networks may include nonhuman actors such as software, hardware, information systems and infrastructure standards (Walsham, 1997). When viewed in this light, networks of human and nonhuman elements represent stable social structures comprised of actors whose interests have been aligned. The stability of the network and any variability in outcomes is theorized to be a result of how goals are translated, resources are enrolled, and irreversibility is established in proposed courses of action (Callon, 1991; Latour, 1996; Walsham, 1997). Traditionally, qualitative methods such as case studies have been used to investigate and explain the relationships between human and non-human actors in a network (see Bloomfield et al., 1992; Bloomfield and Best, 1992; Preston et al., 1992; Walsham and Sahay, 1999). Scholars are now beginning to apply social network analysis to examine the information system as a node in a network. For example, Kane and Alavi (2008) use social network analysis to investigate how human actors and information systems interact as equals in a multi-modal network. Their findings suggest that the centrality of information systems positively impacts the quality and efficiency of organizational outcomes. While using social network analysis in this manner is in the early stages of application, it certainly holds promise when paired with research on interorganizational systems (IOS) that link trading partners. For AIS researchers who subscribe to the view that information systems (such as accounting systems) represent significant actors within a socially constructed network, social network analysis provides an additional tool in investigating the effects of information systems on organizational outcomes and performance by examining the manner in which information systems interact with other nodes in a network. This approach moves information systems beyond simply being viewed and studied as electronic ties between nodes in a network, but rather as autonomous and powerful nodes within a network. Given this view of information systems as nodes, studies aimed at addressing the following research questions could contribute to the AIS literature: • What is the optimal nature and strength of ties between information systems linking nodes to facilitate the flow of information or to foster innovation?
132
J. Worrell et al. / International Journal of Accounting Information Systems 14 (2013) 127–137
• How does the position of the information system in the network affect competitiveness and strategic advantage? • How can information systems that link nodes help manage risks associated with transactions, improve innovation, and increase collaboration? • Can we predict changes in the usage of transaction processing systems and related internal controls that have occurred, signaling adjustments to business processes and/or practices which have not yet been identified nor documented by either the firm's internal or external auditors? 3.3. Research on connections/relations between human actors By far, the most common use of social network analysis in the accounting literature has been as a tool to identify the relationships among and between individuals and groups. Social network analysis has been used to investigate Board memberships across organizations (Conyon and Muldoon, 2006), accounting scholars' connections and productivity (Wakefield, 2008), composition and connections of advice networks (McDonald and Westphal, 2003), socialization of new staff auditors (Morrison, 2002), and the role of accounting and financial experts in organizational decision making under conditions of uncertainty (Chapman, 1998; Masquefa, 2008). While scholarly inquiry in this vein has certainly been the most prolific, future studies could inform the following questions: • How can communication and coordination structures within work units, divisions, firms and/or professional associations be more efficiently and effectively organized? • How effective are ties and relationships at encouraging/discouraging ethical business practices as compared to more formal and traditional means (codes of conduct, training, threat of sanctions, etc.)? • How can accounting and audit personnel better position themselves within their organizations to increase their effectiveness, perceived value, access to resources, and/or access to critical information? 4. Social network analysis to predict, detect and prevent fraud Recently, scholars have turned to social network analysis as a tool to investigate one of the oldest accounting crimes in human history: fraud. The intersection of social network analysis' relative maturity, electronic data capture, and public interest created a perfect storm of sorts with the widely publicized Enron scandal. The e-mail records subpoenaed by the Department of Justice, and subsequently released as the “Enron corpus”, have allowed researchers to investigate the pattern of communication and coordination among the various actors in this fraud. Early efforts have proven insightful as they suggest how communication and coordination patterns changed as the fraud progressed (Murshed et al., 2007). While early work using social network analysis to investigate fraud has been largely exploratory in nature, there is the potential to combine communication and exchange theories with social network analysis to investigate fraud in ways previously unexplored. Rather than simply examining the relationships between members of a network, it may be possible to examine how fraud is facilitated by different types of ties among members of a social network and to identify characteristics of the social structures within those networks that may have an influence on mitigating or prohibiting opportunities for fraud. AIS and audit researchers can look to several applicable theories from the social psychology and criminology domains to better understand how frauds are conceived and executed through the social networks that exist within an organization, or at the boundary of the organization between employees and clients. For instance, social capital theory (Nahapiet and Ghoshal, 1998) has been applied to reveal a negative relationship between social capital and criminal activities within a community (Katz, 2002). Sampson and Laub (1993), Katz (1999), and Katz (2000) demonstrate that as social capital increases within a community, not only is desistance from crime more likely, but the capacity for empathy among community members also increases. Building upon this understanding and applying it to the context of criminal networks may allow us to understand how community members engage in communication and other activities that enhance the social capital available within the community and, subsequently, establish obstacles for behaviors that encourage fraud opportunities, or are otherwise harmful to the community.
J. Worrell et al. / International Journal of Accounting Information Systems 14 (2013) 127–137
133
Perhaps an even more intriguing theoretical lens from which to examine a fraud network is provided by social disorganization theory, which attributes criminal behavior to the breakdown of communal relationships that traditionally foster mutual benefits among community members. First established by Shaw and McKay (1969), social disorganization theory suggests that as interactions among members of a community become less frequent, less structured, and less beneficial to community members, social controls which serve to influence member actions are weakened, thereby resulting in opportunities for negligent or criminal behaviors. In terms of its direct application to fraud activities, social disorganization theory suggests that social networks that are limited in terms of strong ties between members are more likely to be infested with crime. Members of this weak tie network, if given the opportunity and pressure, are positioned to exploit the lack of social controls present within the network and rationalize their behaviors accordingly. See Fig. 2 for an illustration of this relationship. Another potentially rewarding application of social disorganization theory to the examination of fraudrelated criminal networks concerns the identification and understanding of collusion structures among social network members. Social exchange theory posits that all exchanges, including social exchanges, are a subjective cost–benefit analysis comparing alternatives (Cook, 1991). The main assumptions underlying why individuals engage in social exchange include anticipated reciprocity and expected gain in reputation, influence, or other rewards. While dense network structures facilitate the flow of information about the reputations and actions of actors, these strong network ties are also high in expectations of obligation and reciprocity, possibly including obligations to engage in and/or support criminal activity. Although only indirectly supported by evidence from the Enron email corpus (Murshed et al., 2007), the potential for a curvilinear relationship among social activity and criminal activity structures exists, whereby a lack of social ties results in individual criminal activities within a network, but an over-abundance of social ties leads to collusion among the criminally-inclined. See Fig. 3 for an illustration of this potential relationship. Exploratory research in this area may prove beneficial. While the extant literature involving social disorganization theory is primarily concentrated on explaining criminal behaviors within neighborhoods and social networks that occur in physical spaces, we believe that the application of this theory to fraud investigation and prevention within digital social networks is justifiable. An area ripe for future research is the investigation of fraud and crime in digital social networks, such as Facebook, Twitter and other online communities, and how tie strength differs between electronic and physical community relationships. Another area of research that social network
Fig. 2. Social controls and fraud as explained by social disorganization theory.
134
J. Worrell et al. / International Journal of Accounting Information Systems 14 (2013) 127–137
Fig. 3. Social controls and criminal activity structures.
analysis invites is the inclusion of characteristics shared among actors in a dyad, not simply the pattern of transactions, to help identify fraud. For instance, using electronic trace data from accounting information systems, social network analysis can uncover areas of high density transactions, low density transactions and transactions among nodes that are closely tied versus sparsely connected or share some common attributes that may have the potential to identify patterns of fraudulent activities. 5. Conclusion In Section 3, we provided an overview of current accounting research utilizing social network analysis. From this examination, three distinct literature streams emerged: transactions and control mechanisms in the context of inter-organizational dynamics, networks where accounting systems are viewed as nodes, and connections between human actors in an accounting context. From these streams, we presented a series of research questions which could serve to expand our knowledge in these areas and which social network analysis is uniquely positioned to inform in the near-term. In Section 4 we discussed one of these examples in depth, to provide deeper insights into how social network analysis could be applied in the context of fraud detection and investigation. While social network analysis offers opportunities to expand and contribute to the research streams mentioned above, we would be remiss if we didn't expand our discussion of how social network analysis can be leveraged to inform other areas of interest to AIS researchers. To this end, we looked to the topics typically addressed in the International Journal of Accounting Information Systems for areas where social network analysis might provide distinct benefits over other methods and foster a better understanding of these. Although there are numerous topics where we feel this method might prove fruitful, we focused on three that we feel have the most promise. 5.1. Control and auditability of information systems Since the Sarbanes Oxley Act of 2002, accounting researchers have focused anew on investigating issues around control and accountability of information systems. Of particular interest has been evaluating the nature and effectiveness of the control environment (Bowen et al., 2007; Klamm and Watson, 2009; Bart and Turel, 2010). The control environment represents the foundation for an organization's system of
J. Worrell et al. / International Journal of Accounting Information Systems 14 (2013) 127–137
135
internal controls and includes management's ethical values and integrity, organizational structure, and authority and reporting arrangements (COSO, 1992). Social network analysis could prove useful for research aimed at better understanding the nature and effectiveness of the control environment as well as the overall system of internal controls. For example, it could be used • as a method for examining communication and coordination between key stakeholders important to the identification and control of risk (such as auditors, members of the audit committee, risk owners, and management), • as a method for examining the effectiveness of coordination mechanisms inscribed in organizational structures as well as informal coordination mechanisms, and • as a method for examining the effectiveness of traditional controls (such as segregation of duties, approvals, and restricting access) compared to social controls embedded in the fabric of relationships. 5.2. Electronic dissemination of accounting information Information from accounting and transaction processing systems is a necessary component of planning, budgeting and management control. For this information to be useful, it must be accessible to decision makers in a proper format and timely fashion. However, it is often difficult to identify which actors require specific information, especially in globally dispersed organizations or networked enterprises with diverse stakeholder groups. As we have previously discussed, actors can be individuals, work units, organizations, regulatory bodies or even other information systems. Social network analysis provides a useful methodological approach to examine the interconnections between diverse actors and how these linkages can inform questions around which actors need accounting information and in what format. For example, it could be used • as a method to examine flows of information between linkages in inter-organizational systems connecting networked enterprises, • to identify whether key information is kept confidential or is transferred beyond organizational boundaries, and • as a method to identify key sources of accounting information as well as consumers of this information. 5.3. Organizational/social perspectives on impact of technology on accounting Prior research has demonstrated that informal networks augment or supersede formal hierarchies with respect to knowledge sharing. An extension of research aimed at understanding the dissemination of accounting information would be to more fully explore where and to whom people look for advice and expertise. While Murthy and Taylor's (2009) work on examining knowledge sharing practices on the Accounting Education using Computers and Multimedia (AECM) email list takes a first step, further investigation into networks aimed at connecting expertise, such as the American Accounting Association's AAA Commons initiative, might serve to bridge the gap between formal networks and informal networks. Social network analysis provides a new approach to more fully explore how accounting and business process knowledge is dispersed across organizational or professional boundaries, as well as understand the manner in which it is accessed. For example, it could be used • as a method to identify the location of expertise within and outside traditional hierarchies and organizational structures, • as a method to identify informal expertise networks and provide insight on how people perform their work, • as a method to aid in the design of systems aimed at connecting people with expertise (such as AAA Commons), and • as a method to explain attributes of individuals and groups critical to internal control, dissemination and creation of accounting-related information, and policy and standards. To conclude, the purpose of this article was to present social network analysis as an alternative method that has high potential for expanding AIS research. Social network analysis focuses on the pattern of ties
136
J. Worrell et al. / International Journal of Accounting Information Systems 14 (2013) 127–137
connecting nodes within a network, and emphasizes that both the pattern of relationships and characteristics of the dyads influence the actions of network members. While most prior research using social network analysis has focused on network dynamics in face to face networks, advances in information and communication technologies, especially accounting information systems, open new possibilities for the types of network structures and research questions that can be investigated.
References Baker F, Hubert L. The analysis of social interaction data. Sociological Methods and Research 1981;9:339–931. Bart C, Turel O. IT and the board of directors: an empirical investigation into the “governance questions” Canadian board members ask about IT. J Inf Syst 2010;24:147. Bloomfield B, Best A. Management consultant systems development, power and the translation of problems. Sociol Rev 1992;40: 533–60. Bloomfield B, Coombs R, Cooper D, Rea D. Machines and manoeuvres: responsibility accounting and the construction of hospital information systems. Acc Manage Inf Technol 1992;2:197–219. Borgatti SP, Cross R. A relational view of information seeking and learning in social networks. Manage Sci 2003;49:432–45. Borgatti SP, Foster PC. The network paradigm in organizational research: a review and typology. J Manage 2003;29:991–1013. Bowen P, Cheung M-YD, Rohde F. Enhancing IT governance practices: a model and case study of an organization's efforts. Int J Acc Inf Syst 2007;8:191–221. Burt RS. Structural holes: the social structure of competition. Cambridge, MA: Harvard University Press; 1992. Callon M. Techno-economic networks and irreversibility. In: Law J, editor. A sociology of monsters: essays on power, technology and domination. London: Routledge; 1991. Chapman C. Accountants in organisational networks. Acc Organ Soc 1998;23:737–66. Coleman JS. Foundations of social theory. Cambridge, MA: Belknap Press of Harvard University Press; 1990. Conyon M, Muldoon M. The small world of corporate boards. J Bus Finance Acc 2006;33:1321–43. Cook KS. The microfoundations of social structure: an exchange perspective. In: Huber J, editor. Macro-micro lingakes in sociology. Newbury Park, CA: Sage; 1991. p. 219–34. COSO. Internal control-integrated framework. Jersey City, NJ: American Institute of Certified Public Accountants; 1992. Freeman LC. Centrality in social networks: conceptual clarification. Soc Netw 1979;1:215–39. Garton L, Haythornthwaite C, Wellman B. Studying on-line social networks. In: Jones S, editor. Doing internet research: critical issues and methods for examining the net. Thousand Oaks, CA: Sage; 1999. p. 75–105. Gulati R, Gargiulo M. Where do interorganizational networks come from? Am J Sociol 1999;104:1439–93. Hopwood AG. Looking across rather than up and down: on the need to explore the lateral processing of information. Acc Organ Soc 1996;21:589–90. Jones C, Hesterly W, Borgatti S. A general theory of network governance: exchange conditions and social mechanisms. Acad Manage Rev 1997;22:911–45. Kane G, Alavi M. Casting the net: a multimodal network perspective on user–system interactions. Inf Syst Res 2008;19:253–72. Katz R. Building the foundation for a side-by-side explanatory model: a general theory of crime, the age-graded life course theory and attachment theory. Western Criminology Review. 1:http://wcr.sonoma.edu/v1n2/katz.html1999. Katz R. Explaining girls' and women's crime and desistance in the context of their victimization experiences: a developmental test of revised strain theory and the life course perspective. Violence Against Women 2000;6:633–60. Katz R. Re-examining the integrative social capital theory of crime. West Criminol Rev 2002;4:30–54. Klamm B, Watson MW. SOX 404 reported internal control weaknesses: a test of COSO framework components and information technology. J Inf Syst 2009;23:1. Latour B. On actor-network theory: a few clarifications. Soziale Welt 1996;47:369–81. Malone T. The future of work: how the new order of business will shape your organization, your management style, and your life. Boston, MA: Harvard Business School Press; 2004. Masquefa B. Top management adoption of a locally driven performance measurement and evaluation system: a social network perspective. Manage Acc Res 2008;19:182–207. McDonald ML, Westphal JD. Getting by with the advice of their friends: CEOs' advice networks and firms' strategic responses to poor performance. Adm Sci Q 2003;48:1–32. Morrison E. Newcomers' relationships: the role of social network ties during socialization. Acad Manage J 2002;45:1149–60. Mouritsen J, Thrane S. Accounting, network complimentarities and the development of inter-organisational relations. Acc Organ Soc 2006;31:241–75. Murshed STH, Davis JG, Hossain L. Social network analysis and organizational disintegration: the case of Enron corporation. International Conference on Information Systems. Phoenix, AZ: Association for Information Systems; 2007. Murthy U, Taylor E. Knowledge sharing among accounting academics in an electronic network of practice. Acc Horiz 2009;23:151. Nahapiet J, Ghoshal S. Social capital, intellectual capital, and the organizational advantage. Acad Manage Rev 1998;23:242–66. Ostrom E. Governing the commons: the evolution of institutions for collective action. Cambridge: Cambridge University Press; 1990. Preston A, Cooper D, Coombs R. Fabricating budgets: a study of the production of management budgeting in the National Health Service. Acc Organ Soc 1992;17:571–94. Richardson A. Regulatory networks for accounting and auditing standards: a social network analysis of Canadian and international standard-setting. Acc Organ Soc 2009;34:571–88. Sampson RJ, Laub JH. Crime in the making pathways and turning points through life. Cambridge, Massachusetts: Harvard University Press; 1993. Scott J. Social network analysis. Newbury Park, CA: Sage Publications; 1991. Shaw CR, McKay HD. Juvenile delinquency and urban areas. Chicago, Illinois: University of Chicago Press; 1969.
J. Worrell et al. / International Journal of Accounting Information Systems 14 (2013) 127–137
137
Smatt C. An investigation of the impact of the structure and quality of relationships on knowledge exchange and individual performance. Management Information Systems. Tallahassee: Florida State University; 2009. p. 143. Uzzi B. Social structure and competition in interfirm networks: the paradox of embeddedness. Adm Sci Q 1997;42:35–67. Wakefield R. Networks of accounting research: a citation-based structural and network analysis. Br Acc Rev 2008;40:228–44. Walsham G. Actor-network theory and IS research: current status and future prospects. In: Lee A, Liebenau J, DeGross J, editors. Information systems and qualitative research. London: Chapman and Hall; 1997. Walsham G, Sahay S. GIS for district-level administration in India: problems and opportunities (n1). MIS Q 1999;23:39. Wasserman S, Faust K. Social network analysis. Cambridge: Cambridge University Press; 1994.