Accounting, Organizations and Society 24 (1999) 317331

Towards a meta-theory of accounting information systems

Elaine G. Mauldin a, Linda V. Ruchala b, *
a

School of Accountancy, University of Missouri-Columbia, 320 Middlebush Hall, Columbia, MO 65211, USA b University of Nebraska-Lincoln, 380 CBA, Lincoln, NE 68588-0488, USA

Abstract The purpose of this paper is to articulate a model for accounting information systems (AIS) research that synthesizes the primary theoretical perspectives of the extant literature. Building on the three orientations used in prior research (technological, organizational and cognitive approaches) and adopting an explicit systems perspective, we develop a model that links system design alternatives to the three orientations and to task performance. The model places a central focus on the accounting task and suggests a matching process between requirements of the task and system design alternatives at multiple levels of analysis. We also demonstrate how the application of the model suggests future research opportunities, organized around four research propositions. # 1999 Elsevier Science Ltd. All rights reserved.
Keywords: Accounting information systems (AIS); Information technology; Expert systems; Decision support systems; Decision aids

1. Introduction Extant accounting information systems (AIS) research has evolved from the source disciplines of computer science, organizational theory and cognitive psychology. An advantage of this evolution is a diverse and rich literature with the potential for exploring many dierent interrelationships among technical, organizational and individual aspects of judgment and decision performance. AIS research also spans from the macro to the micro aspects of the information system (Birnberg & Shields, 1989). The broad scope of AIS applicability suggests a potentially large sphere of inuence for research done in this area. Conversely, the diused theoretical basis also has resulted in a

``nebulous'' focus for AIS research (McCarthy, 1987) and an ``absence of identity'' (Sutton, 1992). For AIS researchers, the ``absence of identity'' problem has been exacerbated by the fact that initial applications of information technology (IT)1 to accounting systems were transaction processing systems that tended to mirror historically developed manual processes. Sub-disciplines of accounting (e.g. auditing, nancial, managerial, and tax) often did not perceive a need to incorporate systems considerations into their research. Now, as IT grows more advanced, AIS applications are recognized as fundamentally changing
1 We follow March and Smith's (1995, p. 252) denition of information technology as ``Technology used to acquire and process information in support of human processes. It is typically instantiated as IT systemscomplex organizations of hardware, software, procedures, data, and people, developed to address tasks faced by individuals and groups, typically within some organizational setting.''

* Corresponding author. Tel.:+1-402-472-8812; fax:+1402-472-4100; e-mail: lruchala1@unl.edu

0361-3682/99/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved. PII: S0361 -3 682(99)00006 -9

318

E.G. Mauldin, L.V. Ruchala / Accounting, Organizations and Society 24 (1999) 317331

task processes and providing complex decision support, as opposed to simply increasing the speed and accuracy of traditional accounting tasks. As a result, there is a growing need to integrate AIS research more closely with the other sub-disciplines of accounting research. Practicing accountants are being asked to focus their attention on the design of AIS to add competitive value to information and re-dene the scope of AIS (Brecht & Martin, 1996). An integration of AIS research within sub-disciplines could also provide direction for practice. The comparative advantage of accounting researchers within the study of IT lies in their institutional accounting knowledge. Systems researchers can contribute insights into the development of systems utilizing technology, and the other sub-areas can contribute insights into the task characteristics in the environment. For instance, systems researchers have extensively investigated group decision support systems (GDSS), but they have only recently been considered in auditing. On the other hand, auditing research has extensively investigated the role of knowledge and expertise. The merging of the two sets of ndings may be relevant to AIS design, training, and use.2 The changes in the nature and purpose of AIS, combined with the diused theoretical perspectives, suggest a need for an organizing paradigm that provides a means for integrating extant AIS research and mining the synergetic opportunities that diering perspectives hold. We use the term ``meta-theory'' to suggest the integration and synthesis of the technical, organizational, and cognitive orientations into an overarching model for AIS research. Prior AIS research (e.g. Cushing, 1990; Reneau & Grabski, 1987) has adopted or adapted well-regarded frameworks from the management information systems (MIS) and IT areas (e.g. Ives, Hamilton & Davis, 1980; Mason &
2 Bamber (1993, p. 15) reaches a similar conclusion in discussing research opportunities for behavioral AIS research: ``This research will be most fruitful if it focuses on those domain-specic variables unique to accounting systems and on the accounting decision context... Moreover, as new information technology is integrated into all accounting sub-elds, collaborative research will blur the distinctions between subelds.''

Mitro, 1973; Gorry & Scott-Morton, 1971). The meta-theory model extends prior research by addressing the following four limitations of existing IT frameworks suggested by March and Smith (1995, p. 255): . failure to recognize that ``the nature of the task to which the IT is applied is critical''; . failure to ``recognize the adaptive nature of articial phenomena; the phenomena itself is subject to change''; . failure to account for the ``design science research being done in the eld''; . failure to ``provide direction for choosing important interactions to study'' and treating all interactions as equal. At the center of the approach in this paper is a task-orientation and system perspective that recognizes key interrelationships among major categories of contingency factors; technological, organizational and cognitive.3 A primary result of this paper is the proposition that an AIS research model should be reoriented around a task focus. A signicant body of prior research into judgment and decision making performance suggests that performance is highly dependent upon contextual elements of the task (Payne et al., 1993; Solomon & Shields, 1995); the evaluation of an AIS should be considered in conjunction with task achievement. This reorientation around task also aids an understanding of what March and Smith call ``the adaptive nature of articial phenomena''. Task requirements establish the set of system design characteristics4 and alternatives possible for use. Changes in the level of performance needed and/ or requirements of the task motivate adapting IT and changing the set of options available for
3 Similar structures have been suggested by prior research examining decision processes. Newell and Simon (1972) identied three factors in human problem solving as decision-maker characteristics, task demands, and organizational context. Payne et al. (1993) suggested three major classes of factors inuencing decision strategy as the problem, the person, and the social context. 4 Systems design characteristics are dened as the set of hardware, software, procedures, data, and people, developed to address a particular organizational task (March & Smith, 1995).

E.G. Mauldin, L.V. Ruchala / Accounting, Organizations and Society 24 (1999) 317331

319

system design and implementation.5 As March and Smith (1995, p. 252) state: ``technologies are often developed in response to specic task requirements using practical reasoning and experiential knowledge''. The task orientation of the proposed model also suggests an explicit role for design research within the AIS domain. March and Smith (1995) suggest that prior frameworks have omitted a role for ``design science''the building and evaluating of technological models, methods and instantiations. A focus on the task and on task performance makes explicit the need to develop systems that improve task outcomes and to evaluate implementations based upon their incremental task improvement. Finally, integrating the three source disciplines traced in this paper around a task focus is also intended to help identify which interactions might be more important to study than others and the forms that interactions may take. This approach recognizes key interrelationships between technology, organization science and cognitive processes. AIS technologies, in the form of databases, expert systems and decision aids, represent tools for solving accounting problems. As such, these tools must be selected and implemented considering the task requirements and cognitive and organizational features embedded within that implementation. Users of AIS have diering problem and organizational constraints that necessitate diering AIS approaches. Hopwood (1989, p. 15) states: ``. . .accounting oriented knowledge is more likely to be gained by research which systematically attempts to study the interdependency and interpenetration of the technical and the human and the organizational.'' The remainder of this paper proceeds as follows. The paper begins by providing an overview of the AIS meta-theory model with illustrations of the model's design principles. Then, four research
5 Changes in the level of performance needed by a rm could include the need to achieve an equivalent level of performance for less cost as well as an increase in performance. Value-chain analysis and business process reengineering are two examples of initiatives motivating the adapting of IT to make task processes and outcomes more ecient or eective.

propositions are developed and future research implications presented. 2. A meta-theory model for AIS research AIS encompass systems that are either used by accountants, by other decision-makers employing accounting information, or in tasks that involve the application of accounting data (Reneau & Grabski, 1987). The model developed in this paper builds on previous frameworks in the management information systems (MIS) (e.g. Gorry & ScottMorton, 1971; Ives et al., 1980; Mason & Mitro, 1973), and AIS (Cushing, 1990; Reneau & Grabski, 1987) literature. The meta-theory model is illustrated in Fig. 1 and will be explicated through a discussion of four organizing principles, one for each major component of the model. The discussion of each organizing principle includes a description of the theoretical roots of each component. Representative examples of prior research are used to underscore the need for an extended integrated framework. Taken together, these four principles constitute a nomological network with task characteristics at the core. 2.1. Organizing principle No. 1: task focus The rst organizing principle of the framework is that AIS research should be task focused. Fig. 1 illustrates this principle through the centrality of task characteristics to system design alternatives and task performance. The importance of task focused research is increasingly recognized, both in psychology and accounting (Einhorn & Hogarth, 1981, 1993; Waller & Jiambalvo, 1984). Gibbins and Jamal (1993, p. 453) argue that research without task considerations may be uninformative: ``Without careful observation of the task, it is dicult to relate the research results to important questions about the task and accountants' performance of it, such as eectiveness, eciency, expertise, and value.'' Payne et al. (1993) emphasize that decision-makers can invoke dierent strategies depending upon even seemingly minor task features, such as the number of alternatives or outcomes. A task focus is necessary for

320

E.G. Mauldin, L.V. Ruchala / Accounting, Organizations and Society 24 (1999) 317331

Fig. 1. A meta-theory model for AIS research.

designing and implementing systems in consonance with human judgment and decision processes and organizational strategies. Thus, to understand and improve AIS strategies, research must rst understand the task demands. This organizing principle extends prior AIS models, such as Reneau and Grabski (1987), which have subsumed task elements within broader categories of environmental factors constraining IS development. 2.1.1. Dening task characteristics The concept of task is multidimensional. AIS serve many types of business tasks, including strategic, managerial, and operational (Gorry & Scott-Morton, 1971), and each type of task can be found in all of the accounting sub-disciplines. Although classication by business purpose is intuitively appealing, shifting the focus to core task characteristics may provide new insights for understanding systems. There exists no denitive taxonomy of task characteristics, although several accounting researchers have stressed the importance of task characteristics in understanding cognitive processes (e.g. Bonner & Pennington, 1991; Gibbins & Jamal, 1993). The meta-theory model of AIS denes four core dimensions of task characteristics that have been considered in prior accounting and AIS research. The rst is the mental processes involved in the task, such as external information search, knowledge

retrieval, hypothesis generation and evaluation, problem solving design, estimation or prediction, and choice. This dimension is analogous to distinctions made in the behavioral auditing and judgment and decision making literatures, such as in Bonner and Pennington (1991). The second is complexity, considered to be a function of both the amount and clarity of inputs, processing, or outputs (Bonner, 1994). The third dimension is task demands as suggested by a task demand continuum developed by Stone and Gueutal (1985). One end of the continuum is anchored on physical labor and the other end is symbolic, abstract knowledge. While many accounting studies focus purely on the knowledge end of the spectrum, the entire spectrum is important since AIS increasingly impacts a variety of non-accountant decision-makers utilizing accounting information, such as production and clerical personnel. The fourth and nal dimension is frequency of occurrence. Regularities in the task over time lead to dierent adaptations in the cognitive process and in the evolution of systems (Simon, 1981; Gibbins & Jamal, 1993). 2.1.2. Contributions of a task focus Carefully dening the underlying task requirements, as well as comparing and contrasting those requirements to tasks previously studied, is a critical event necessary to further our understanding of information systems in accounting

E.G. Mauldin, L.V. Ruchala / Accounting, Organizations and Society 24 (1999) 317331

321

contexts (Peters, 1993). More clearly focused, task oriented research may improve the ability to select appropriate technology to study, a major problem identied in previous AIS research. Technological advances arrive at an ever-increasing pace such that it is extremely dicult to determine, a priori, what should and should not be studied. Weber (1987, p. 7) suggests that researchers have been ``seduced'' by the excitement of new technology: As each wave of new technology occurs, the literature shifts to the predictable descriptions of the technology, examination of implications, case studies of use, exhortations to adopt, etc. . . . The end result is a eld devoid of theory and general principles. March and Smith (1995) argue that the challenge for future AIS research is to categorize technology so that research eorts will not be wasted building and studying systems that have already been built and studied ``in kind''. They oer guidance by dividing research activities into four categories (build, evaluate, theorize, and justify), and identifying criteria for research within each domain. For instance, research that builds systems should be judged based on value or utility to a community of users. Building the rst application has utility only if the task application is important. Subsequent build studies must justify signicant improvement. The rst organizing principle of the meta-theory model provides additional guidance for the design science research challenge in two ways. First, analysis of task requirements needed for a particular accounting function may assist in identifying the potential of a new technology for improving that function. Second, task analysis may assist in determining how well research using a particular task in one area will translate to a task in another area, thereby helping researchers focus on extensions in the application of technology rather than replications of ``in kind'' systems. 2.2. Organizing principle No. 2: design process The centrality of task leads to the second organizing principle illustrated in Fig. 1 by the left

arrow leading from task characteristics to system design alternatives. We suggest that, for AIS, task requirements become the start of a process that establishes the set of system design characteristics needed. It is important to realize that this organizing principle is about designing an Accounting Information System and not just about computers and electronic networks. Simon (1997) states: ``in the post-industrial society, the key problem is how to organize to make decisionsthat is to process information'' and points out that decision making is shared between the human and mechanized components of systems. The metatheory model suggests that task needs must be considered in choosing among system alternatives for a division of labor between the human and computer components. System design alternatives, however, operate at dierent levels of analysis. We adopt March and Smith's (1995) classication of constructs, models, methods, and instantiations. Constructs are conceptualizations used to describe problems and specify their solutions. Constructs operate at a high level of abstraction. For instance, in an enterprise wide accounting system, constructs would include ``data should be captured that supports the organization's critical events'' and ``detailed data should be maintained in one place that can be accessed by many dierent users''. Models are propositions expressing relationships among constructs. The Resources Events Agents (REA) model developed by McCarthy (1982) and discussed below is an example of an enterprise wide model. Methods are a set of steps used to perform a task. Continuing the example, database tables and data dictionaries are two of the many methods that would be used in an enterprise wide accounting system. Instantiations are realized implementations of technology, such as SAP R/3 for an enterprise wide accounting system example. The model accommodates research at any of the various levels of analysis. The model also recognizes, however, that a system design alternative choice made at a higher level (such as construct) will often subsequently limit choices at lower levels (models, methods, or instantiations). Organizing principle No. 2 suggests a matching

322

E.G. Mauldin, L.V. Ruchala / Accounting, Organizations and Society 24 (1999) 317331

process between requirements of the task and design alternatives at each level of analysis. 2.3. Organizing principle No. 3: contingency factors The task and system design alternatives may be central to the model, but they do not exist in a vacuum. The third organizing principle of the model is that the eects of AIS on task performance must be examined within the context of cognitive, technological, and organizational contingency factors. Fig. 1 illustrates the three contingency factors as jointly aecting task characteristics. We adopt an explicit systems perspective where the focus is on the interdependence of the elements of the system (Frances & Garnsey, 1996). 2.3.1. Cognitive factors AIS cognitive research focuses on the impact of AIS on individual and group decisions and draws on the judgement and decision making and psychology literatures.6 Reviews by Huber (1983), Reneau and Grabski (1987), Amer, Bailey, and De (1987) and Weber (1987) all characterize behavioral AIS research through the mid-80's as lacking a theoretical basis and generating inconsistent results. More recent AIS research has moved to a closer investigation of judgment and decision processes and strategies, mirroring trends in behavioral research within other accounting areas. The comparative advantage of the cognitive perspective is its recognition that technology aects the decisionmaker in three dierent ways: intended technical eects, transient eects associated with assimilation, and unintended social eects (Kiesler, 1986). An example of the investigation of unintended eects is cognitive research focusing on identifying changes in cognitive strategies in the presence of automated decision aids designed to improve task performance. Chu (1991) nds that a decision aid
We thank a reviewer for noting that there is also an impact of cognitive factors on AIS design and evaluation in that the design and evaluation are by people. The cognitive functioning of individuals doing the AIS design and evaluation also has interesting potential research implications. In a sense, the model captures this issue if we view AIS design and evaluation as a task having its own alternative design structures and cognitive, technological and organizational components.
6

does not reduce the potentially biasing use of simplifying strategies as task complexity increases. Subjects using a decision aid generated alternatives using an incremental, ``satiscing'', approach signicantly more often than non-users, suggesting that the decision aid may induce users to adopt a particular process in a manner that system designers may not have intended. Kottemann, Davis and Remus (1994) nds that subjects making production-planning decisions with a decision aid expressed inated condence beliefs in their performance, even though their actual performance was lower than subjects without the aid. One possible cause of this eect is a tendency to assume that the computer is ``right''. The ndings of these studies illustrate that the intended technical eect of improved performance is not automatic with the introduction of a decision aid. Subtle task and training dierences can aect both performance and research results. The meta-theory model helps to extend prior AIS cognitive research in two ways. First, Libby and Luft's (1993) expertise paradigm is adopted to represent the cognitive contingency factors in the model. The expertise paradigm concludes that task performance is contingent upon ability, knowledge, motivation and environment. The Libby and Luft model is useful to explicate some of the interrelations among the cognitive characteristics (including knowledge and motivation) with the other two contingent factors identied in the metatheory model (technological and organizational factors) and with the task characteristics. The expertise paradigm has been widely embraced as the basis for current cognitive auditing research in expertise development (e.g. Bonner & Walker, 1994; Nelson, 1993) and is generalized to AIS. Second, the model provides a clearer focus on accounting specic tasks, as well as organizational contexts, a research extension suggested by Bamber (1993). 2.3.2. Technological factors Information technology refers to the tools and techniques that have been developed specically to acquire and process information in support of human purposes (March & Smith, 1995). The meta-theory model incorporates Sutton's (1992) representation of technology as consisting of three

E.G. Mauldin, L.V. Ruchala / Accounting, Organizations and Society 24 (1999) 317331

323

categories of primary interest to AIS researchers: information economics, diusion, and presentation. Information economics is concerned with the determination of content, considering expected costs and benets of the system. Information diffusion is concerned with how to get information to users, while information presentation is concerned with format issues. The modeling of a multidimensional accounting system is one example of information economics research, highly regarded because it is theoretically grounded in database management research and has often been ahead of practice (Sutton, 1992; Amer et al., 1987). Multidimensional accounting systems are based on the idea that accounting should provide disaggregated data about economic events that can allow individual users to generate information needed for a variety of judgments and decisions. Although the basic idea has been around for many years (Goetz, 1939; Schrader, 1962; Sorter, 1969), it was not until the 1970s that technology began to emerge to enable implementation of a new accounting model. Godfrey and Prince (1971) demonstrated how computerbased information networks could be used to provide disaggregated data to users. More explicit models of multidimensional accounting systems incorporated advances in database technology (Colantoni, Manes & Whinston, 1971; Lieberman & Whinston, 1975; Haseman & Whinston, 1976). McCarthy (1982) introduced the ResourcesEvents-Agents (REA) accounting model, extending prior database models by clearly including semantic modeling (based on economic exchanges not debits and credits) and more fully developing the structure of the accounting model. The REA model can be used as a starting point in enterprise wide database designs. Advances in future research may be stimulated through the meta-theory model's integration of technology with organization and cognitive perspectives. A problem identied in prior technology-based research is the fact that technological performance is related to the environment in which it operates. Incomplete understanding or consideration of that environment may lead to inappropriately designed applications of technology, resulting in undesirable side eects, such as under-utilization

of technology (March & Smith, 1995). The challenge for future research is to not only determine what technology works, but also understand why technology utility may dier in dierent applications. For example, Dunn and McCarthy (1997) suggest that validation of the REA model at either the individual or organization level is the greatest need for future research. The double-sided arrow between technology and cognitive characteristics in Fig. 1 recognizes that, on the one hand, technology may enhance knowledge and provide decision support while on the other hand, cognitive strategies and individual-level factors could aect the technological design and mode of operation chosen. For example, one area of MIS research involves system design features to enhance exibility. Advances in technology may also increase the ease in adapting technology to individual preferences, analogous to the ``smart'' systems found in automobile design in which individual climate, lighting and ergonomic preferences can be accommodated. 2.3.3. Organizational factors The organizational level interactions are represented within Fig. 1 as the set of linkages from technology to task performance through the organizational and task boxes. The link between technological and organizational factors is also two-way since technology may allow changes to the organization structure, for example, and the organization context will determine the technology chosen. Organizational factors include organizational strategy, structure, and the internal and external business environment. Prior organizational research focuses on the design and implementation of AIS, as well as the relationship between AIS choice and organizational performance. Accounting research into the organizational determinants of AIS have generally been based on three theories: contingency, agency and transaction cost economics. 2.3.4. Contingency theory Early contingency research, summarized by Otley (1980), found no universally appropriate management accounting system (MAS) applicable to all organizations in all circumstances. Conceptual frameworks developed by Waterhouse and Tiessen

324

E.G. Mauldin, L.V. Ruchala / Accounting, Organizations and Society 24 (1999) 317331

(1978) and Reneau and Grabski (1987) agree that the internal and external environment and production technology are two primary categories of contingent variables to be considered in choosing an AIS. Empirically, Chenhall and Morris (1986), Gul (1991), and Gul and Chia (1994) examine the contingent relationship between internal and external environmental factors (organizational interdependence, decentralization, and perceived environmental uncertainty) and choice of an AIS (information scope, timeliness, aggregation, and integration). The general conclusion from these studies is that greater organizational interdependence, decentralization, and perceived environmental uncertainty are factors associated with either a greater perceived need for more sophisticated AIS or higher rm performance with more sophisticated AIS. Other contingency research (Chong, 1996; Kim, 1988; Mia & Chenhall, 1994) includes the role of production technology and task variables in assessing the performance and satisfaction with MAS. Kim (1988) nds, in a hospital setting, that when tasks are highly predictable and impersonal communication is used, MAS user satisfaction is higher. Mia and Chenhall (1994) and Chong (1996) nd that the use of broad scope MAS positively aects performance only in the presence of high task uncertainty. These contingency theory ndings are particularly germane in the current environment of rapid technological developments. Many business articles seem to imply that all organizations should adopt the latest technology (Brecht & Martin, 1996; Magretta, 1998; Piturro, 1998), yet empirical ndings suggest that this could be counter-productive. For instance, Gul and Chia (1994) and Chong (1996) argue that, where either environmental or task uncertainty is low, broad scope MAS may have an adverse eect on performance due to information overload. The meta-theory model provides a clearer task focus with a systems perspective that may guide future MAS choice research. While recognizing that contingency theory is a valid approach to the study of organizational factors and technology, contingency theory has been criticized for lacking a substantive basis to suggest which variables, and which combinations of variables, are important (Fisher, 1995; Otley, 1980;

Weber, 1987). That is, it is dicult to argue against inclusion of any of the contingent variables, yet equally dicult to determine their completeness or to know which combinations of factors make sense and are more important. For management control design in general, Fisher (1995) calls for more comprehensive models that provide a theoretical basis for multiple interactions among variables. 2.3.5. Economics based theories The meta-theory model adopts suggestions from Tiessen and Waterhouse (1983) and Spicer and Ballew (1983) that transaction cost economics (TCE) and agency theory may sharpen the focus of contingency theory by providing a justication for identifying factors that aect MAS choice. TCE and agency theory share similar roots in economic theory but dier somewhat in their assumptions and research focus. Agency theory already enjoys a prominent role in the investigation of information system choice. Analytical research has focused on the role and value of information in organizations and choice of information systems under conditions of moral hazard (e.g. Arya, Glover & Sivaramakrishnan, 1997; Baiman, 1975). TCE focuses on the structure of the underlying economic transactions: standardization, frequency, uncertainty and investment in unique, or specic, assets. TCE is a comparative theory and, so, would be suited to empirical cross-sectional investigations of organizational dierences in the design and implementation of MAS (Williamson, 1985). Reference to TCE and agency theory could also improve the operationalization of constructs used in prior contingency research. For example, environmental uncertainty has been consistently operationalized using a validated instrument measuring perceived uncertainty. However, a wider range of uncertainty could be examined through comparison of multiple industries in varying locations. All three of the theories that comprise the organization science orientation bring to AIS research increased recognition of the ways in which technology may be impacted by, and may impact, strategic, contractual and competitive decisions of the rm. Their competitive advantage is in situating the rms' information needs in a

E.G. Mauldin, L.V. Ruchala / Accounting, Organizations and Society 24 (1999) 317331

325

broader strategic and dynamic business context. The meta-theory model presents a comprehensive view of AIS by integrating theoretically defensible categories of variables with a task focus. 2.4. Organizing principle No. 4: task performance The nal organizing principle is that the AIS outcome is task performance. While principle one begins with the task orientation, principle four ends with an outcome orientation, representing the fulllment of the task. Task performance is chosen as the outcome variable because both organizational and IT researchers dene information in terms of the value it holds for human action (March & Smith, 1995; Simon, 1997). However, task performance may be operationalized from multiple standpoints, including the eciency, eectiveness (cost- or process-), output quantity and/or quality, or timeliness of the action taken. Performance criteria will vary according to the level of analysis being conducted, corresponding to the system design alternatives discussed above. For instance, at the construct level, task performance may be strategic eectiveness while at the instantiation level it may be individual decision-maker performance. The meta-theory model for AIS research depicted in Fig. 1 is a comprehensive, dynamic model that formally recognizes complex interactions between the four components of the model, task demands, system design alternatives, contingency factors, and task performance. The model provides an overview perspective to understand prior research in AIS. The expected contribution of the metatheory model is the stimulation of research emphasizing the systemic nature of AIS interactions among the components of the model, as outlined in the organizing principles above. To that end, the following section illustrates four research propositions suggested by the meta-theory model. 3. Research propositions 3.1. Research proposition No. 1 The rst research proposition is that a temporal ordering of technology, cognitive and organizational

inuences does not exist; these coexist and evolve simultaneously. This proposition is based on the systemic perspective of interdependence described in the third organizing principle of the model. One striking example of the coexistence of multiple inuences is found in the application of process improvement (Davenport, 1993) or business process reengineering (Hammer & Champy, 1993). These techniques are heavily dependent upon information technology to transform business processes. However, the central point in such analyses is not to change the fundamental business task, but rather to gain synergy through the use of technology in the process of task accomplishment. Hammer and Champy (1993, p. 47) clearly state this point: ``Information technology acts as an enabler that allows organizations to do work in radically dierent ways.'' Research in MIS has found that the biggest obstacle to successful reengineering is change management (Clemens, Thatcher & Row, 1995). Change management involves both organization and individual factors, indicating the importance of the systemic perspective instead of a temporal ordering. The meta-theory model can guide future research examining the interdependence of the three contingency factors. For example, future research could identify how control systems are impacted by, and impact, reengineering eorts at either a macro or a micro level. Specically on the micro level, agency theory could provide one approach to examining interdependencies between technology, cognitive, and organizational inuences. The idea is that technology may be used as a control mechanism to ensure that the agent's actions are consistent with the goals of the principal by placing limits on the action decision-makers may take. Using AIS for increased monitoring has implications for reducing information asymmetry, changing existing relationships between the principal and agent. This approach, largely unexplored in the extant literature, is particularly complex since it involves technology, organization control structures, and ability, motivation, and knowledge issues from the cognitive perspective. Mauldin (1999) examines compensation schemes in an expert system environment and concludes that the introduction of technology changes not only the eort eects of

326

E.G. Mauldin, L.V. Ruchala / Accounting, Organizations and Society 24 (1999) 317331

incentives, but also the type of individual attracted to the organization. She also identies task dierences in these relationships. Much remains to be done in this area including examination of other control structures, such as amount and types of rm supervision. A dierent approach to testing the rst research proposition would be to hold both task and technology constant to more carefully identify the eects of cognitive and organizational inuences. For example, a bond-rating task7 shares many common task demands with a variance analysis task. Both require the integration of multiple cues including retrieval, recognition, comparison, and projection, and both are done on a recurring basis. In spite of these task commonalties, there are likely to be dierences in organization structure or cognitive experiences (such as dierences in performance evaluation, incentive structures, or ability) that lead to dierences in the eects of technology on performance. Technology for the common task characteristics could be examined in the two contrasting environments to begin to understand the complexities arising in the real world. 3.2. Research proposition No. 2 The second research proposition is that the relative signicance of technological, cognitive, and organizational factors dier for dierent tasks. A good illustration of this research proposition is the use of decision aids for dierent tasks. The accounting profession has devoted considerable attention to the development and implementation of structured decision aids (Messier, 1995). Benets commonly associated with aided judgment include enhanced performance, reduced decision variability, and facilitation of knowledge transfer (Libby & Luft, 1993). Two types of problems in decision-aided tasks have been identied by prior research. The rst is non-reliance on the aid even though the aid has been shown to produce, on average, better outcomes than un-aided judgment (Ashton, 1990; Eining, Jones & Loebbecke, 1997;
7 Such as in the information presentation study by Amer (1991).

Boatsman, Moeckel & Pei, 1997). The second is inappropriate reliance where the aid's recommendation is used even when the decision-maker is aware that the aid does not incorporate all relevant features of the environment (Glover, Prawitt & Spilker, 1997). The non-reliance problem has been found in rating tasks where multiple items of information are used to infer states of the world in a probabilistic setting (determination of bond ratings or management fraud risk assessments). The inappropriate reliance problem was found in a computation problem (calculation of income tax). The two tasks diered on at least two dimensions of task demands, cognitive and complexity. However, incentive structures and knowledge and ability of subjects also diered in the studies. A likely explanation of the dierences in outcome is that there are dierent weights on cognitive, organizational, and technology factors related to each task. The meta-theory model could guide future research explaining the weightings of each of the factors for each type of task, to provide direction for system design alternatives by task. 3.3. Research proposition No. 3 The third research proposition is that there is a relationship between the dierent levels of analysis such that design choices made at a higher level (such as construct) will aect those at a lower level (such as model). March and Smith (1995) suggest that system development methods move from abstract constructions of problems (constructs) to increasingly concrete representations of user needs (models), system requirements (methods) and implementations (instantiations). This hierarchy logically implies that judgments made at any one level will limit the subsequent choices available at a lower level. Within the accounting arena, an important research question is whether and how the problem denition at the construct level changes alternative sets of models, methods and instantiations. Brecht and Martin (1996) suggest a redenition of the scope of AIS, at the construct level, to include strategic, management and task/operational information. Future research is needed to evaluate the

E.G. Mauldin, L.V. Ruchala / Accounting, Organizations and Society 24 (1999) 317331

327

merits of this revised denition against not only internal requirements, but also external requirements for information, such as real-time, on-line information retrieval by investors. Future research is also needed to investigate specic changes that might stem from a change in the construct level denition of AIS. Implementation problems, at the instantiation level, associated with realizing a redenition such as that described above are largely due to the way accounting information has traditionally been dened. David (1997) investigates how the changing denition at the construct level requires changes at the model level. She rst denes three types of events: economic events (actual transactions changing resource quantities), business events (information for planning and control) and information events (means of communicating information about the former two events). Then, using a REA framework, she illustrates how a traditional accounting system (i.e. limited to economic events) could evolve into a very dierent model if business and information events were included. She also discusses the implications of this modeling for process reengineering, noting that cost-benet analyses are vital in considering alternative models and methods. Future research could empirically validate the match between a revised denition of accounting information systems at the construct level and an REA model for implementation. Just as the construct level eects the model level, decisions made at the model level eect the methods and instantiations level. For example, instantiations of articial intelligence (AI), using neural nets and case-based reasoning, are able to quickly discern patterns in data that would take years to discover using older techniques. Several banks and credit card companies are now using AI to study patterns of individual credit card usage to detect transactions that are potentially fraudulent (Widrow, Rumelhart & Lehr, 1994). It is a revised accounting model (maintaining a wealth of detail transaction data) that makes AI methods possible. The decision to model the accounting system multidimensionally allows even traditional transaction data to be used in new instantiations. The meta-theory model could guide future research in identifying and examining the eects of other such applications.

3.4. Research proposition No. 4 The fourth research proposition states that system design alternatives should explicitly consider complementarities and t between strategy and structure. Complementarities in organizations, introduced by Milgrom and Roberts (1995), are based on the idea that activities are complements if doing more of any one of them increases the returns to doing more of the others. In a system with strong complementarities, changing only a few of the system elements at a time may not achieve all the benets that are available through a fully coordinated move, and may even have negative payos (Milgrom & Roberts, 1995). Frances and Garnsey (1996) demonstrate how capitalizing on complementarities in technology, distribution, and employment practices allowed supermarkets in the United Kingdom to change the structure of market control, resulting in dramatically higher net margins than in many other countries. Greater recognition of, and empirically testing, research proposition four is particularly important in research building or evaluating emerging technologies. For example, the Internet has the potential to completely redene external nancial reporting as we know it today. Elliott (1995) predicts real-time business reporting and investor access to corporate databases. Research dening complementarities in rm strategy and structure governing existing external relationships and how they are changing seems particularly appropriate to avoid negative consequences of such an implementation. There is a huge gap between having the technology available to provide any user any data within the rm and having a system that will control both the accuracy of the data (important to the external user) and the legitimacy of, and limitations on, data requests (important to the rm). AI is theoretically capable of providing a vehicle of control. The meta-theory model could guide future research analyzing the eect of AI on nancial reporting concerns. Another example of capitalizing on complementarities involves group support systems (GSS), a category of technology that has been extensively studied within the MIS domain and

328

E.G. Mauldin, L.V. Ruchala / Accounting, Organizations and Society 24 (1999) 317331

recently considered in audit research. Bamber, Hill and Watson (1998) review prior audit group research as well as recent audit GSS research and provide a series of propositions for future AIS research, specic to auditing. They note inconsistent ndings in MIS research and suggest the importance of evaluating potential GSS applications in a context-specic basis. The framework they develop for further GSS audit research explicitly encompasses task characteristics as well as the group setting and expected GSS process gains and losses (Bamber, Watson & Hill, 1996). The authors note a lack of GSS research in the areas of conict resolution and execution tasks, which are potentially important tasks in the audit function. Research proposition four goes one step further to recommend examination of the adoption and diusion of GSS. The process gain and process loss perspective can help identify complementarities. For example, the eects of networking and workpaper automation through GSS could be positive, in allowing reviewers more freedom, or negative in having less personal contact with audit sta and client representatives. With networking technology, GSS is becoming an important part of audit environments due to the need for interaction among the audit team and the geographic dispersion of audit activity. Reaping the benets of this technology might also require complementary procedures to mitigate the negative impacts. The meta-theory model can guide research investigating how the process gain and loss tradeos are made within organizations. Such research would be relevant to increasing the success of system innovations, not only in auditing, but also in other areas. Research proposition four applies to models as well as instantiations of system design alternatives. For example, the development of the REA model discussed earlier is predicated on reduction of redundancy and making data structures more ecient. However, Van Alstyne, Brynjolfsson and Madnick (1995) argue that, under some current organization structures, an REA type model may not be eective. Future research identifying complementarities in organization structures could, once again, impact implementation strategies.

4. Conclusions Technology is a pervasive and growing component of accounting tasks and has been shown to change work processes. The meta-theory model for AIS research contained in this paper explicitly recognizes this factor. Currently, multiple research methods are being employed to investigate dual AIS/accounting problems. This is especially evident in managerial accounting, where experimental, eld, and analytical work has addressed the interrelationships between managerial tasks and AIS design and use. Less evident is an explicit combined strategy to employ multi-methods in a manner that most eectively uses the strengths of each approach (Peters, 1993). The meta-theory model develops a comprehensive framework that can be used to help organize and interpret future research. The meta-theory model may also provide a vehicle to encourage the integration of AIS research with the other sub-disciplines of accounting research. Explicitly considering task commonalities and dierences between sub-disciplines may avoid unnecessary duplication of research as well as accelerate the advance of AIS research. This paper develops a meta-theory model for AIS research that begins with a task focus and suggests a matching process between requirements of the task and system design alternatives at four dierent levels of analysis. Further, the model suggests that technological, organizational, and cognitive contingency factors impact the outcome, task performance. The three contingency factors are developed from a review of the extant literature and are complimentary, each focusing on an important aspect of AIS design and use in addressing accounting issues. The model provides a nomological structure that can be used in future research to identify variable combinations and critical interactions for specic task settings. Four propositions for future research are also identied using the proposed model. These propositions illustrate questions that have been under-researched regarding the temporal ordering of contingency factors, the relative importance of contingency factors for dierent tasks, the consequences of system design choices and the complementarities in AIS.

E.G. Mauldin, L.V. Ruchala / Accounting, Organizations and Society 24 (1999) 317331

329

Acknowledgements We appreciate helpful comments received from Vicky Arnold, John Barrick, Jean Cooper, Severin Grabski, Scott Jackson, Uday Murthy, Bob Ramsay, Mark Taylor, the 1997 AIS Research Symposium participants, and the anonymous reviewers. The second author also gratefully acknowledges the nancial support of a KPMG Peat Marwick Research Fellowship. References
Amer, T., Bailey, A. D., & De, P. (1987). A review of the computer information systems research in accounting and auditing. Journal of Information Systems, 2, 328. Amer, T. (1991). An experimental investigation of multi-cue nancial information display and decision making. Journal of Information Systems, 5, 1834. Arya, A., Glover, J., & Sivaramakrishnan, K. (1997). The interaction between decision and control problems and the value of information. The Accounting Review, 72, 561574. Ashton, R. H. (1990). Pressure and performance in accounting decision settings: Paradoxical eects of incentives, feedback, and justication. Journal of Accounting Research, 28, 148186. Baiman, S. (1995). The evaluation and choice of internal information systems within a multi-person world. Journal of Accounting Research, 13, 115. Bamber, E. M. (1993). Opportunities in behavioral accounting research. Behavioral Research in Accounting, 5, 129. Bamber, E. M., Hill, M. C., & Watson, R. T. (1998). Audit groups and group support systems: A framework and propositions for future research. Journal of Information Systems, 12 (2), 4574. Bamber, E. M., Watson, R. T., & Hill, M. C. (1996). The eects of group support system technology on audit group decision making. Auditing: A Journal of Practice and Theory, 15, 122134. Birnberg, J. G., & Shields, J. F. (1989). Three decades of behavioral accounting research: A search for order. Behavioral Research in Accounting, 1, 2374. Boatsman, J. R., Moeckel, C., & Pei, B. (1997). The eects of decision consequences on auditors' reliance on decision aids in audit planning. Organizational Behavior and Human Decision Processes, 71, 211247. Bonner, S. E. (1994). A model of the eects of audit task complexity. Accounting, Organizations, and Society, 19, 213234. Bonner, S., & Pennington, N. (1991). Cognitive processes and knowledge as determinants of auditor expertise. Journal of Accounting Literature, 10, 143. Bonner, S., & Walker, P. E. (1994). The eects of instruction and experience on the acquisition of auditing knowledge. The Accounting Review, 69, 157178. Brecht, H., & Martin, M. P. (1996). Accounting information systems: The challenge of extending their scope to business and information strategy. Accounting Horizons, 10, 1622.

Chenhall, R. H., & Morris, D. (1986). The impact of structure, environment, and interdependence on the perceived usefulness of management accounting systems. The Accounting Review, 61, 1635. Chong, V. K. (1996). Management accounting systems, task uncertainty and managerial performance: A research note. Accounting, Organizations and Society, 21, 415421. Chu, P. C. (1991). A study of the inuence of a decision support aid on decision processes: Exploring the blackbox. Journal of Information Systems, 5, 117. Clemens, E. K., Thatcher, M. E., & Row, M. C. (1995). Identifying sources of reengineering failures: A study of the behavioral factors contributing to reengineering risks. Journal of Management Information Systems, 12, 936. Colantoni, C. S., Manes, R. P., & Whinston, A. (1971). A unied approach to the theory of accounting and information systems. The Accounting Review, 46, 90102. Cushing, B. E. (1990). Frameworks, paradigms, and scientic research in management information systems. Journal of Information Systems, 4, 3859. Davenport, T. H. (1993). Process improvement. Cambridge, MA: Harvard Business School Press. David, J. S. (1997). Three events that dene an REA methodology for systems analysis, design and implementation. Working paper, Arizona State University. Dunn, C. L., & McCarthy, W. E. (1997). The REA accounting model: Intellectual heritage and prospects for progress. Journal of Information Systems, 11, 3151. Einhorn, H. J., & Hogarth, R. M. (1981). Behavioral decision theory: Processes of judgment and choice. Journal of Accounting Research, 19, 131. Eining, M. M., Jones, D. R., & Loebbecke, J. K. (1997). Reliance on decision aids: An examination of auditors' assessment of management fraud. AuditingA Journal of Practice & Theory, 16, 119. Elliott, R. K. (1995). The future of assurance services: Implications for academia. Accounting Horizons, 9, 118127. Frances, J., & Garnsey, E. (1996). Supermarkets and suppliers in the United Kingdom: System integration, information and control. Accounting, Organizations and Society, 21, 591610. Fisher, J. (1995). Contingency-based research on management control systems: Categorization by level of complexity. Journal of Accounting Literature, 14, 2453. Gibbins, M., & Jamal, K. (1993). Problem-centered research and knowledge-based theory in the professional accounting setting. Accounting, Organizations and Society, 18, 451 466. Glover, S., Prawitt, D. F., & Spilker, B. C. (1997). The inuence of decision aids on user behavior: Implications for knowledge acquisition and inappropriate reliance. Organizational Behavior and Human Decision Processes, 72, 232 255. Godfrey, J. T., & Prince, T. R. (1971). The accounting model from an information systems perspective. The Accounting Review, 46, 7589. Goetz, B. E. (1939). What's wrong with accounting? Advanced Management, Fall, 151157.

330

E.G. Mauldin, L.V. Ruchala / Accounting, Organizations and Society 24 (1999) 317331 McCarthy, W. E. (1982). The REA accounting model: A generalized framework for accounting systems in a shared data environment. The Accounting Review, 57, 554578. McCarthy, W. E. (1987). Accounting information systems: Research directions and perspective. Journal of Information Systems, 2, 2932. Messier Jr., W. F. 1995. Research in and development of audit decision aids, in R. H. Ashton, & A. H. Ashton, (Eds.), Judgment and decision research in accounting and auditing (pp. 207228). New York, NY: Cambridge University Press. Mia, L., & Chenhall, R. (1994). The usefulness of management accounting systems, functional dierentiation, and managerial eectiveness. Accounting, Organizations and Society, 19, 113. Milgrom, P., & Roberts, J. (1995). Complementarities and t: strategy, structure, and organizational change in manufacturing. Journal of Accounting and Economics, 19, 179208. Nelson, M. W. (1993). The eects of error frequency and accounting knowledge on error diagnosis in analytical review. The Accounting Review, 68, 803824. Newell, A., & Simon, H. A. (1972). Human problem solving. Englewood Clis: Prentice Hall. Otley, D. T. (1980). The contingency theory of management accounting: achievement and prognosis. Accounting, Organizations and Society, 5, 413428. Payne, J. W., Bettman, J. R., & Johnson, E. J. (1993). The adaptive decision-maker. New York, NY: Cambridge University Press. Peters, J. M. (1993). Decision making, cognitive science and accounting: An overview of the intersection. Accounting, Organizations, and Society, 18, 383405. Piturro, M. (1998). Solutions in nance: The power triangle. Accounting Technology, Supplement, 28. Reneau, J. H., & Grabski, S. V. (1987). A review of research in computer-human interaction and individual dierences within a model for research in accounting information systems. Journal of Information Systems, 2, 3353. Schrader, W. J. (1962). An inductive approach to accounting theory. The Accounting Review, 37, 645649. Simon, H. A. (1981). The sciences of the articial. Boston, MA: MIT Press. Simon, H. A. (1997). Administrative behavior. New York, NY: The Free Press. Solomon, I., & Shields, M. D. (1995). Judgment and decision making research in auditing. In R. H. Ashton, & A. H. Ashton, (Eds.) Judgment and decision research in accounting and auditing (pp. 137175). New York, NY: Cambridge University Press. Sorter, G. (1969). An `events' approach to basic accounting theory. The Accounting Review, 44, 1219. Spicer, B., & Ballew, V. (1983). Management accounting systems and the economics of internal organizations. Accounting, Organizations and Society, 8, 7396. Stone, E. F., & Gueutal, H. G. (1985). An empirical derivation of the dimensions along which characteristics of jobs are perceived. Academy of Management Journal, 28, 376396.

Gorry, G. A., & Scott-Morton, M. S. (1971). A framework for management information systems.. Sloan Management Review, 13, 5570. Gul, F. (1991). The eects of management accounting systems and environmental uncertainty on small business managers' performance. Accounting and Business Research, 22, 5761. Gul, F., & Chia, Y. M. (1994). The eects of management accounting systems, perceived environmental uncertainty and decentralization on managerial performance: A test of three-way interaction. Accounting, Organizations and Society, 19, 413426. Hammer, M., & Champy, J. (1993). Reengineering the corporation: a manifesto for business revolution. New York, NY: HarperCollins. Haseman, W. D., & Whinston, A. B. (1976). Design of a multidimensional accounting system. The Accounting Review, 51, 6579. Hogarth, R. (1993). Accounting for decisions and decisions for accounting. Accounting, Organizations and Society, 18, 407 424. Hopwood, A. G. (1989). Behavioral accounting in retrospect and prospect. Behavioral Research in Accounting, 1, 122. Huber, G. P. (1983). Cognitive style as a basis for MIS and DSS design: Much ado about nothing?. Management Science, 29, 567579. Ives, B., Hamilton, S., & Davis, G. B. (1980). A framework for research in computer-based management information systems. Management Science, 26, 910934. Kiesler, S. (1986). Thinking ahead. Harvard Business Review, 64, 4660. Kim, K. K. (1988). Organizational coordination and performance in hospital accounting information systems: An empirical investigation. The Accounting Review, 63, 472489. Kottemann, J. E., Davis, F. D., & Remus, W. E. (1994). Computer-assisted decision making: performance, beliefs, and the illusion of control. Organizational Behavior and Human Decision Processes, 57, 2637. Libby, R., & Luft, J. (1993). Determinants of judgment performance in accounting settings: Ability, knowledge, motivation and environment. Accounting, Organizations and Society, 18, 425450. Lieberman, A. Z., & Whinston, A. B. (1975). A structuring of an events-accounting information system. The Accounting Review, 50, 246258. Magretta, J. (1998). The power of virtual integration: An interview with Dell Computer's Michael Dell. Harvard Business Review, 76, 7284. March, S. T., & Smith, G. F. (1995). Design and natural science research on information technology. Decision Support Systems, 15, 251266. Mason, R. O., & Mitro, L. L. (1973). A program for research on management information systems. Management Science, 19, 475587. Mauldin, E. G. (1999). The eects of information technology on incentive contracting, Working Paper, University of Missouri-Columbia.

E.G. Mauldin, L.V. Ruchala / Accounting, Organizations and Society 24 (1999) 317331 Sutton, S. G. (1992). Can we research a eld we cannot dene?. Advances in Accounting Information Systems, 1, 113. Tiessen, P., & Waterhouse, J. H. (1983). Towards a descriptive theory of management accounting. Accounting, Organizations and Society, 8, 251267. Van Alstyne, M., Brynjolfsson, E., & and Madnick, S. (1995). Why not one big database? Principles for data ownership. Decision Support Systems, 15, 267284. Waller, W., & Jiambalvo, J. (1984). The use of normative models in human information processing research in accounting. Journal of Accounting Literature, 3, 201223.

331

Waterhouse, J. H., & Tiessen, P. (1978). A contingency framework for management accounting systems research. Accounting, Organizations and Society, 3, 6576. Weber, R. (1987). Toward a theory of artifacts: A paradigmatic base for information systems research. Journal of Information Systems, 1, 319. Widrow, B., Rumelhart, D. E., & Lehr, M. A. (1994). Neural networks: Applications in industry, business, and science. Communications of the ACM, 37, 93105. Williamson, O. E (1985). The economic institutions of capitalism. New York, NY: The Free Press.