Вы находитесь на странице: 1из 12
INTERNATIONAL JOURNAL OF ACCOUNTING INFORMATION SYSTEMS International Journal of Accounting Information Systems 1 (2000) 54-65 NORTH-HOLLAND Incorporating Information Technology Considerations Into an Expanded Model of Judgment and Decision Making in Accounting Tanya L. Benford**, James E. Hunton "Texas A&M University, College Station, Texas, USA “University of South Florida, Tampa, Florida, USA Manuscript received September 1, 1998; revised manuscript May 1, 1999 Abstract Due to an increasingly sophisticated economic environment, accounting tasks and supporting information technology are becoming more complex. Under these circumstances, achieving congruence between task require- ‘ments and computer-supported information systems (i., task/technology fit) increases in importance, However, the layering of task and technology complexities may impose inordinately high mental workloads on individual decision-makers, thereby detracting from performance. Thus, judiciously designing computer-supported IS to si multaneously optimize task/technology fit, and minimize mental workload, is critical While a significant amount of research in accounting has examined the individual determinants of judgment and decision making (Libby and Luft, 1993), only recently has this line of inquiry been extended to incorporate the impacts of task and technology on performance (Amold and Sutton, 1998). The theoretical model proposed in this article complements and extends this research thrust by incorporating the concepts of task/technology fit (Goodhue and Thompson, 1995) and mental workload (Jex, 1988; Vidulich, 1988) into an expanded set of deter- minants of judgment and decision making. The proposed research model has theoretical and practical implica- tions in a variety of accounting-related decision domains. © 2000 Elsevier Science Inc. ‘Keywords: Computer supported information systems; Cognitive complexity; Mental workload: Task technology fit; Judgment and decision makers Introductio non human decision processes. When developing such in- formation systems, designers must be cognizant of the importance of achieving congruence between task requirements and technology functionality (task/ Globalization of economic commerce has led to an increasingly complex. business environment. In response, many organizations have substantially ex- panded the amount of information they collect, inte- grate, and analyze in an effort to generate high- quality decisions. Given the dynamic nature of the global environment and the escalating volume of in- formation-processing activities in economic entities, the requirements and complexity of accounting-related tasks are in a constant state of transformation, One method of dealing with changing task requirements and increasing task complexity is to integrate the use of computer-supported information systems (IS) into technology fit). While a significant amount of research in account- ing has examined the impact of task characteristics on judgment and decision-making performance (Libby 1995), only recently has this line of inquiry been ex- tended to incorporate the impact of task and technol- ‘ogy on performance (Arnold and Sutton, 1998). Com- plementary IS research (Goodhue and ‘Thompson, 1995; Goodhue, 1995) suggests that increases in task/ technology fit are positively associated with human * Corresponding author. Tel: (409) 845-5014; Fax: (409) 845-0028; E-mail: tbenford@coba.usf.edu 1467-0895/00/S ~ see front matter © 2000 Elsevier Science Inc. All rights reserved. PII: $1467-0895(99)00004-4 Benford and Hunton / International Journal of Accounting Information Systems 1 (2000) 54-65 55 performance. The IS research argues that technology functionality must be designed to match task requit ments, and that the proper alignment of technology functionality and task requirements is a key factor as- sociated with performance. However, prior research has not considered the simultaneous effects of increas- ing task and technology complexity on cognitive pr cessing activities associated with judgment and deci- sion making. ‘The central premise of this conceptual article is that the layering of task and technology complexities can impose high mental workloads on individual de- cision-makers, and these higher mental workloads can negatively impact performance. Thus, the possi- bility exists that the positive impact of task/technol- ogy fit may be more than offset by the negative effect of increased mental workloads. Because individuals possess limited cognitive processing resources (Si- mon, 1990), they often employ simplifying tech- niques, such as cue recognition and satisficing heu- ristics, in the face of increasing complexity. Accordingly, the layering of task and technology ‘complexities can unduly increase mental workloads to the point where decision makers will resort to higher-order approximations of decision solutions, which can lead to a deterioration of performance. Hence, the designers of computer-supported IS should attempt to simultaneously accomplish two ob- jectives: optimize task/technology fit, and minimize ‘mental workload. ‘The purpose of this article is to develop a theory of performance that explicitly considers the environ- mental impact of task/technology fit. Accordingly, the theoretical model presented expands the set of in- dividual determinants of judgment and decision mak- ing in accounting to include task/technology fit and ‘mental workload. The expanded model integrates de- cision theories in accounting (Libby and Luft, 1993; Libby, 1995) and cognitive psychology (Einhorn and Hogarth, 1981; Jex, 1988; Vidulich, 1988; Simon, 1990; Payne et al., 1993) with system performance theories from the 1S literature (Benbasat and Dexter, 1985; Vessey and Galletta, 1991; Cheney et al., 1986; Goodhue and Thompson, 1995). As a result, accounting researchers and decision-support system designers are provided with a more comprehensive iew of individual decision-making processes—a view that incorporates the impact of task/technology fit and mental workload on performance. The model presented herein is applicable to many research streams within accounting, such as designing com- puter-supported IS, auditing financial records, plan- ning tax strategies, budgeting capital expenditures, and providing assurance services. ‘The next section reviews relevant research in ac- counting, information systems, and cognitive psy- chology, and defines the variables used in the current theoretical model. This overview is followed by a proposed theory of performance that explicitly incor- porates task/technology fit and mental workload. The article concludes with a discussion of possible re- search questions, inherent model implications, and potential theoretical extensions. An example of how ‘one might test the theoretical model developed in this article is presented in Appendix A. Individual Determinants of Performance ‘The Libby and Luft (1993) model of judgment and decision making, which evolved from prior re- search in both cognitive psychology and auditing (Einhorn and Hogarth, 1981; Libby, 195), is the pre- dominant theoretical model currently used in ac- counting research. The model asserts that knowledge is a function of experience and ability, and that per- formance is a function of knowledge and ability. Al- though most accounting judgment and decision-mak- ing research has focused on auditing tasks, the Libby and Luft model can be generalized beyond the audit context (Hunton et al., 1998). Hence, the Libby and Luft model forms the core of the theoretical research model developed in this article (Figure 1). The fol- Jowing subsections review literature relevant to the development of model propositions regarding task experience, task knowledge, ability, and motivation. Defining Experience and Knowledge Early studies in auditing used experience as a sur- rogate for knowledge (Frederick, 1991; Bonner, 1990), In a more complete conceptualization of hu- man decision making, Libby (1995) categorized ex- perience as either firsthand (e.g., task completion, re- viewing work of others, process feedback, and ‘outcome feedback) or secondhand (e.g., education, training, reading audit guides, and discussion with colleagues). Libby recognized that experience is not a suitable proxy for knowledge; rather, experience begets knowledge. In particular, task-specific experi- ence yields task-specific knowledge. This relation- ship has been widely accepted in judgment and deci- sion-making research in accounting and cognitive psychology. Prior research on auditor expertise has paid par- ticular attention to two aspects of knowledge: (1) knowledge content—what information is encoded in memory (Bonner, 1990; Bonner and Walker, 1994; Nelson, 1993; Kaplan et al., 1992; Tan and Libby, 56 Benford and Hunton / international Journal of Accounting Information Systems 1 (2000) 54-65 ea, Technology Functionality Technology Experience Task/Technology Fit ‘Motivation Pit PX Proposition X Figure 1. Research Model 1997) and (2) knowledge organization—how infor- mation is encoded and structured in memory (Bedard and Mock, 1992; Choo and Trotman, 1991; Frederick, 1991), Extant research on knowledge organization suggests that experts tend to organize information schematically, while novices tend to organize infor- ‘mation taxonomically (Choo and Trotman, 1991; Fre= derick, 1991; Bedard and Mock, 1992). Since experts organize information differently than novices, they also tend to employ different search strategies (Choo and Trotman, 1991; Hunton and McEwen, 1997), Re- search suggests that a mismatch between task organi- zation and knowledge organization may lead to a de- crease in performance (Nelson et al., 1995). Knowledge content has been defined as informa- tion encoded in memory, where memory is either ep- isodic (the memory of particular experiences or facts) ‘or semantic (understanding the meanings of con- cepts) (Tulving, 1972; Libby. 1995). For instance, the definition of a particular financial ratio would be in episodic memory, while the relation between pos- sible errors and omissions in the financial data and a fluctuation in a particular financial ratio would be in semantic memory. Knowledge content may consist of world know!- edge, general domain knowledge, or task-specific knowledge (Bonner and Lewis, 1990). The specific task to be performed will dictate the importance of, each of these knowledge subsets. Each knowledge subset consists of declarative and procedural know!- edge. Bonner and Walker (1994) suggest that dectar- ative knowledge (i.e., specific factual items of infor- mation) is acquired prior to procedural knowledge. Declarative knowledge may be learned through read- ing, education, and training. Procedural knowledge (i.e., how one applies factual information items in a given circumstance) is acquired primarily through practice and feedback. Thus, firsthand experiences are a primary determinant of procedural knowledge and secondhand experiences are a primary determi- nant of declarative knowledge. Therefore, knowledge may be thought of as a combination of declarative and procedural information that is imbedded in an individual's schematic or taxonomic information structure, Defining Ability Ability is a determinant of an individual's capac- ity to perform physical or mental tasks. Performance of mental tasks, or thinking capacity, is a function of both skill (ability) and knowledge (Simon, 1990). Ability is defined by Campbell and Campbell (1988, 89) as “stable characteristics of individual differ ences, such as general verbal ability, mathematical aptitude, eye-hand coordination, or upper body strength, that make it possible for people to learn or Benford and Hunton / International Journal of Accounting Information Systems 1 (2000) 54-65 37 execute certain cognitive, psychomotor, or physical behaviors.” Thus cognitive ability may be thought of as the skill necessary to perform a mental task, which is consistent with Libby’s (1995) definition of ability as an individual's innate capacity to complete infor- mation-processing tasks (encoding, retrieval, analy- sis, ete.) that contribute to problem solving, I is important to distinguish between knowledge, which may be increased through experience (c.g., task completion, feedback, training, etc.), and an in- dividual’s innate capacity to lear and perform, which is not generally considered a function of expe- rience. Prior research has used problem-solving skills as a surrogate for ability (Bonner and Lewis, 1990; Libby and Tan, 1994; Hunton et al., 1998). However, whether problem-solving skills are a suitable proxy for ability is questionable. Problem-solving_ skills may not adequately reflect an individual's ability to earn, as such skills are confounded with his/her knowledge and experience, In the current context we are interested in an individual's innate ability to inte- grate complex information sets, which is essential for successful performance in a complex task environ- ment (Schroder et al., 1967). We argue that this di mension of ability is represented by cognitive com- plexity, rather than problem-solving skills, Sternberg (1994, 286) defines cognitive complex- ity as a concept “that deals with the ways people ap- ply the intellectual power that they possess . . . a per- sonality factor that varies from one person to the next but is stable within at least adult individuals.” Cogni- tive complexity is consistent with the Campbell and Campbell definition of ability as a measure of the ca- pacity to leam or perform a mental task. The cogni- tive complexity dimension of ability also coincides with Libby and Luft’s (1993) conceptualization (Ta- ble 1). Cognitive complexity can affect the acquisi- tion of declarative knowledge and the application of procedural knowledge. Additionally, cognitive com- Table 1 Individual Determinants of Performance (from Libby 1995) plexity can directly impact human performance (Schroder et al., 1967). Using cognitive-complexity ability rather than problem-solving skills is an impor- tant issue to researchers who use the theoretical model developed in this article, as cognitive com- plexity provides a higher degree of independence among the experience, knowledge, and ability constructs. Defining Motivation Motivation is comprised of three factors: the choice to perform a given task, the level of effort ap- plied in the performance of the task, and the persis- tence of effort in the performance of the task (Camp- bell and Campbell, 1988; IIgen and Klein, 1988). Given that an individual chooses to perform a task, in the absence of environmental and behavioral con- straints the level of effort and persistence can com- pensate, in some instances, for less knowledge and ability (Libby, 1995), Motivation may be either intrinsic or extrinsic. In- trinsic motivation reflects the effort expended to sat- isfy an internal need, while extrinsic motivation re- flects the effort expended in response to external incentives such as money (Pinder, 1984). Deci (1975) identifies three general approaches to intrinsic moti- vation: the individual's desired level of arousal (stim- ulation to the brain and central nervous system), the individual's desired level of uncertainty, and the indi- vidual’s desire to be competent. Although these ap- proaches may appear disparate, Deci (1975) suggests that they are in fact consistent, and that individuals will only feel competent when they are able to seek out challenges (stimulation) and manage uncertainty. ‘There are many theoretical approaches to intrinsic and extrinsic motivation, such as expectancy theory (Vroom, 1964; Naylor et al., 1980), equity theory (Ad- ams, 1963; Carrell and Dittrich, 1978), goal-setting theory (Locke et al., 1981), cognitive evaluation the- Experiences—task-related opportunities for learning, + Firsthand experiences include task completion, reviewing the work of others, process feedback, and outcome feedback. + Secondhand experiences include education, training, reading relevant literature, and discussions with colleagues. Knowledge—declarative or procedural inform: encoded in memory where memory cither episodic or semantic. + Knowledge subsets include world knowledge, general domain knowledge, or task specific knowledge. * Episodic memory is the memory of specific facts. + Semantic memory is understanding the meaning of concepts. Abilities—an individual's innate capacity to perform mental-processing tasks (encoding, retrieval, analysis, etc.) that contributes to problem solving, * Cognitive complexity —how indi iduals apply their intellectual power (Stemberg 1994). Motivation—the choice to perform a task, the level of effort applied in the performance of the task, and the persistence of effort. 58 Benford and Hunton / International Journal of Accounting Information Systems 1 (2000) 54-65 ory (White, 1959; DeCharms, 1968; Deci, 1975), so- cial learning theory (Bandura, 1977), attribution the- ory (McMahan, 1973), and schema theory (Barlett, 1932; Neisser, 1976; Markus and Zajonc, 1985). All of these theories examine factors that influence an in- dividual’s choice to perform a target behavior, his! her tevel of effort, and his/her degree of persistence. Each of these theoretical approaches to motivation consistent with the notion that motivation can moder ate human performance (Libby, 1995). The indi ual determinants of performance, discussed in this section, are summarized in Table 1. Environmental Determinants of Performance Libby and Luft (1993) theorized that environmen- tal factors such as time pressure, task structure, countability, and technology. can influence the level of experience, ability, and knowledge necessary to ield high-quality decisions. While prior behavioral esearch in accounting has examined the impact of several environmental factors on judgment and dei sion making in accounting (Kennedy, 1995; Cushing and Ahlawat, 1996; Cuccia et al., 1995; McDaniel, 1990; Libby and Tan, 1994), the influences of infor- mation technology and cognitive processing limita- tions have not been fully investigated, Accordingly, this article considers two key environmental factors; task/technology fit and mental workload. Defining Task/Technology Fit Extant IS research has proposed several theoreti- cal models linking tasks, information technology, and performance. For example, some theoretical models suggest that cognitive fit is a primary deter minant of system performance (Benbasat and Dex- ter, 1985; Vessey, 1991; Vessey and Galletta, 1991), while other models posit that system performance is 4 function of utilization (Cheney et al., 1986). A task/ technology fit model has evolved from these fit and utilization theories (Goodhue, 1995; Goodhue and Thompson, 1995). Each of these theoretical perspec tives considers the impact on performance of the in- teraction between the individual and information technology. The following subsections review rele- vant IS literature and consider theoretical implic tions for the research model developed in this article Theories of Cognitive Fit. Vessey (1991) proposes a theory of cognitive fit that suggests the following: “For most effective and effi- cient problem solving to occur, the problem represen tation and any tools or aids employed should all sup- port the strategies (methods or processes) required to perform the task” (Vessey and Galletta, 1991, 64), For the most part, cognitive fit research (Benbasat and Dexter, 1985; Vessey and Galletta, 1991) has fo- cused on information presentation and the associated system design implications. Early studies on cogni- tive fit did not examine complex judgment and deci- sion-making tasks, such as those found in auditing and other accounting scenarios. Rather, these early studies usually considered the impact of data repre- sentation (graphs versus tables) on an individual's judgment and decision-making performance. Vesey and Galletta (1991, 65) noted that “. .. simply match- ing task to technology is insufficient to achieve the desired effects, and that the user must also use appro- priate processes, and thus develop appropriate mental representations, for performance effects to occur.” Building on this observation, Goodhue and Thomp- son (1995) incorporated cognitive fit theory into a task-to-performance chain (TPC). Task-to-Performance Chain. Task-to-performance chain theory proposes relation- ships among technology functionality, technology experience, and task requirements. Technology func- tionality refers to the function the system serves in the organization. Markus (1984) identifies the fol- lowing five primary functions of systems: (1) to structure work (operational systems); (2) to evaluate performance and motivate people (monitoring and control systems); (3) to support intellectual processes (planning and decision systems); (4) to augment hu- man communication (communication systems); and (5) to facilitate interorganizational transactions (interor- ganizational systems), Markus notes that these pri- mary systems may be combined to form a compos system with two or more subsystems. For example, a CPA firm's audit IS could serve to organize audit work papers (structure work); monitor and control the audit process; model client processes or simulate expert (auditor) analysis; and augment communica- tion between members of the audit team. The research model developed in this article fo- euses on IS that use technology to aid managers in performing judgment and decision-making tasks. These systems do not solve problems, rather they provide models and reports that enable decision mak ers 10 expand their scope of analysis (Markus, 1984). In addition to technology functionality, TPC also in- corporates the impact of task characteristics, ‘Wood (1986) postulates that the essential compo- nents of a task are inputs (required acts and informa- tion cues), and outputs (e.g., variance analysis, griev- ance settlement). Required acts (task requirements) Benford and Hunton / Intemational Journal of Accounting Information Systems I (2000) 54-65 39 are the specific processes or activities performed by an individual in order to complete the task. Informa- tion cues are the pieces of information used to per- form the task. The task output, or product, determines the appropriate task inputs. Therefore, a change in task inputs yields a new task. Task inputs also deter- mine the nature and scope of resources (e.g., knowl edge, ability, and tools) necessary for task perfor- mance. That the use of particular tools is, by definition, also a task, is an important concept. For stance, the task may be to use a particular com- puter-supported information system to produce a spe- cific report Another important aspect of individual decision- making involves mental workload. As accounting- related tasks and supporting information technolo- gies become more complex, an individual's cognitive resources may be pushed to their limits. The follow ing section addresses this key concern. Defining Mental Workload Both knowledge and ability are subject to limita tions of processing power and processing speed (Si- mon 1990). Simon suggests that individuals are sub- ject to a variety of performance limitations, therefore they often employ simplifying problem-solving tech- ‘niques—such as pattern recognition, schematic orga- nizing, decision heuristics, and cue recognition—to arrive at approximate solutions. Simon (1990, 6) notes that, “Because of the limits on their computing Table 2 Selected Environmental Determinants of Performance speeds and power, intelligent systems must use ap- proximate methods to handle most tasks. Their ratio- nality is bounded.” Indeed, as problems become more complex, individuals will often satisfice, thus sacrificing one performance measure for another (c.g., accuracy for speed) in an effort to find a sat factory approximate solution, Higher-order approxi- mations tend to be less thorough and accurate than lower-order approximations; thus, there tends to be @ negative relationship between higher-order approxi- mations and performance. According to Simon, when given a particular task to perform, an individual's capacity for thought is a function of his/her stored knowledge and innate abil- ity (ie., processing speed and power). Therefore, the knowledge and ability needed to solve a simple prob- Jem would be less than that needed to solve a more complex problem. As tasks become more complex (see Table 2 for a definition of complexity), the cog- nitive resources (knowledge and ability) needed for problem solving would be expected to increase. Mental workload is an individual’s evaluation of their attentional load margin (Jex, 1988). That is, mental workload is the individual's assessment of the difference between their capacity for performance and the demands of the task. As task demands in- crease, individuals will perceive increases in their mental workload, and excessive mental workload is associated with suboptimal performance (Vidulich, 1988). The environmental determinants of perfor- ‘Technology functionality —the function(s) a system serves in an organization. Two or more subsystems may be combined to form ‘composite systems (Markus 1984). + Operational systems structure work. ‘+ Monitoring and control systems evaluate performance and motivate people. + Planning and decision systems support intellectual processes. + Communication systems augment human communication + Inter-organizational systems facilitate inte-organizatioal transactions. “Task requirements—specific processes or activities performed by an individual in order to complete a task (Wood 1986). ‘Task/technology fit—the correspondence between task requirements, technology functionality. technology experience (Goodhue and Thompson 1995) and task knowledge. ‘Complexity —deseribes relationships between inputs, and places cognitive demands on individuals (Wood 1986), + Component complexity isthe number of distnet (non-redundant) inputs (information cues and acts) necessary for task performance. + Coordinative complexity is a measure of the relationship between task products and inputs including: (1) the number of acts which must be performed in sequence, (2) the degree to which there isa linea relationship between the frequency or intensity of the act and the task product, and (3) the frequency with which a task input is associated witha task product, + Dynamic complexity reflects the magnitude, the pattern, and the predictability of intertemporal changes in component and coordinative complexity. Thus the greater the intertemporal association (i... the less change overtime), the less complex the task. Mental workload—the individual's evaluation oftheir attentional load margin, which is the difference between their motivated capacity and the demands of the tsk (Jex 1988); verbal reports ofthe information load experienced (Vidulich 1988). 60 Benford and Hunton / International Journal of Accounting Information Systems 1 (2000) 54-65 mance, discussed in this section, are summarized in Table 2, An Expanded Model of Judgment and Decision Making in Accounting The first five propositions presented in the theo- retical model (see Figure 1) reflect the Libby and Luft (1993), Libby (1995) theoretical framework, which forms the core of the expanded model of judg ment and decision making in accounting. The core (baseline model) is extended to reflect: (1) the pro- posed relationship between motivation and ability, (2) the proposed relationship between the baseline model and task/technology fit and, (3) the proposed relationship between the baseline model and mental workload. Baseline Model of Judgment and Decision Making in Accounting Many studies of expertise have focused on the re- lationship between experience, knowledge and expert performance (Bedard, 1989; Bedard and Chi, 1993; Bedard and Mock, 1992; Bonner and Walker, 1994; Davis, 1996; Libby and Tan, 1994; Bonner and Lewis, 1990; Bonner, 1990; Choo and Trotman, 1991; Hunton et al., 1997). These studies suggest, consistent with the Libby and Luft (1993) theoretical framework, that knowledge is both a function of experience and a determinant of human performance, thus the follow- ing relationships are proposed (Figure 1): Proposition 1 (P1): Task experience is positively associated with task knowledge. Proposition 2 (P2): Task knowledge is positively associated with performance. Ability is a multifaceted construct that represents ‘an individual's capacity to perform mental or physi- cal tasks, Prior research examining the link between ability and performance has yielded mixed results (Libby and Tan, 1994; Hunton and McEwen, 1997), However, in these studies, ability was operational- ized using a measure of the individual's problem- solving skills. If the participants had prior experience with problems similar to the metric employed in th study, the results imay have been confounded. Al- though there are many different ability traits (Pinder, 1984), consistent with Schroder et al. (1967), the fo- cus in this model is on the individual’s innate capac- ity to perform complex information-processing tasks. Cognitive complexity reflects the individual's ability to integrate and differentiate complex information sets. This suggests that ability (cognitive complexity) impacts how individuals organize knowledge as well as the search strategies employed in task perfor- mance (Frederick, 1991; Nelson et al., 1995; Choo and Trotman, 1991; Hunton and McEwen, 1997). In addition, cognitive complexity is considered stable (not confounded by experience or knowledge) among adults (Sternberg, 1994), Based on the discussion just presented, the following relationships are proposed (see Figure 1): Proposition 3 (P3): Ability is positively associ- ated with task knowledge Proposition 4 (P4): Ability is positively associ- ated with performance. Motivation is also generally considered to be a determinant of performance (Pinder, 1984). Individu- als that possess large quantities of knowledge and/or ability but who are not sufficiently motivated to bring those resources to bear on the task, may not perform as well as individuals who are highly motivated but possess lesser knowledge and/or ability. This sug- gests that, given the same level of knowledge and ability, more highly motivated individuals would be expected to outperform their less-motivated counter- parts. Thus, the following propositions are offered (Figure 1): Proposition 5 (5): Motivation moderates the im- pact of task knowledge on performance. Proposition 6 (P6): Motivation moderates the im- pact of ability on performance. Extensions to the Baseline Model of Judgment and Decision Making in Accounting ‘Task-to-performance chain theory states that hu- man performance is optimized when there is a proper matching among task requirements, technology func- tionality, and experience using the technology. This three-way dependency yields a construct termed Task/Technology Fit (TTF) and, according to TPC theory, TTF directly impacts performance. However, Libby and Luft (1993) suggest that a match between ask requirements and task knowledge is also neces- sary for high-quality decision making. Thus we ex- pand the determinants of task/technology fit to in- clude task knowledge. This suggests the following propositions (Figure 1): Proposition 7 (P7): Task requirements, task knowledge, technology functionality, and technology experience are positively associ: ated with task/technology fit, Benford and Hunton / International Journal of Accounting Information Systems 1 (2000) 54-65 61 Proposition 8 (P8): Taskitechnology fit is posi- tively associated with performance. While extant accounting research has not incorpo- rated task/technology fit concepts into decision-mak- ing models, neither has IS research explicitly consid- ered the impact of task-specific knowledge on performance. The research model presented in this paper integrates both concepts to form a more com- plete view of individual decision-making processes when computer-supported information systems are employed. Given the same level of task complexity, individuals possessing higher levels of cognitive resources will likely experience less mental workload than individu- als possessing lower cognitive resource levels. This is consistent with the U-curve hypothesis (Schroder et al. 1967), which suggests that an individual’s level of in- formation processing will decrease as environmental complexity becomes either extremely high or ex- tremely low. Thus, we would expect that as mental workload becomes highly taxed, an individual's infor- ‘mation processing will decrease, In a decision environment requiring the use of computer-supported information systems, mental workload can become further taxed. That is, to the extent that supporting information technology is more complex for individuals to use, the level of cognitive resources required will be correspondingly higher. Holding technological complexity constant, an indi- vidual’s experience using a computer-supported infor- ‘mation system can reduce mental workload. Overall, the concem is that increased technological complex- ity can result in higher levels of mental workload and satisficing, thereby deteriorating performance. The issue of technological complexity is critically impor- tant to developers of computer-supported IS, for they should strive to minimize the apparent complexity of information technology to users. The layering of both task and technology com- plexity can impose inordinately high levels of mental workload on individual decision-makers. However, holding task and technology complexity constant, mental workload can vary from individual to individ- ual, depending on their experience using the support- ing technology, requisite task knowledge, and innate ability (e.g., cognitive complexity). The issue of mental workload is incorporated into the research ‘model and the following propositions are set forth (Figure 1): Proposition 9 (P9): Task and technology com- plexities are positively associated with mental workload. Proposition 10 (P10): Technology experience, task knowledge, and ability are negatively as- sociated with mental workload Extant research in cognitive psychology (Simon, 1990; Payne et al., 1993) and auditing (Bonner, 1994) suggests that as mental workload increases there is often a corresponding decrease in performance, The theory of cognitive effort and accuracy (Payne et al., 1993) assumes that individuals desire to conserve their cognitive resources, yet maximize the accuracy of their decision. This theory posits that individuals’ decision strategies will change (e.g., trading accuracy for speed) as their mental workload increases or de- creases. Thus mental workload (which is a function of complexity and individual differences) is also a determinant of performance, which leads to the final proposition (Figure 1): Proposition 11 (PII): Mental workload is nega- tively associated with performance, The following section presents possible research questions Research Questions In the proposed research model, task knowledge, ability, task/technology fit, and mental workload all directly influence performance. The latter two con- structs expand the Libby and Luft model of the deter- ‘minants of judgment performance in accounting set- tings, The proposed model also incorporates. the moderating influence of motivation on performance. The central research question focuses on whether the expanded theoretical model presented in this article significantly improves the explanatory power of the baseline Libby and Luft model in situations where computer-supported IS are used in arriving at deci- sions. Additionally, when computer-supported IS are employed, is the explanatory power of the baseline model significantly enhanced by including: motivation only, task/technology fit only, ‘mental workload only, motivation and task/technology fit, motivation and mental workload, task/technology fit and mental workload, and/or 7. motivation, task technology fit, and mental workload? Aube Future studies could focus on answering these re- search questions in an effort.to find the most parsi- 2 Benford and Hunton / International Journal of Accounting Information Systems 1 (2000) 54-65 monious model that best represents individual deter- minants of performance in accounting where information technology is used to enable and support the decision-making process. ‘The model presented in this article can be applied to a variety of accounting-related situations, such as planning tax strategies, budgeting capital expendi- tures, and providing assurance services. In Appendix ‘A, an auditing context has been selected to illustrate how researchers might operationalize the research ‘model. While research designs could employ surveys and field studies to test the model's generalizability, this example describes how an experiment could be used to assess the internal validity of the research model and to compare proposed models to the base- ine model (Libby and Luft, 1993). The researchers recognize that there are many possible experimental treatments and conditions that can be developed to test the research model. In Appendix A, we describe one possible experimental scenario. Discussion The environment of economic commerce across the globe is becoming increasingly complex and de- pendent on information technology. As a result, ac- counting-related task requirements are rapidly evolv- ing and task complexity is increasing. Under these circumstances, there are many decision-making situ- ations where computer-supported IS are a necessary part of decision processes. Heretofore, cognitive de- cision-making models in accounting have not ade- quately addressed the impact of computer-supported IS on performance. Accordingly, the objective of t article is to develop a theoretical model of individual decision-making applicable to situations where com- puter-supported IS are used as an integral part of the decision process. ‘The research model developed in this article com- plements and extends a theoretical model of judg- ‘ment and decision making in accounting provided by Libby and Luft (1993). The Libby and Luft model suggests that both experience and ability determine knowledge, and both knowledge and ability deter- mine performance. Their model forms the core of the research model proposed in this paper. The Libby and Luft model is extended by explicitly including the constructs of motivation, task/technology fit, and mental workload into an expanded set of individual determinants of performance. The impact of motivation on performance has been discussed by Libby (1995). Libby encapsulates motivation in his conceptualization of knowledge and performance, noting that motivation is an inter- vening variable that can impact the strength, but not the direction, of a relationship. The proposed model, consistent with Libby's conceptualization, considers motivation to be an important construct that is inde- pendent from knowledge. Hence, it is explicitly rep- resented as a moderator between task knowledge and performance, as well as between ability and perfor- mance. Inclusion of task/technology fit in the re~ search model is derived from theory presented in the IS literature. Goodhue and Thompson (1995) devel- oped a theoretical model that explicitly considered the relationship between task requirements and tech- nology functionality, and the joint impact of task and technology on human performance. Mental workload theory (Jex, 1988: Vidulich, 1988) is also incorporated into the research model. Cognitive theory asserts that cognitive information-processing capacity is limited, and that individuals will often sacrifice performance when mental workloads become too burdensome. Task/technology fit and mental workload consider ations are incorporated into the baseline model, as these constructs are salient factors to consider when the decision environment requires the use of a com- puter-supported information system. Implications for accounting education and prac- tice include training and systems design issues. Infor- mation technology is having an enormous impact on society in general, and on accounting in particular. As information technology is more deeply integrated into accounting practice, understanding the funda- mental determinants of performance is an important prerequisite to developing effective training pro- grams. Designers of computer-supported IS can also benefit from a more comprehensive model. Fo stance, systems designers should focus on how to de- velop computer-supported IS capable of handling increasingly complex information collection, integra- tion, and analyses, while making such complexity transparent to users, Attending to this critical issue will help to increase technology functionality and re- duce apparent technological complexity. AS a result, designers can help to maximize task/technology fit and minimize mental workload. In addition, the research model presented in this article can be applied to new areas of assurance s vices, such as providing assurance over the reliability of IS. Based on this model, one can argue that the de- sign of computer-supported IS used by organiza- tional managers can impact the quality of their de~ cisions and, by extension, the efficacy of the organization as a whole. Accountants can use the re- search model as a basis for providing IS reliability assurance over the design, function, and use of com- puter-supported IS in organizations, Specifically. a Benford and Hunton / International Journal of Accounting Information Systems 1 (2000) 54-65 6 countants can develop assurance procedures de- signed to assess the extent to which computer- supported IS are used in decision making, the degree of matching between task requirements and function- ality of the computer-supported IS, and the perceived complexity of using the computerized aids. Only recently have psychology-based models of accounting judgment and decision making incorpo- rated critical information technology factors, such as task/technology fit, technology experience, and tech- nology complexity. For instance, the theory of tech- nology dominance (TTD) offered by Arnold and Sutton (1998) posits that task experience, task com- plexity, decision aid familiarity, and cognitive fit are key factors influencing a user's reliance on an in- telligent system. These factors are consistent with many of the constructs examined in the current re- search model. While the TTD examines an individ- ual’s reliance on intelligent systems, the proposed re- search model focuses on performance. Thus, it is not surprising that some of the proposed relationships in the TTD framework differ from the current model. For example, the TTD suggests that a low level of task experience is positively associated with reliance on an intelligent system, whereas the current model posits that a low level of task experience is negatively related to performance. Similarities and differences such as those just noted suggest that addi- tional research endeavors might attempt to integrate both the use of and reliance on information technol- ogy into more complete and internally consistent framework. Appendix A Experimental Example The example set forth in this appendix proposes a two (high and low task/technology fit) by two (high and low mental workload) fixed-factor, between-sub- jects experimental design. Task/technology fit would be manipulated by providing subjects with a rela- tively large and diverse set of task requirements aimed at performing a preliminary analytical review for a new audit client (a constant condition). Subjects would be required to use a computer-supported infor- mation system to conduct their analytical review. One version of the support aid would be improperly matched to the task requirements. While providing a computer-supported IS that does not adequately sup- port task requirements may appear nonsensical, in many instances improper matching exists in the workplace (e.g., a task that requires analysis in Euro- dollars and a computer-supported IS that is unable to handle Eurodollar translations and calculations). A second version of the experimental computer-sup- ported IS would be properly matched to the task at hand (e.g., providing a computer-supported IS with the capability of Eurodollar translations and calculations Mental workload could be manipulated by creat- ing a relatively complex task environment for sub- jects, such as the preliminary analytical review, and holding task complexity constant across. subjects. Technological complexity could then be manipulated as either high or low. In the high condition, the com- puter-supported IS would be relatively difficult to use and in the low condition, the support aid would be relatively easy to use. For instance, many com- puter-supported IS provide general purpose tools, such as spreadsheets or databases. Technological complexity could be manipulated by varying the number of macros and templates available to the user ie., the less complex computer-supported IS would have more built in macros and templates). As a re- sult, mental workload would be relatively higher for one group of subjects as compared with the other group. Participants would be chosen to ensure sufficient variation in task experience, task knowledge, and technology experience. This could be achieved by so- liciting both student subjects and professional audi- tors as experimental participants. Individual differ- ences would be assessed via postexperimental survey questions, Ability would be measured using stan- dardized psychological evaluations of cognitive com- plexity. Motivation would be aroused by using an in- centive scheme, and measured after the experimental task is completed. Participants would be randomly assigned to one of four conditions: (1) high task/technology fit and high mental workload, (2) high task/technology fit and low mental workload, (3) low task/technology fit and high mental workload, or (4) low task/technol- ogy fit and low mental workload. Manipulation check questions would be used to assess the subjects’ perceived levels of task/technology fit and mental workload. The constructs depicted in the research model are latent and as such, are not directly observable. Thus, these constructs must be inferred through the use of measured variables. Structural equation modeling (SEM) is deemed an appropriate method of data anal- ysis in this proposed study for several reasons. First, SEM permits simultaneous consideration of multiple unobserved theoretical relationships and “is particu- larly useful when one dependent variable becomes an independent variable in subsequent dependency rela- tionships” (Hair et al., 1995, 617), as is the case with oF Benford and Hunton / International Journal of Accounting Information Systems 1 (2000) 54-65 the proposed model (Figure 1). SEM improves statis- tical estimation by considering the correlation attrib- utable to the reliability of the predictor (measured) variable. Thus, SEM permits explicit consideration of measurement error in the estimation process. Fi nally, SEM permits evaluation of the overall model fit, measurement model fit, and structural model fit, as well as comparison of competing models. SEM as- sumes that all relationships are linear, all observa- tions are independent, and that there is random sam- pling of the participants (Hair et al., 1995). In addition, SEM is sensitive to multivariate normality, although, EQS (Bentler, 1985) incorporates elliptical and arbitrary distribution theories and is fairly robust to violations of normality, which permits modeling of data that has any distributional form (Bentler et al. 1987), A major drawback of SEM is the large sample size required. The proposed experiment assumes that a sufficiently large number of participants would be available to test the complete model and to compare competing submodels. Competing models can be sta- tistically compared to one another using the LaGrange multiplier and Wald tests (Bentler, 1986). During the data analysis phase of the experiment, manipulation checks would be performed to assess the extent to which the intended experimental treat- ments matched the reported experiences of subjects. Next, the complete research model would be tested using SEM to determine the overall goodness of fit of this data to the model. The following competing sub- models would then be individually tested: - baseline model (i.e.. task experience, task knowledge, ability, and performance), 2. mental workload model (i.e., task complexity, technology complexity, technology. experi- ence, task experience, task knowledge, ability, mental workload, and performance), 3. task/technology fit model (i.e., task require- ments, technology functionality, technology experience, task experience, task knowledge, task/technology fit, and performance), 4, baseline plus mental workload models, baseline plus task/technology fit models, and 6, mental workload plus task/technology fit models. All possible combinations of models would then’ be compared to determine the best-fitting model. If two or more models were equally best fitting, the ‘most parsimonious model would be identified as the ‘most eloquent representation (in this study) of indi- vidual decision-making where a computer-supported information system is used during the decision pro- cess. Future research experiments could test the model under various accounting-related decision sce~ narios in an effort to further assess the models’ inter- nal, external, and ecological validity References ‘Adams JS. Toward an understanding of inequity. Abnorm Soc Psychol 1963; 67:422-436, Amold V, Sutton SG. The theory of technology dominance: Understanding the impact of intelligent decision aids on decision makers’ judgments. Adv Account Behav Res 1998; 1:175-194, Bandura A. Social learning theory. Englewood Cliffs (NJ) Prentice-Hall, 1977, Barlett FC. Remembering: A study in experimencal and so: cial psychology. New York: Cambridge University Press, 1932. Bedard J. Expertise in auditing: myth or reality? Account Org Soe 1989314:113-131, Bedard J. Chi M. Expertise in auditing. Auditing: A Journal Practice Theory 1993;12 Suppl:21—45, Bedard J, Mock TT. Expert and novice problem-solving be- havior in audit planning. Audit J Prac Theor 1992; 11 Suppl:2-32 Benbasat I, Dexter AS. An experimental evaluation of ‘graphical and color-enhanced information presentation, ‘Manag Sci 1985; 31:1348-1364, Bentler PM. Theory and implementation of EQS, A struc tural equations program, Los Angeles: BMDP Stati cal Software, 1985. Bentler PM. Lagrange multiplier and Wald tests for EQS ‘and EQS/PC. Los Angeles: BMDP Statistical Software, 1986. Bentler PS ance structure a Lee SY, Weng LJ. Multiple population covari- ralysis under arbitrary distribution the- ‘ory. Commun Stat Theor 1987; 16:1951-1964, Bonner SE. Experience effects in auditing: The role of task- specific knowledge. Account Rev’ 1990; 65:72-92, Bonner SE, A mode! of the effects of audit task complexity Account Org Soc 1994; 19:213-234. Bonner SE, Lewis BL. Determinants of auditor expertise. J Account Res 1990: 30 Suppl: 1-20. Bonner S, Walker P. The effects of instruction and experi tence on the acquisition of auditing knowledge. Account Rev 1994; 69:157-178. Campbell JP, Campbell RJ. Industrial-organizational psy- chology and productivity: The goodness of fit. In ‘Campbell IP, Campbell RJ, eds. Productivity in organi- ations. San Franciseo (CA); Jossey-Bass Ine, 1988:82-93. Carrell MR, Dittrich JF. Equity theory: the recent literature, methodological considerations, “and new directio Acad Manag J 1978; $5:523-524, Cheney PH, Mann RI, Amoroso DL. Organizational factors affecting the success of end-user computing. J Manag Inform Syst 1986; 3:65-80. Choo F, Trotman K. The relationship between knowledge Benford and Hunton / International Journal of Accounting Information Systems 1 (2000} 54-65 65 structure and judgments for experienced and inexperi- enced auditors. Account Rev 1991; 66:464-485. Cuccia AD, Hakenbrack K, Nelson MW. The ability of pro- fessional standards to mitigate aggressive reporting. Ac- count Rev 1995; 70:227-248, Cashing BE, Ablawat SS. Mitigation of recency bias in au- dit judgment: The effect of documentation, Audit J Prac Theor 1996; 15:110-122. Davis JT. Experience and auditors’ selection of relevant in- formation for preliminary contro risk. Auditing: J Prac~ tice Theory 1996;15:16-37. Deci, EL. Intrinsic motivation. New York: Plenum Press, 1975, DeCharms R. Personal causation. Orlando, FL: Academic Press, 1968. Einhom HJ, Hogarth RM. Behavioral decision theory: Pro- ccess of judgment and choice. Annu Rev Psychol 1981; 32:53-88 Frederick D. Auditors’ representation and retrieval of inter- nal control knowledge. Account Rev 1991; 66:240-258. Goodhue DL. Understanding user evaluations of informa- tion systems. Manag Sci 1995; 41:1827-1844, Goodhue DL, Thompson RL. Task-technology fit and indi- Vidual performance, MISQ 1995:19:213-233. Hair JF, Anderson RE, Tatham RL, Black WC. Multivariate data analysis with readings. Englewood Cliffs (NJ): Prentice Hall, 1995. Hunton JE, McEwen RA. An assessment of the relation be- tween analysts’ earnings forecast accuracy, motivation incentives and cognitive information search strategy. Account Rev 1997; 72:497-515, Hunton JE, Stone DN, Weir B. Succeeding in managerial accounting: A structural equations analysis. Working paper, University of Ilinois, 1998, igen DR, Klein HJ. Individual motivation and_perfor- ‘mance: Cognitive influences on effort and choice, In: ‘Campbell JP, Campbell RJ, eds. Productivity in organi- zations, San Francisco (CA): Jossey-Bass, 1988:143— 176. Jex HR, Measuring mental workload: Problems, progress and promises. In: Handcock PA, Meshkati N, eds. #u- ‘man mental workload, North Holland: Elsevier Science, 1988:5-39 Kaplan $, Moeckel E, Williams C, Deahl J. Auditors’ hy- pothesis plausibility assessments in an analytical review setting. Audit J Prac Theor 1992; 11:50-65. Kennedy J. Debiasing the curse of knowledge in audit judg- ment. Account Rev 1995: 70:249-273, Libby R. The role of knowledge and memory in audit judg~ ‘ment. In; Ashton RH, Ashton AH, eds. Judgment and decision-making research in accounting and auditing New York: Cambridge University Press, 1995. 176— 206. Libby R, Luft J. Determinants of judgment performance in accounting settings: Ability, knowledge, motivation, and environment. Account Org Soc 1993; 18:425-450. Libby R, Tan H. Modeling the determinants of audit exper- tise. Account Org Soc 1994; 19:701-116. Locke EA, Shaw KN, Saari LM, Latham GP, Goal setting and task performance: 1969-1980. Psychol Bull 1981; 90:125-152. ‘Markus ML. System design features. In: Systems in organi= ‘zations, Marshfield (MA): Pittman Publishing, 1984. Markus H, Zajone RB. The cognitive perspective in social psychology. In: Lindsey S, Hanson E, eds. The hand- book of social psychology. New York: Random House, 1985:137-230. McDaniel L. The effects of time pressure and audit program structure on audit performance. J Account Res 1990; 28:267-285. McMahan ID. Relationships between causal attributions and expectancy of success. J Pers Soc Psychol 1973; 28:108-114, Naylor JD, Pritchard RD, flgen DR.A theory of behavior in ‘organizations. Orlando (FL): Academie Press, 1980. Neisser U. Cognition and reality: Principles and implica- tions of cognitive psychology. New York: W. H. Free ‘man, 1976. Nelson MW. The effects of error frequency and accounting knowledge on error diagnosis in analytical review. Ac- count Rev 1993; 68:804-824. Nelson MW, Libby R, Bonner SE. Knowledge structure and the estimation of conditional probabilities in audit plan- ring. Account Rev 1995; 70:27-47, Payne JW, Bettman JR, Johnson EJ. The adaptive decision ‘maker. New York: Cambridge University Press, 1993. Pinder CC. Work motivation theory, issues, and applica- tions. London: Scott, Foresman, 1984. Schroder HM, Driver MJ, Streufert S. Human information processing. New York: Holt, Rinehart and Winston, 1967. ‘Simon HA, Invariants of human behavior. Ann Rev Psychol 1990; 41:1-19. Sternberg RJ. Cognitive complexity. In: Sternberg RJ, ed. Encyclopedia of human intelligence. New York: Mac- Millan, 1994:286-290. Tan H, Libby R. Tacit managerial versus technical knowl- edge as determinants of audit expertise inthe field. J Ac: count Res 1997; 35:97-113. Tulving E. In: Tulving E, Donaldson W, eds. Organization cof memory. New York: Academic Press, 1972:381-403, Vessey I. Cognitive fit; A theory-based analysis of the graphs versus tables literature. Decision Sci 1991; 22:219-241 Vessey 1, Galletta D. Cognitive fit: An empirical study of information acquisition. Inform Syst Res 1991; 2:63-84. Vidulich MA. The cognitive psychology of subjective men- tal workload, In; Handcock PA, Meshkati N, eds. Hu- ‘man mental workload, North Holland: Elsevier Science, 1988:219-229, Vroom VH. Work and motivation, New York: Wiley, 1964. White RW. Motivation reconsidered: The concept of com- petence. Psychol Rev 1959; 66:297-333. ‘Wood R. Task complexity: Definition of the construct. Org Behav Hum Decision Proc 1986; 37:60-82.

Вам также может понравиться