Predicting information technology adoption in small businesses: an extension of the technology acceptance model.

Hayes, Thomas P., Jr.

INTRODUCTION

There are only a handful of studies that examine information technology (IT) adoption in small businesses. This is quite surprising given that smaller firms far outnumber larger ones and contribute significantly to the economy. According to the U.S. Small Business Administration (2009), small businesses are responsible for creating many new jobs and innovations as well as contributing close to half of the U.S. GDP (1).

Moreover, research suggests that small companies face unique IT issues, such as reliance on external IT expertise (Thong, Yap & Raman, 1996). Thus, the implication is that many studies that look at IT adoption may not be applicable to small businesses. Further, the studies that have examined IT adoption in small firms all imply that an important determinant is attitude toward the technology (Riemenschneider, Harrison, & Mykytyn, Jr., 2003; Caldeira & Ward, 2003; Mirchandani & Motwani, 2001).

Attitude toward technology is an integral component of the Technology Acceptance Model (TAM). Specifically, the TAM predicts that a user's attitude toward a particular technology ultimately affects whether or not they accept that technology. In fact, the TAM has already been used to study IT adoption in small businesses (e.g., Dembla, Palvia & Krishnan, 2007; Riemenschneider, Harrison, & Mykytyn, Jr., 2003).

However, by itself, the TAM only explains about 40% of the variance in computer usage, suggesting that additional factors may help explain IT acceptance (Legris, Ingham & Collerette, 2003). Thus, the purpose of this paper is twofold. Obviously, one goal is to improve upon the TAM by proposing a revised model that incorporates elements from the mental model literature. Most importantly, however, the goal is to offer a model that ultimately better explains IT adoption in a small business environment.

The rest of the paper is laid out as follows. First, a review of the TAM and mental model literature is provided followed by the revised model. Next, the revised model is empirically tested. Finally, limitations and future aspirations are discussed.

TECHNOLOGY ACCEPTANCE MODEL

The original form of the Technology Acceptance Model (TAM) (Davis, 1989; Davis, Bagozzi & Warshaw, 1989) is derived from the Theory of Reasoned Action (TRA), a commonly used theory from social psychology (Fishbein & Ajzen, 1975) (Figure 1). The TRA can be described as a conceptual framework that predicts whether or not an individual performs a certain behavior based on their behavioral intention (BI) to perform that behavior (16). Further, one's BI is determined by the individual's attitude (A) and subjective norm (SN) with respect to the behavior, where A is determined by one's beliefs and evaluations of the consequences related to that behavior (16). SN is determined by the individual's perception that one's referents have opinions about whether or not to perform the behavior and by the individual's motivation to comply with those referent opinions (16).

The TAM also asserts that one's behavior is determined by their intention to perform that behavior. The TAM, however, is specifically adapted to model users' acceptance of information systems (Davis, Bagozzi & Warshaw, 1989) (Figure 2). TAM posits that computer users' usage behavior is indirectly determined by two particular beliefs, perceived usefulness (U) and perceived ease of use (EOU) (985). This differs from the TRA where all beliefs are aggregated into a single construct (988).

Specifically, U is defined as the "user's subjective probability that using a specific application system will increase his or her job performance within an organizational context" (985). In other words, U refers to users' beliefs that the system will help them increase their job performance. EOU is defined as "the degree to which the prospective user expects the target system to be free of effort" (985). In other words, EOU refers to users' beliefs that the system is easy to use. Both these constructs directly affect a user's attitude (A) toward using the system, which, in turn, affects the user's behavioral intention (BI) to use the system.

There are a several additional differences to note regarding the TAM. First, the TAM excludes the social norm (SN) construct from the TRA. Fishbein and Ajzen (1975) note that little research has been done regarding normative beliefs or motivation to comply with those beliefs (304), and Davis, Bagozzi, and Warshaw (1989) concur, noting that SN is the least understood component of TRA (986). As cited in Davis, Bagozzi, and Warshaw (1989), the direct effects of the SN component are difficult to disentangle from its indirect effects through A (986). Thus, because of the problematic nature of SN, the authors chose to drop this construct from their model.

Second, Davis, Bagozzi, and Warshaw (1989) posit that BI may be directly affected not only by A, but also by U. That is, their model suggests a direct link between the user's perceived usefulness of the system and his/her intention to use the system. Third, they include external variables in their model, items that directly affect U and EOU. System features, training, user support consultants, and documentation are all examples of external variables in the TAM (988). Finally, the TAM's U and EOU are expected to generalize across other systems and users. In contrast, the TRA identifies new beliefs with every new context (988).

Several studies have used the TAM to study IT adoption in a small business setting (e.g., Chatzoglou, Vraimaki, Diamantidis & Sarigiannidis, 2010; Dembla, Palvia & Krishnan, 2007; Riemenschneider, Harrison, & Mykytyn, Jr., 2003). For example, in their study of web-enabled transaction processing by small businesses, Dembla, Palvia & Krishnan (2007) find that consistent with the TAM, perceived usefulness is a major determinant of adoption (10). Further, in their study of small and medium-sized businesses in Greece, Chatzoglou, Vraimaki,

[FIGURE 1 OMITTED]

[FIGURE 2 OMITTED]

Diamantidis and Sarigiannidis (2010) find that perceived usefulness as well as perceived ease of use are important determinants of computer acceptance. Overall, despite its potential for studying IT adoption in small businesses, the TAM still only explains, at most, about 40% of the variance in computer usage, suggesting that the current model does not include significant factors (Legris, Ingham & Collerette, 2003). Indeed, in their own study of IT adoptions in small businesses, Riemenschneider, Harrison & Mykytyn (2003) use a "combined" model that incorporates elements from the TAM and the Theory of Planned Behavior (TPB). Thus, in the spirit of furthering our ability to explain IT adoption in small businesses, it makes sense to consider additional constructs. One such construct, mental models, is discussed in the next section.

MENTAL MODELS

Mental models are defined as the user's internal representations of an object, which guide their interaction with that object (Staggers & Norcio, 1993). More specifically, a user's mental model of a computer system is their mental representation of that system, which guides their actions and helps them interpret the system's behavior (Young, 1981). Overall, mental models provide users with predictive and explanatory power for understanding their interaction with the system (Norman, 1983).

Unfortunately, prior research has coined various other terms that are used synonymously with mental models, such as conceptual models, cognitive models, mental models of discourse, component models, and causal models (Staggers & Norcio, 1993). Norman (1983) provides some clarification on this matter. He defines the conceptual model as an accurate, consistent, and complete model of the system that is created by teachers, designers, scientists, or engineers (7). In contrast, a user's mental model represents what they actually "have in their heads" and may not be the same as the conceptual model (12).

Norman's distinction between these two terms shall be used throughout the remainder of the paper. Namely, the term conceptual model will be used to mean a model that describes how the system should work, i.e., a blueprint. Conceptual models are thus external to the individual. On the other hand, the term mental model will be used to refer to the user's understanding of how the system works. In other words, a mental model is internal to the individual, representing the individual's mental template of the system.

Some of the past literature on mental models looks at how individuals form their mental models. One notion is that people use analogies to structure unfamiliar domains (Gentner & Gentner, 1983). As cited in Staggers and Norcio (1993), Douglas (1982) finds that subjects create a typewriter model when they are learning to use a text editor, indicating that individuals transfer familiar knowledge to similar, yet unfamiliar, domains (590).

Much of the past literature, however, examines how users' mental models aid in their learning process. For example, Brandt (2001) examines how users employ mental models to obtain task-specific knowledge and to help in solving problems. The literature supports the general notion that users' mental models serve as templates for understanding in a variety of contexts, including where users are using or learning to use a computer system.

Further, past research has looked at ways to help users create more useful mental models. In particular, studies have examined the benefits from providing users with a conceptual model of a system prior to training them on the system. For example, Bayman and Mayer (1984) report evidence that providing users with conceptual models helps them develop more useful mental models of the system (197). In other words, providing users with a diagram of how the system works helps them to better understand the system. Young (1981) also offers evidence that supports this conclusion. In addition to having a better understanding of the system, these studies provide evidence that users perform better as well (e.g., Bayman & Mayer, 1984; Mayer, 1981; Young, 1981).

REVISED MODEL

As mentioned in the previous section, research on mental models suggests that in order for users to interact effectively with a system, they need to create a mental representation of how the system works (e.g., Young, 1981; Brandt, 2001). This model guides users' actions and helps them interpret system behaviors (Young, 1981). Further, researchers have argued that individuals employ mental models to build their knowledge base (e.g., task-specific knowledge) and to aid in problem-solving (Brandt, 2001).

Additional evidence indicates that providing users with a conceptual model of the system beforehand aids in their understanding of the system (i.e., developing their mental model of the system) and improves their performance (Bayman & Mayer, 1984; Mayer, 1981; Young, 1981). Intuitively, this result makes sense, for it implies that users perform better when they understand how the system works. Subjects that receive conceptual models before their interaction with the system are able to develop more useful mental models of the system (i.e., a better understanding of the system), which, in turn, improves their performance.

Overall, the evidence signifies that mental models are an important aspect of the user's interaction with the computer system. Additionally, the evidence suggests that providing users with a conceptual model of the system aids them in developing a more useful mental model of the system, i.e., facilitating their understanding of how the system works. More importantly, however, this stream of research provides insight into one determinant of technology acceptance. Specifically, the results from these studies imply that providing subjects with a conceptual model of a computer system will facilitate a greater understanding of the system, which, in turn, may make the system seem easier to use (i.e., perceived ease of use).

Based on evidence provided by the mental model literature, the following revised model is proposed (Figure 3). In the revised model, the original TAM is extended to include mental models. For simplifying purposes, the original model is used as the basis for my revised model. Venkatesh and Davis (2000) do propose an extension of the Technology Acceptance Model, namely, TAM2, which adds several additional items, including subjective norm. Legris, Ingham and Collerette (2003), however, cite that this most recent version of the TAM still only accounts for about 40% of users' acceptance behaviors, suggesting that the TAM is missing significant factors. Thus, for this paper, the original TAM is used.

Specifically, it is posited that a user's mental model of the computer system affects their perceived ease of use. That is, it is expected that a user's knowledge of how a system works will impact their perceived ease of use of the system. Referring back to Norman's (1983) distinction, a mental model represents a user's understanding of how the system works.

EMPIRICAL TESTING

To operationalize this new construct, users could be provided with a conceptual model of the system. In turn, with the knowledge of how the system works, users can create a more useful mental model of the system, which should make it easier for them to use. In other words, users' greater understanding of the system should positively affect their perceived ease of use, which should positively affect their actual system use.

[FIGURE 3 OMITTED]

Intuitively, it makes sense that including mental models can improve the predictive power of the original TAM. After all, Caldeira and Ward (2003) report that a contributing factor for successful IT adoption in small businesses is IS/IT training (132). Similarly, in interviews with small business owners and managers, Mirchandani and Motwani (2001) note that employees' knowledge of computers was a significant factor in whether they adopted electronic commerce technology. In other words, users are more likely to adopt a particular technology when they understand it. According to Bayman and Mayer (1984), one way to aid this understanding is to provide users with a conceptual model of the system.

To test the above revised model, 132 subjects were provided with two case scenarios, both which included a decision aid technology. In both cases, subjects worked through the given task where they had to assess fraud risk for a fictitious company. On the first case, they also had the requirement that they must use the decision aid. Effectively, this first case served as training on the decision aid. It also served to help familiarize them with the case layout and the assigned task.

Subjects then completed the second case with the option to use the decision aid. In both cases, subjects were provided with a conceptual model of the decision aid. Subjects also completed questionnaires after completing each case. The questionnaires were used to collect demographic data, as well as perceived ease of use (EOU) with respect to the decision aid and intent to use the aid (INTENT).

For both EOU and INTENT, the scales were self-reported on a 6-point Likert scale from "Strongly Disagree" to "Strongly Agree." Further, for analysis purposes, summated scales were created for EOU and INTENT, and reliability analysis was performed on both scales. For EOU and INTENT, the Cronbach's alpha coefficients are 0.73 and 0.87, respectively, suggesting that both scales are internally consistent.

A simple average of the responses for EOU was 4.8, meaning that on average, subjects agreed that the decision aid was easy to use. More importantly, the data suggests that as a result of the conceptual model provided to subjects, they found the aid easier to use.

Additionally, a simple linear regression was performed to determine if EOU predicts INTENT. Results of this regression were significant (t = 3.673, p < 0.001). In other words, the results suggest that subjects were more likely to use the decision aid if they found it easy to use.

Finally, it is important to note that INTENT serves as a good proxy for actual usage. Over 90% of the subjects actually chose to use the decision aid on the second case. Another simple regression was performed to determine if INTENT predicts actual usage. Results of this regression were significant (t = 2.975, p < 0.01). In other words, the results suggest that intent to use generally translates to actual usage.

DISCUSSION AND LIMITATIONS

Per the revised model, it is expected that users' mental models of the system affect their perceived ease of use, which, in turn, affects their intent to use the system. Specifically, users' mental models are operationalized as providing a conceptual model of the system. Empirical testing was performed to test this model. Results suggest that subjects found the decision aid easier to use, and accordingly, were more likely to use the decision aid.

The results of the current study have important implications for studying IT adoption in small businesses. First of all, the results continue to support the link between perceived ease of use (EOU) and intent to use (INTENT) a particular technology. Specifically, users are more likely to use the technology if it's easy to use. In a small business environment, this conclusion is particularly important given the significant cost associated with implementing most technologies. Second, and most importantly, the results suggest that providing a conceptual model of the system makes the system appear easier to use because users gain a better understanding of the system. Again, this conclusion is especially important for small businesses as it ultimately increases the likelihood of successfully adopting a technology.

Despite the encouraging results, there are still several limitations to consider. First, the generalizability of the results is limited since the current study greatly simplified the research setting to include only those variables of interest. Along similar lines, it is important to note that the adjusted R-square of the model (i.e., INTENT regressed on EOU) is fairly low (9%), suggesting that other important factors are missing. Future research should seek to uncover additional factors, as well as test the significance of other factors already in the revised model (e.g., perceived usefulness). Finally, there is the potential for measurement error since both EOU and INTENT are self-reported measures. While the use of multi-item scales somewhat mitigates this likelihood, it is still possible for measurement error to occur.

Overall, the present study adds to the TAM literature by extending the original TAM model. Specifically, the addition of mental models enriches the original TAM model, and theoretically can be applied across a variety of contexts. Finally, the study adds to the small business literature by offering a model that contributes to our knowledge of the factors that lead to successful IT adoption in a small business environment.

REFERENCES

Al-Gahtani, S. (2001). The applicability of TAM outside North America: An empirical test in the United Kingdom. Information Resources Management Journal 14(3), 37-46.

Bayman, P. & R.E. Mayer (1984). Instructional manipulation of users' mental models for electronic calculators. International Journal of Man-Machine Studies 20(2), 189-199.

Brandt, D.S. (2001). Information technology literacy: Task knowledge and mental models. Library Trends 50 (1), 73-86.

Caldeira, M.M. and J.M. Ward (2003). Using resource-based theory to interpret the successful adoption and use of information systems and technology in manufacturing small and medium-sized enterprises. European Journal of Information Systems 12(2), 127-141.

Chatzoglou, P.D., E. Vraimaki, A. Diamantidis, and L. Sarigiannidis (2010). Computer acceptance in Greek SME's. Journal of Small Business and Enterprise Development 17(1), 78-100.

Davis, F.D. (1989). Perceived usefulness, perceived ease of use, and user acceptance of information technology. MIS Quarterly 13(3), 319-340.

Davis, F.D., R.P. Bagozzi, and P.R. Warshaw (1989). User acceptance of computer technology: A comparison of two theoretical models. Management Science 35(8), 982-1003.

Dembla, P., P. Palvia, and B. Krishnan (2007). Understanding the adoption of web-enabled transaction processing by small businesses. Journal of Electronic Commerce Research 8(1), 1-17.

Douglas, S.A. (1982). Learning to text edit: semantics in procedural skill acquisition. Unpublished doctoral dissertation, Stanford University.

Fishbein, M. and I. Ajzen (1975). Belief, attitude, intention, and behavior. Reading, MA: Addison-Wesley Publishing.

Gentner, D. and D.R. Gentner (1983). Flowing waters or teeming crowds: mental models of electricity. In D.Gentner & A.L. Stevens (Eds.), Mental Models (pp. 99-129). Hillsdale, NJ: Lawrence Erlbaum Associates.

Hayes, T. (2008). An examination of the factors that influence an auditor's decision to use a decision aid in their assessment of management fraud. International Journal of Business, Accounting, and Finance 2(1), 67-82.

Legris, P., J. Ingham, and P. Collerette (2003). Why do people use information technology? A critical review of the technology acceptance model. Information and Management 40(3), 191-204.

Mayer, R.E. (1981). The Psychology of how novices learn computer programming. ACM Computing Surveys 13 (1), 587-605.

Mirchandani, D.A. and Motwani, J. (2001). Understanding small business electronic commerce adoption: An empirical analysis. The Journal of Computer Information Systems 41(3), 70-73.

Norman, D.A. (1983). Some observations on mental models. In D. Gentner & A.L. Stevens (Eds.), Mental Models (pp. 7-14). Hillsdale, NJ: Lawrence Erlbaum Associates.

Riemenschneider, C.K., D.A. Harrison, and P.P. Mykytyn, Jr. (2003). Understanding IT adoption decisions in small business: integrating current theories. Information & Management 40(4), 269-284.

Staggers, N. and A.F. Norcio (1993). Mental models: Concepts for human-computer interaction research. International Journal of Man-Machine Studies 38(4), 587-605.

Thong, J., C. Yap, and K. Raman (1996). Top management support, external expertise and information systems implementation in small businesses. Information Systems Research 7(2), 248-267.

U.S. Small Business Administration (2009). The small business economy: A report to the President. Washington, DC: U.S. Government Printing Office.

Venkatesh, V. and F.D. Davis (2000). A theoretical extension of the technology acceptance model: Four longitudinal field studies. Management Science 46(2), 186-204.

Young, R.M. (1981). The machine inside the machine: users' models of pocket calculators. International Journal of Man-Machine Studies 15(1), 51-85.

Thomas P. Hayes, Jr., University of Arkansas--Fort Smith