- Email: [email protected]
A contingency model of computer and Internet self-efficacy
Information & Management 43 (2006) 541–550 www.elsevier.com/locate/dsw
A contingency model of computer and Internet self-efficacy Gholamreza Torkzadeh *, Jerry Cha-Jan Chang, Didem Demirhan Department of MIS, College of Business and Economics, University of Nevada, Las Vegas, USA Accepted 9 February 2006
Abstract Information system researchers have recently devoted considerable attention to the concept of computer self-efficacy in order to understand computer user behavior and system use. This article reports on the development and examination of a contingency model of computer and Internet self-efficacy. User attitude and computer anxiety were assumed to influence the development of computer and Internet self-efficacy. Measures of user attitude, computer anxiety, computer self-efficacy, and Internet self-efficacy were used in a university environment to collect 347 responses at both the beginning and end of an introductory computer course. Results suggested that training significantly improved computer and Internet self-efficacy. Respondents with ‘favorable’ attitudes toward computers improved their self-efficacy significantly more than respondents with ‘unfavorable’ attitudes. Respondents with ‘low’ computer anxiety improved their self-efficacy significantly more than respondents with ‘high’ computer anxiety. The interaction effect between attitude and anxiety was significant for computer self-efficacy scores but not for Internet self-efficacy scores. The implications of these findings are discussed. # 2006 Elsevier B.V. All rights reserved. Keywords: Computer self-efficacy; Internet self-efficacy; Computer anxiety; Computer user attitude; Computer user training; Gender; Contingency model
1. Introduction The ultimate question about information technology effectiveness relates to its impact on the individual and organizations. The effective use of an IS is influenced by not only system design features but also by the user’s ability to use the system effectively in making decisions, plan work, service customers, or control events. Self-efficacy reflects the belief that individuals have about their ability to use systems effectively. Research studies suggest that the higher the induced level of self-efficacy, the greater is performance
* Corresponding author at: Department of MIS, College of Business and Economics, University of Nevada, Las Vegas, 4505 Maryland Parkway, Box 456034, NV, USA. Tel.: +1 702 895 3796; fax: +1 702 895 0802. E-mail address: [email protected] (G. Torkzadeh). 0378-7206/$ – see front matter # 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.im.2006.02.001
achievement [1]. Individuals with high self-efficacy work harder and longer than individuals with low selfefficacy [44]. Computer self-efficacy is defined as an individual’s belief regarding their ability to use a computer [10]. Research suggests that it plays a significant role in an individual’s decision to use computers and how comfortable users are in learning skills related to effective use [26,32]. MIS researchers have, in recent years, devoted much effort in studying computer user training, user attitude, and computer anxiety as they relate to computer self-efficacy. Marakas et al. provided a comprehensive review of related literature and the path that research on computer self-efficacy has traveled. Although computer self-efficacy constructs have been the subject of research studies, contingency models that examined the influencing effect of user attitude and computer anxiety on computer self-efficacy
542
G. Torkzadeh et al. / Information & Management 43 (2006) 541–550
have not been considered. User attitude and computer anxiety are important variables and are expected to influence the outcome of self-efficacy development efforts [27,39]. Improving our understanding about influence on training programs should help in making better decisions regarding technology implementation, acceptance, and use. In our study, the pattern of change in computer and Internet self-efficacy was examined as individuals learned about computers and interacted with them. Survey responses were collected from 347 students at the beginning and end of an introductory computer course. Data were analyzed to examine the relationship between training and self-efficacy and how this relationship was influenced by user attitude and computer anxiety. Our specific goals were to examine the relationships between training and (a) computer self-efficacy, (b) Internet self-efficacy, (c) computer self-efficacy controlling for user attitude, (d) Internet self-efficacy controlling for user attitude, (e) computer self-efficacy controlling for computer anxiety, (f) Internet self-efficacy controlling for computer anxiety, (g) computer self-efficacy controlling for the interaction effect between attitude and anxiety, and (h) Internet self-efficacy controlling for the interaction effect between attitude and anxiety. 2. Self-efficacy construct Research on self-efficacy concept has a long tradition in social sciences with notable works by Bandura [2,4,5], Schunk [36,37], Gist [14,15], and others. Self-efficacy arises from the gradual acquisition of complex cognitive, social, linguistic and/or physical skills through experience [3]. It is expected to affect task effort, persistence, expressed interest, and the level of goal difficulty selected for performance. Individuals appear to evaluate information about their abilities and then regulate their choices and efforts accordingly. The strength of their conviction in their own effectiveness is likely to influence whether they will try to cope with a given situation. Individuals with high efficacy expectations have a greater chance of success in a given task [30]. Self-efficacy is a dynamic construct that changes as new information and experiences are acquired. It is generally described as having three components: magnitude—the levels of task difficulty that people believe they can attain; strength—their conviction about its magnitude; and generality—the degree to which the expectation is generalized across situations. The purpose in evaluating these components is to discover
the type of questions that will best explain and predict a person’s dispositions, intentions, and actions. Schunk described how self-efficacy influenced academic learning processes. At the start of an activity, students hold differing beliefs about their ability to acquire knowledge, perform skills, master the material, etc. Initial self-efficacy varies as a function of aptitude (such as abilities and attitudes) and prior experience. Such personal factors as goal setting and information processing, along with situational factors, such as rewards and teacher feedback, affect students while they are learning. From these, students derive cues signaling how well they are learning, which they use to assess efficacy for further learning. Motivation is enhanced if students perceive that they are making progress in learning. In turn, as students work on tasks and become more skillful, they maintain a sense of self-efficacy for performing well. Thus training improves self-efficacy and individual differences moderate the outcome of this training: indeed computer training may be more effective for people who have a positive attitude toward computers and no computer anxiety. This suggests a contingency model of self-efficacy where individual differences influence learning outcomes. 3. Research model and hypotheses In our contingency model, user attitude and computer anxiety were the conditional variables that moderated the outcome of computer training. Here, we measured training outcome in terms of pattern of change in computer and Internet self-efficacy as trainees went through a semester long computer course. Part (a) in Fig. 1 shows our assumed moderating effect of computer anxiety and user attitude on computer and Internet selfefficacy. Part (b) shows the assumed moderating effect of the interaction between computer anxiety and user attitude on computer and Internet self-efficacy. Research studies in cognitive modeling, behavioral modeling, and self-management suggest that training programs enhance self-efficacy [12,13,16]. MIS research has focused on computer self-efficacy as a special case of general self-efficacy. MIS researchers, similarly, have illustrated the importance of training for computer self-efficacy. Training is also suggested to enhance Internet self-efficacy, considered an extension of computer self-efficacy construct in the domain of World Wide Web [31]. Thus we proposed: Hypothesis 1. Computer training improves computer self-efficacy.
G. Torkzadeh et al. / Information & Management 43 (2006) 541–550
543
Fig. 1. A contingency model of self-efficacy.
Hypothesis 2. Computer training improves Internet self-efficacy. Studies in the field of psychology suggest that people experience anxiety when performing behaviors that they do not feel they can perform competently. When trainees experience anxiety, they cannot concentrate on learning; they focus on inner feelings and thoughts [9,38]. Studies have suggested that computer anxiety reduces the effectiveness of computer-based training [17,22]. We, therefore, expected computer anxiety to adversely affect training outcome measured in terms of improvement in computer or Internet self-efficacy. Thus, we proposed: Hypothesis 3. Computer anxiety negatively influences computer training outcome measured in terms of pattern of change in computer self-efficacy. Hypothesis 4. Computer anxiety negatively influences computer training outcome measured in terms of pattern of change in Internet self-efficacy. User attitude also plays an important role in training and learning. Studies have suggested that user attitude toward computer influenced the outcome of training programs [7,41,45]. A positive user attitude toward computer is expected to positively influence the outcome of training programs and the level of selfefficacy improvement. Thus, we proposed: Hypothesis 5. User attitude influences computer training outcome measured in terms of pattern of change in computer self-efficacy.
Hypothesis 6. User attitude influences computer training outcome measured in terms of pattern of change in Internet self-efficacy. We have argued that anxiety and attitude, separately, influence the outcome of computer training. Research also suggested a significant relationship between computer anxiety and user attitude [6,18,23]. For example, Harris suggested that computer anxiety had negative influence on attitude toward end-user-computing. However, this relationship has not been widely tested and there is thus a need to examine the interaction effects between computer anxiety and user attitude on computer training. We expected the interaction between anxiety and attitude to influence the computer training outcome and thus constructed the following hypotheses: Hypothesis 7. The interaction between computer anxiety and user attitude influences computer training outcome measured in terms of pattern of change in computer self-efficacy. Hypothesis 8. The interaction between computer anxiety and user attitude influences computer training outcome measured in terms of pattern of change in Internet self-efficacy. These hypotheses are summarized in Table 1, which also provides literature support for each. 4. The instruments Arguably the most important aspect of computer self-efficacy to MIS research is the development of
544
G. Torkzadeh et al. / Information & Management 43 (2006) 541–550
Table 1 Summary of hypothesis and supporting literature Hypothesis
Statement
Literature support
H1 and H2 H3 and H4
Training improves computer and Internet self-efficacy Computer anxiety influences training outcome in terms of pattern of change in computer self-efficacy and Internet self-efficacy Computer user attitude influences training outcome in terms of pattern of change in computer self-efficacy and Internet self-efficacy The interaction between computer anxiety and user attitude influences training outcome in terms of pattern of change in computer self-efficacy and Internet self-efficacy
[10,12,13,15,16,26,31,39] [9,17,22,7]
H5 and H6 H7 and H8
valid and reliable measures that can be used with confidence to accumulate findings and facilitate substantive hypotheses testing. We have not developed a sufficient pool of research instruments in this domain and this has hampered research on computer self-efficacy. Two broad perspectives have emerged for measures of computer self-efficacy. One encourages developing task specific measures, while another encourages developing general measures. Works of Murphy et al. [29] and Compeau et al. [11] are examples of general and task specific measures, respectively. Marakas et al. suggested the need for specific computer self-efficacy measures while Torkzadeh et al. validated and used general measures. The argument for use of specific measures of computer self-efficacy suggested that it was important to align computer self-efficacy with the particular task domain. Such a strong alignment would be likely to improve the exploratory power of the instruments. This perspective served to motivate the development of measures that assessed the individual’s level of selfefficacy with specific task such as the use of a spreadsheet, suggesting that a new measure is needed every time a new area is explored. The argument for general measures of computer self-efficacy is that it is difficult to establish a research tradition if new measures are to be developed every time research is attempted: task specific measures may not be stable over time for use in longitudinal or follow up studies. This perspective has served to motivate the development of general measures that are likely to be stable over time and applicable for longitudinal studies. The inherent nature of computer technology involves constant and often rapid change, most of which is not accurately predicted. This unpredictable nature has posed challenges for research in IS in general and instrument development in particular. This does not mean that existing measures should not be used, but that they must be used with caution and their reliability must be reported each time they are used.
[7,39,40,45] [6,18,23]
4.1. Computer self-efficacy We used an instrument developed by Murphy et al. based on the work of Bandura, in which he conceptualized self-efficacy as individualized selfperception varying across activities and situational circumstances, rather than as a goal disposition that can be assessed by an omnibus test. Domain specific measures, such as computer self-efficacy, allow researchers to be more accurate in evaluating computer training outcome than general instruments. Owen [33] also suggested that self-efficacy could be reliably measured and that such measures might be used to assess a composite of affect, cognition, and performance in the attainment of program and course objectives. Murphy et al. suggested that their instrument could be used to evaluate skill attainment at both preand post-training. Later, Harrison and Rainer [19] used this instrument to measure the respondent’s perceptions regarding specific computer-related knowledge and skills and reported an overall reliability coefficient of 0.95 for the instrument. In a more recent study, Torkzadeh et al. used it in a study of training and computer self-efficacy and also reported a reliability of 0.95. Our study used 25 of the original 32-item instrument developed by Murphy et al. (see Appendix A). Items related to mainframe system were deemed irrelevant here, and were dropped. 4.2. Internet self-efficacy The computer self-efficacy instrument pre-dated the rise and importance of Internet related skills. In a more recent study, Torkzadeh and Van Dyke [40] developed a 17-item instrument for measuring the individual’s self-perception and self-competency in interacting with the Internet. They based their measurement development on Bandura’s conceptualization of self-efficacy and other studies of social and cognitive psychology. Later, they used this 17-item instrument in a study of training and Internet self-
G. Torkzadeh et al. / Information & Management 43 (2006) 541–550
efficacy and reported a reliability of 0.96. We used 15 items of the instrument developed by Torkzadeh and Van Dyke for measuring Internet self-efficacy. Two items in the original instrument were deemed redundant and were dropped. Both computer selfefficacy and Internet self-efficacy instruments used a five point Likert-type scale where 1 = strongly disagree and 5 = strongly agree. 4.3. Computer anxiety and user attitude Our study used measures of computer anxiety developed by Heinssen et al. [21] and used by Compeau et al. to study computer self-efficacy and system use. It also used measures of user attitude developed by Loyd and Gressard [25] and used by Compeau et al. In their study, Compeau et al. used five items for measuring user attitude or ‘‘affective response’’ and four for measuring computer anxiety. Affect or user attitude, they argued, represented the positive side (where a person enjoyed interacting with a computer) and anxiety represented the negative side (where a person experienced feelings of apprehension or anxiety when interacting). Compeau et al. used partial least square tests (PLS) to assess their measurement model and reported internal consistency greater than 0.7 for measures of computer anxiety and user attitude. We used the five items of user attitude and four items of computer anxiety of Compeau et al. Both computer anxiety and user attitude instruments used a five point Likert-type scale, where 1 = strongly disagree and 5 = strongly agree. 5. The study A survey instrument containing measures of computer self-efficacy, Internet self-efficacy, user attitude, and computer anxiety was administered to business undergraduates at a large state university in the southwest region of the United States. Students in multiple sections of an introductory course participated in the study. The course is required for all business students. Class sizes ranged from 45 to 60. Course content and grade distribution were similar for all sections. The course covered information technology infrastructure, management and organizational support systems, information systems in the enterprise, electronic business, telecommunications and networks, management of hardware and software, file and database management, systems development, decision support systems, and aspects of a computer career. The measures were administered at both the beginning and end of the course. The time between
545
pre- and post-test ranged from 11 to 12 weeks. Respondents had no prior knowledge that they would be asked to repeat the survey at the end and were asked to respond to questions focusing on their current belief. Participation in the study was voluntary; student names were used to match pre- and post-training responses. A total of 347 usable matching responses consisting of 201 male (58%) and 146 female students were obtained. Incomplete responses and those with matching problems were omitted. The survey also asked respondents about their educational interest and age. The sample was diverse in terms of respondents’ educational interest. They were majoring in accounting (19.7%), MIS (18.3%), management (16.8%), finance (15.7%), marketing (13.0%), economics (2.3%), and other. The respondents’ age distribution was: under 20 years (10.1%), 20–29 (77.8%), 30–39 (9.5%), and greater than 40 (2.6%). This sample was considered large enough to confirm the measures and test the research hypotheses. 6. Data analysis and results The reliability (internal consistency) of items in each scale was examined using Cronbach’s alpha to confirm the adequacy of the measures for testing the hypotheses (see Table 2). These coefficients confirmed results of earlier studies and provided confidence in testing the hypotheses. The reliability results are based on pretraining data. However, post-training data had very similar results. Respondents were grouped as people with ‘high’ or ‘low’ computer anxiety using the mean score of 1.95 (S.D. = 0.83) on the pre-test computer anxiety scale. Similarly, respondents were grouped as people with ‘favorable’ or ‘unfavorable’ attitude toward computers using the mean score of 3.98 (S.D. = 0.74) on the pre-test computer user attitude scale. Overall, computer user attitude scores were high for all respondents. This is consistent with recent studies and different from what was reported in earlier studies [34], suggesting an upward change in user’s attitude toward computers over the years. Table 2 Reliability of measures Factors
No. of items
Alpha
Computer self-efficacy Internet self-efficacy Attitude Anxiety
25 15 5 4
0.95 0.91 0.81 0.81
546
G. Torkzadeh et al. / Information & Management 43 (2006) 541–550
Table 3 The effect of training on computer and Internet self-efficacy
Computer self-efficacy Internet self-efficacy
Pre, mean (S.D.)
Post, mean (S.D.)
4.24 (0.57) 3.59 (0.81)
4.34 (0.58) 3.77 (0.79)
Source
Table 3 provides pre- and post-training scores for computer self-efficacy and Internet self-efficacy for all respondents. The score for each scale was equal to the total points divided by the number of items. The difference in mean scores was examined using a paired t-test procedure. Both differences in mean scores were significant at p < 0.001. The results suggested that respondents entered training programs with a relatively high level of computer and Internet self-efficacy but that the levels improved significantly after the course. These findings supported Hypotheses 1 and 2. Data were also analyzed to examine the influence of computer anxiety and user attitude on computer and Internet self-efficacy as students went through the training program. The ANOVA method was used to examine the difference between subject responses as well as possible interaction effects of computer anxiety and user attitude on computer and Internet selfefficacy. Pre- and post-training data were used to examine the pattern of change in computer and Internet self-efficacy. Table 4 presents the ANOVA results; they suggested computer and Internet self-efficacy scores were significantly different for respondents with ‘high’ and ‘low’ computer anxiety. In the between subjects tests, there were significant main effects of computer anxiety on both computer and Internet selfefficacy. In the within-subjects tests, there was a significant main effect for computer and Internet selfefficacy; consistent with the t-test results. The interaction effect for within-subjects is not applicable Table 4 The effect of computer anxiety on computer and Internet self-efficacy
Within subject effects Training Training anxiety Between subject effects Anxiety
Sig. 4.25 5.22
0.000 0.000
Table 5 The effect of user attitude on computer and Internet self-efficacy
6.1. Individual differences and self-efficacy development
Source
t
Computer S.E.
Internet S.E.
F-value
Sig.
F-value
Sig.
16.16 5.17
0.000 0.024
27.47 0.29
0.000 0.593
146.80
0.000
59.93
0.000
Within subject effects Training Training attitude Between subject effects Attitude
Computer S.E.
Internet S.E.
F-value
Sig.
F-value
Sig.
20.82 5.84
0.000 0.016
25.63 0.61
0.000 0.436
125.20
0.000
55.00
0.000
here; we used the interaction effect for betweensubjects throughout the paper. These results supported Hypotheses 3 and 4. Table 5 presents the ANOVA results that suggested that computer and Internet self-efficacy scores were significantly different for respondents with ‘favorable’ and ‘unfavorable’ attitudes toward computers. In the within-subject tests, there was a significant main effect for both computer self-efficacy and Internet selfefficacy. In the between-subjects tests, there were significant main effects of user attitude on both computer self-efficacy and Internet self-efficacy. These results supported Hypotheses 5 and 6. 6.2. Interaction effects Table 6 presents the ANOVA results; they suggested that the interaction between user attitude and
Table 6 The interaction effect on computer and Internet self-efficacy training Source
Within subject effects Training Training attitude Training anxiety Training attitude anxiety Between subject effects Anxiety Attitude Attitude anxiety
Computer S.E.
Internet S.E.
F-value
Sig.
F-value
Sig.
14.19 2.45 2.19 0.06
0.000 0.119 0.140 0.811
17.88 0.64 0.00 0.61
0.000 0.423 0.970 0.435
88.25 53.42 4.08
0.000 0.000 0.044
33.22 20.02 2.89
0.000 0.000 0.090
G. Torkzadeh et al. / Information & Management 43 (2006) 541–550
547
Table 7 Scenarios for individual differences
computer anxiety influenced the outcome of training for computer self-efficacy (at 0.05) but not for Internet self-efficacy. These results supported Hypotheses 7 but not 8. Results supported earlier analyses. Data were analyzed to further examine the interaction effects of possible scenarios for individual differences. Respondents were grouped as having ‘high’ or ‘low’ computer anxiety as well as having ‘favorable’ or ‘unfavorable’ attitude toward computers. This enabled us to analyze four possible scenarios of individual differences. Table 7 depicts these scenarios and shows the sample size for each. The sample size is sufficiently large to warrant test of difference in mean scores for the pattern of change in computer and Internet self-efficacy as individuals go through training. Both Tukey and Scheffe multiple comparisons procedures were used to test differences between possible mean scores; they produced similar results. Among the four possible situations, scenarios 2 and 3 were expected to have the most influence on selfefficacy development, in an opposite direction. Results suggested significant differences (at 0.01) between mean score for scenario 2 and for the other three scenarios for both computer and Internet self-efficacy. Results also suggested significant differences (at 0.01) between mean score for scenario 3 and for the other three scenarios for computer self-efficacy and scenarios 1 and 2 of Internet self-efficacy; there was no significant difference between mean scores for scenario 3 and 4 for Internet self-efficacy. 7. Discussion User acceptance and the effective use of information technology have been considered essential success factors for technology management [43]. In order to manage information technology better, we need to
understand why individuals readily accept and use computer applications, persist at improving their computer skills, select challenging projects, search for innovative and new ways of using computer systems, and eventually have high rates of success. The construct of computer self-efficacy is thus important to our understanding of computer user behavior: it helps predict user perceptions and subsequent acceptance and use of computer systems [42]. There has been increased research to help understand computer self-efficacy construct and its correlates. Studies have examined the role of training in computer self-efficacy development. The results of our study support earlier findings that training programs influence self-efficacy. More specifically, we examined the role of training in computer and Internet selfefficacy. We examined the influence of individual differences on the outcome of self-efficacy development. User attitude and computer anxiety were shown to influence the pattern of change in computer and Internet self-efficacy as students went through a computer course. The results suggested that individuals with ‘favorable’ attitudes toward computers improved their computer and Internet self-efficacy significantly more than individuals with ‘unfavorable’ attitudes. The results also suggested that individuals with ‘low’ computer anxiety improved their computer and Internet self-efficacy significantly more than individuals with ‘high’ computer anxiety. The results support the hypothesis that the interaction between user attitude and computer anxiety influence the level of improvement in computer and Internet self-efficacy. Individuals with ‘low’ computer anxiety and ‘favorable’ attitudes toward computers improved their computer and Internet self-efficacy significantly more than any other group. In contrast, individuals with ‘high’ computer anxiety
548
G. Torkzadeh et al. / Information & Management 43 (2006) 541–550
Table 8 Interaction effect by gender Source
Computer self-efficacy Male (n = 201) F-value
Within subject effects Training Training anxiety Training attitude Training attitude anxiety Between subject effects Anxiety Attitude Attitude anxiety
Sig.
Internet self-efficacy Female (n = 146)
Male (n = 201)
Female (n = 146)
F-value
Sig.
F-value
Sig.
F-value
Sig.
6.24 0.37 5.76 0.21
0.013 0.545 0.017 0.645
5.65 2.89 0.37 0.13
0.019 0.091 0.542 0.724
10.09 0.00 0.14 0.07
0.002 0.963 0.707 0.798
5.73 0.01 3.19 1.59
0.018 0.906 0.076 0.209
58.47 59.54 3.22
0.000 0.000 0.074
30.00 5.52 2.24
0.000 0.020 0.137
21.71 33.50 1.39
0.000 0.000 0.239
11.99 0.05 3.33
0.001 0.817 0.070
and ‘unfavorable’ attitudes toward computers improved their computer and Internet self-efficacy less than any other group. Analyses of scenarios for individual differences suggested that computer anxiety exerted more influence on self-efficacy development than did user attitude. Studies have also examined gender issues in selfefficacy development. Earlier findings suggested that males reported higher self-efficacy than females [8,24,28] or males feel more in control when interacting with computers [20]. More recent studies suggested that there were no significant differences in responses for males and females or that the gender difference in computer skill is diminishing [35]. It seemed appropriate to examine the relationship between training and self-efficacy for males and females here, especially since the sample included sufficient responses for each sex. Analysis suggested that the pattern of change in computer and Internet self-efficacy is similar for males and females (see Table 8): both improved significantly for all students. Computer anxiety significantly influenced computer and Internet self-efficacy for both male and female respondents. User attitude significantly influenced computer and Internet self-efficacy for male respondents only. User attitude significantly influenced computer self-efficacy, but not Internet self-efficacy, for female respondents. The results are mixed and nonsignificant when data are split. Results of our study represent several implications. They underlined the importance of individual differences and their effects on computer training and selfefficacy development. Evaluating individuals’ attitude toward computers may be necessary at the beginning of a training program, given the eventual influence of
attitude on the outcome. Favorable attitudes may be reinforced through user involvement in system development activities or through job enrichment and career development programs. User involvement in the design and development of IT applications is broadly accepted as an essential factor influencing user acceptance and system use. Self-efficacy development may also be influenced by other conditions, such as required versus voluntary involvement in training programs. Individuals may not want to participate if they do not feel that it will help their career objectives. College students show greater interest in courses and programs that they feel will help them with job opportunities or desired career paths. As with other such studies, there is a limitation in using student responses. 8. Conclusion We examined the effect of computer training on computer self-efficacy and Internet self-efficacy and the influence of user attitude and computer anxiety on training outcome in terms of pattern of change in computer and Internet self-efficacy. Results suggested that computer training significantly influenced computer and Internet self-efficacy development and further suggested that user attitude and computer anxiety significantly influenced computer and Internet self-efficacy development. Data analyses suggested that there was an interaction between user attitude and computer anxiety and the effect of that interaction on computer self-efficacy but not on Internet selfefficacy. These findings have implications for the design as well as the evaluation of computer training programs.
G. Torkzadeh et al. / Information & Management 43 (2006) 541–550
Appendix A. Self-efficacy, user attitude, and computer anxiety measures Computer self-efficacy I feel confident. . . working on a personal computer (microcomputer) getting software up and running using the user’s guide when help is needed entering and saving data (numbers or words) into a file escaping/exiting from a program or software calling up a data file to view on the monitor screen understanding terms/words relating to computer hardware understanding terms/words relating to computer software handling a disk correctly learning to use a variety of programs (software) making selections from an on-screen menu using a printer to make a ‘‘hardcopy’’ of my work copying a disk copying an individual file adding and deleting information from a data file moving the cursor around the monitor screen writing simple programs for the computers using the computer to write a letter or essay describing the function of computer hardware (keyboard, monitor, disk drives, computer processing unit) understanding the three stages of data processing: input, processing, output getting help for problems in the computer systems using the computer to organize information getting rid of files when they are no longer needed organizing and managing files troubleshooting computer problems Internet self-efficacy I feel confident. . . browsing the World Wide Web (WWW) surfing the World Wide Web (WWW) encrypting my email messages encrypting my email messages before sending them over the Internet decrypting email messages that I receive decrypting email messages creating a home page for the World Wide Web (WWW) making changes on a home page downloading from another computer scanning pictures to save on the computer sending a fax via the computer receiving a fax on my computer recovering a file I accidentally deleted editing (size, color) a scanned picture finding information on the World Wide Web (WWW) User attitude I like working with computers. I look forward to those aspects of my job that require me to use a computer Once I start working on the computer, I find it hard to stop Using a computer is frustrating for me I get bored quickly when working on a computer
549
Appendix A (Continued ) Computer anxiety I feel apprehensive about using computers It scares me to think that I could cause the computer to destroy a large amount of information by hitting the wrong key I hesitate to use a computer for fear of making mistakes I cannot correct Computers are somewhat intimidating to me
References [1] A. Bandura, Toward a unifying theory of behavioral change, Psychological Review 84, 1977, pp. 191–215. [2] A. Bandura, The self system in reciprocal determinism, American Psychologist 33, 1978, pp. 344–358. [3] A. Bandura, Self-efficacy in mechanism human agency, American Psychologist 1982, pp. 122–147. [4] A. Bandura, Social Foundations of Thought and Action: A Social Cognitive Theory, Prentice-Hall, Englewood Cliffs, NJ, 1986. [5] A. Bandura, Regulation of cognitive processes through perceived self-efficacy, Developmental Psychology 25, 1989, pp. 729–735. [6] J.J. Beckers, H.G. Schmidt, The structure of computer anxiety: a six-factor model, Computers in Human Behavior 17, 2001, pp. 35–49. [7] G.G. Bitter, S.J. Davis, Measuring the development of computer literacy among teachers, AEDS Journal 18, 1985, pp. 243–253. [8] R.D. Carlson, B.L. Grabowski, The effects of computer selfefficacy on direction-following behavior in computer assisted instruction, Journal of Computer-Based Instruction 16, 1992, pp. 6–11. [9] H.W. Chou, Effects of training method and computer anxiety on learning performance and self-efficacy, Computers in Human Behavior 17, 2001, pp. 51–69. [10] D.R. Compeau, C.A. Higgins, Computer self-efficacy: development of a measure and initial test, MIS Quarterly 19, 1995, pp. 189–211. [11] D.R. Compeau, C.A. Higgins, S. Huff, Social cognitive theory and individual reactions to computing technology: a longitudinal study, MIS Quarterly 23, 1999, pp. 145–158. [12] P.C. Earley, Self or Group? Cultural effects of training on selfefficacy and performance Administrative Science Quarterly 39, 1994, pp. 89–117. [13] C.A. Frayne, G.P. Latham, Application of social learning theory to employee self-management of attendance, Journal of Applied Psychology 72, 1987, pp. 387–392. [14] M. Gist, Self-efficacy: implications for organizational behavior and human resource management, Academy of Management Review 2, 1987, pp. 472–486. [15] M. Gist, The influence of training method on self-efficacy and idea generation among managers, Personnel Psychology 42, 1989, pp. 787–805. [16] M. Gist, C. Schwoerer, B. Rosen, Effects of alternative training methods on self-efficacy and performance in computer software training, Journal of Applied Psychology 74, 1989, pp. 884–891. [17] K.V. Harrington, J.C. McElroy, P.A. Morrow, Computer anxiety and computer based training: a laboratory experiment, Journal of Educational Computing Research 6, 1990, pp. 343–358. [18] R.W. Harris, Attitudes towards end-user-computing: a structural equation model, Behaviour and Information Technology 18, 1999, pp. 109–125.
550
G. Torkzadeh et al. / Information & Management 43 (2006) 541–550
[19] A. Harrison, K. Rainer, The influence of individual differences on skill in end user-computing, Journal of Management Information Systems 9, 1992, pp. 93–111. [20] J. Hattie, The computer and control over learning, Education 110, 1990, pp. 414–417. [21] R.K. Heinssen, C.R. Glass, L.A. Knight, Assessing computer anxiety: development and validation of the computer anxiety rating scale, Computers and Human Behavior 3, 1987, pp. 49–59. [22] C.M. Henry, B.D. Holtan, Introduction novices to the computer: hands-on vs. demonstration, School Science and Mathematics 87, 1987, pp. 598–607. [23] M. Igbaria, S. Parasuraman, A path analytic study of individual characteristics, computer anxiety and attitudes toward microcomputers, Journal of Management 15, 1989, pp. 373–388. [24] P. Jorde-Bloom, Self-efficacy expectations as a predictor of computer use: a look at early childhood administrator, Computers in the Schools 5, 1988, pp. 45–63. [25] B.H. Loyd, C. Gressard, Reliability and factorial validity of computer attitude scales, Educational and Psychological Measurement 44, 1984, pp. 5001–5505. [26] G.M. Marakas, Y.Y. Mun, R.D. Johnson, The multilevel and multifaceted character of computer self-efficacy: toward clarification of the construct and an integrative framework for research, Information Systems Research 9, 1998, pp. 126–163. [27] G.A. Marcoulides, The relationship between computer anxiety and computer achievement, Journal of Educational Computer Research 4, 1988, pp. 151–157. [28] I.T. Miura, The relationship of self-efficacy expectations to computer interest and course enrollment in college, Sex Roles 16, 1987, pp. 303–311. [29] C. Murphy, D. Coover, S. Owen, Development and validation of the computer self-efficacy scale, Educational and Psychological Measurement 49, 1989, pp. 893–899. [30] T.A. Oliver, F. Shapiro, Self-efficacy and computers, Journal of Computer-Based Instruction 20, 1993, pp. 81–85. [31] M. O’Malley, T. Kelleher, Papayas and pedagogy: geographically dispersed teams and Internet self-efficacy, Public Relations Review 28, 2002, pp. 175–184. [32] C. Ong, J. Lai, Y. Wang, Factors affecting engineers’ acceptance of asynchronous e-learning systems in high-tech companies, Information and Management 41, 2004, pp. 795–804. [33] S.V. Owen, Using self-efficacy in program evaluation, in: Proceedings of the paper presented at the Annual Meeting of the American Educational Research Association, San Francisco, CA, 1986. [34] P. Popovich, K. Hyde, T. Kakrajsek, C. Blumer, The development of the attitudes toward computer usage scale, Educational and Psychological Measurement 47, 1987, pp. 261–269. [35] P. Schumacher, J. Morahan-Martin, Gender, Internet and computer attitudes and experiences, Computers in Human Behavior 17, 2001, pp. 95–110. [36] D.H. Schunk, Self-efficacy perspective on achievement behavior, Educational Psychologist 19, 1984, pp. 48–58. [37] D.H. Schunk, Self-efficacy, Cognitive skill learning, in: C. Ames, R. Aims (Eds.), Research on Motivation in Education, (vol. 3), Academic Press, San Diego, CA, 1989. [38] G. Torkzadeh, I. Angulo, The concept and correlates of computer anxiety, Behaviour and Information Technology 11, 1992, pp. 99– 108. [39] G. Torkzadeh, K. Pflughoeft, L. Hall, Computer self-efficacy, training effectiveness and user attitudes: an empirical study, Behaviour and Information Technology 18, 1999, pp. 299–309.
[40] G. Torkzadeh, T.P. Van Dyke, Development and validation of an Internet self-efficacy scale, Behaviour and Information Technology 20, 2001, pp. 275–280. [41] G. Torkzadeh, T.P. Van Dyke, Effects of training on Internet selfefficacy and computer user attitudes, Computers in Human Behavior 18, 2002, pp. 479–494. [42] V. Venkatesh, F.D. Davis, A model of the antecedents of perceived ease of use: development and test, Decision Sciences 27, 1996, pp. 451–481. [43] V. Venkatesh, M.G. Morris, G.B. Davis, F.D. Davis, User acceptance of information technology: toward a unified view, MIS Quarterly 27, 2003, pp. 425–478. [44] R.E. Wood, A. Bandura, Social cognitive theory of organizational management, Academy of Management Review 14, 1989, pp. 361–384. [45] E. Zoltan, A. Chapanis, What do professional persons think about computers? Behaviour and Information Technology 1, 1982, pp. 55–68. Gholamreza Torkzadeh is Professor and Chair of MIS at the University of Nevada, Las Vegas. He has published on management information systems issues in academic and professional journals including Management Science, Information Systems Research, MIS Quarterly, Communications of the ACM, Decision Sciences, Journal of MIS, Omega, Journal of Operational Research, Information & Management, Journal of Knowledge Engineering, Educational and Psychological Measurement, Behaviour & Information Technology, Long Range Planning, and others. His current research interests include the impact of information technology, measuring e-commerce success, computer self-efficacy, and information systems security. He holds a Ph.D. in Operations Research from The University of Lancaster, England and is a member of The Institute for Operations Research and the Management Science, Association for Information Systems, and Decision Sciences Institute. Jerry Cha-Jan Chang is Assistant Professor in the Department of MIS, College of Business, University of Nevada, Las Vegas. He has a BS in Oceanography from National Ocean University (Taiwan), an MS in Computer Science from Central Michigan University, an MBA from Texas A&M University, an MS in MIS and Ph.D. in MIS from the University of Pittsburgh. His research interests include performance measurement, management of IS, computer self-efficacy, software piracy, IS offshoring, and organizational learning. His articles have appeared in MIS Quarterly, Journal of MIS, Information & Management, Decision Support Systems, DATA BASE, Communications of the ACM, and others. Didem Demirhan is Assistant Professor in the Department of MIS, College of Business, University of Nevada, Las Vegas. She has a BA in Business Administration from Bogazici University (Turkey) and Ph.D. in MS and IS from the University of Texas at Dallas. Her research interests include IT investment and technology cost. Her research articles have appeared in Journal of MIS and JITTA.
A contingency model of computer and Internet self-efficacy Gholamreza Torkzadeh *, Jerry Cha-Jan Chang, Didem Demirhan Department of MIS, College of Business and Economics, University of Nevada, Las Vegas, USA Accepted 9 February 2006
Abstract Information system researchers have recently devoted considerable attention to the concept of computer self-efficacy in order to understand computer user behavior and system use. This article reports on the development and examination of a contingency model of computer and Internet self-efficacy. User attitude and computer anxiety were assumed to influence the development of computer and Internet self-efficacy. Measures of user attitude, computer anxiety, computer self-efficacy, and Internet self-efficacy were used in a university environment to collect 347 responses at both the beginning and end of an introductory computer course. Results suggested that training significantly improved computer and Internet self-efficacy. Respondents with ‘favorable’ attitudes toward computers improved their self-efficacy significantly more than respondents with ‘unfavorable’ attitudes. Respondents with ‘low’ computer anxiety improved their self-efficacy significantly more than respondents with ‘high’ computer anxiety. The interaction effect between attitude and anxiety was significant for computer self-efficacy scores but not for Internet self-efficacy scores. The implications of these findings are discussed. # 2006 Elsevier B.V. All rights reserved. Keywords: Computer self-efficacy; Internet self-efficacy; Computer anxiety; Computer user attitude; Computer user training; Gender; Contingency model
1. Introduction The ultimate question about information technology effectiveness relates to its impact on the individual and organizations. The effective use of an IS is influenced by not only system design features but also by the user’s ability to use the system effectively in making decisions, plan work, service customers, or control events. Self-efficacy reflects the belief that individuals have about their ability to use systems effectively. Research studies suggest that the higher the induced level of self-efficacy, the greater is performance
* Corresponding author at: Department of MIS, College of Business and Economics, University of Nevada, Las Vegas, 4505 Maryland Parkway, Box 456034, NV, USA. Tel.: +1 702 895 3796; fax: +1 702 895 0802. E-mail address: [email protected] (G. Torkzadeh). 0378-7206/$ – see front matter # 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.im.2006.02.001
achievement [1]. Individuals with high self-efficacy work harder and longer than individuals with low selfefficacy [44]. Computer self-efficacy is defined as an individual’s belief regarding their ability to use a computer [10]. Research suggests that it plays a significant role in an individual’s decision to use computers and how comfortable users are in learning skills related to effective use [26,32]. MIS researchers have, in recent years, devoted much effort in studying computer user training, user attitude, and computer anxiety as they relate to computer self-efficacy. Marakas et al. provided a comprehensive review of related literature and the path that research on computer self-efficacy has traveled. Although computer self-efficacy constructs have been the subject of research studies, contingency models that examined the influencing effect of user attitude and computer anxiety on computer self-efficacy
542
G. Torkzadeh et al. / Information & Management 43 (2006) 541–550
have not been considered. User attitude and computer anxiety are important variables and are expected to influence the outcome of self-efficacy development efforts [27,39]. Improving our understanding about influence on training programs should help in making better decisions regarding technology implementation, acceptance, and use. In our study, the pattern of change in computer and Internet self-efficacy was examined as individuals learned about computers and interacted with them. Survey responses were collected from 347 students at the beginning and end of an introductory computer course. Data were analyzed to examine the relationship between training and self-efficacy and how this relationship was influenced by user attitude and computer anxiety. Our specific goals were to examine the relationships between training and (a) computer self-efficacy, (b) Internet self-efficacy, (c) computer self-efficacy controlling for user attitude, (d) Internet self-efficacy controlling for user attitude, (e) computer self-efficacy controlling for computer anxiety, (f) Internet self-efficacy controlling for computer anxiety, (g) computer self-efficacy controlling for the interaction effect between attitude and anxiety, and (h) Internet self-efficacy controlling for the interaction effect between attitude and anxiety. 2. Self-efficacy construct Research on self-efficacy concept has a long tradition in social sciences with notable works by Bandura [2,4,5], Schunk [36,37], Gist [14,15], and others. Self-efficacy arises from the gradual acquisition of complex cognitive, social, linguistic and/or physical skills through experience [3]. It is expected to affect task effort, persistence, expressed interest, and the level of goal difficulty selected for performance. Individuals appear to evaluate information about their abilities and then regulate their choices and efforts accordingly. The strength of their conviction in their own effectiveness is likely to influence whether they will try to cope with a given situation. Individuals with high efficacy expectations have a greater chance of success in a given task [30]. Self-efficacy is a dynamic construct that changes as new information and experiences are acquired. It is generally described as having three components: magnitude—the levels of task difficulty that people believe they can attain; strength—their conviction about its magnitude; and generality—the degree to which the expectation is generalized across situations. The purpose in evaluating these components is to discover
the type of questions that will best explain and predict a person’s dispositions, intentions, and actions. Schunk described how self-efficacy influenced academic learning processes. At the start of an activity, students hold differing beliefs about their ability to acquire knowledge, perform skills, master the material, etc. Initial self-efficacy varies as a function of aptitude (such as abilities and attitudes) and prior experience. Such personal factors as goal setting and information processing, along with situational factors, such as rewards and teacher feedback, affect students while they are learning. From these, students derive cues signaling how well they are learning, which they use to assess efficacy for further learning. Motivation is enhanced if students perceive that they are making progress in learning. In turn, as students work on tasks and become more skillful, they maintain a sense of self-efficacy for performing well. Thus training improves self-efficacy and individual differences moderate the outcome of this training: indeed computer training may be more effective for people who have a positive attitude toward computers and no computer anxiety. This suggests a contingency model of self-efficacy where individual differences influence learning outcomes. 3. Research model and hypotheses In our contingency model, user attitude and computer anxiety were the conditional variables that moderated the outcome of computer training. Here, we measured training outcome in terms of pattern of change in computer and Internet self-efficacy as trainees went through a semester long computer course. Part (a) in Fig. 1 shows our assumed moderating effect of computer anxiety and user attitude on computer and Internet selfefficacy. Part (b) shows the assumed moderating effect of the interaction between computer anxiety and user attitude on computer and Internet self-efficacy. Research studies in cognitive modeling, behavioral modeling, and self-management suggest that training programs enhance self-efficacy [12,13,16]. MIS research has focused on computer self-efficacy as a special case of general self-efficacy. MIS researchers, similarly, have illustrated the importance of training for computer self-efficacy. Training is also suggested to enhance Internet self-efficacy, considered an extension of computer self-efficacy construct in the domain of World Wide Web [31]. Thus we proposed: Hypothesis 1. Computer training improves computer self-efficacy.
G. Torkzadeh et al. / Information & Management 43 (2006) 541–550
543
Fig. 1. A contingency model of self-efficacy.
Hypothesis 2. Computer training improves Internet self-efficacy. Studies in the field of psychology suggest that people experience anxiety when performing behaviors that they do not feel they can perform competently. When trainees experience anxiety, they cannot concentrate on learning; they focus on inner feelings and thoughts [9,38]. Studies have suggested that computer anxiety reduces the effectiveness of computer-based training [17,22]. We, therefore, expected computer anxiety to adversely affect training outcome measured in terms of improvement in computer or Internet self-efficacy. Thus, we proposed: Hypothesis 3. Computer anxiety negatively influences computer training outcome measured in terms of pattern of change in computer self-efficacy. Hypothesis 4. Computer anxiety negatively influences computer training outcome measured in terms of pattern of change in Internet self-efficacy. User attitude also plays an important role in training and learning. Studies have suggested that user attitude toward computer influenced the outcome of training programs [7,41,45]. A positive user attitude toward computer is expected to positively influence the outcome of training programs and the level of selfefficacy improvement. Thus, we proposed: Hypothesis 5. User attitude influences computer training outcome measured in terms of pattern of change in computer self-efficacy.
Hypothesis 6. User attitude influences computer training outcome measured in terms of pattern of change in Internet self-efficacy. We have argued that anxiety and attitude, separately, influence the outcome of computer training. Research also suggested a significant relationship between computer anxiety and user attitude [6,18,23]. For example, Harris suggested that computer anxiety had negative influence on attitude toward end-user-computing. However, this relationship has not been widely tested and there is thus a need to examine the interaction effects between computer anxiety and user attitude on computer training. We expected the interaction between anxiety and attitude to influence the computer training outcome and thus constructed the following hypotheses: Hypothesis 7. The interaction between computer anxiety and user attitude influences computer training outcome measured in terms of pattern of change in computer self-efficacy. Hypothesis 8. The interaction between computer anxiety and user attitude influences computer training outcome measured in terms of pattern of change in Internet self-efficacy. These hypotheses are summarized in Table 1, which also provides literature support for each. 4. The instruments Arguably the most important aspect of computer self-efficacy to MIS research is the development of
544
G. Torkzadeh et al. / Information & Management 43 (2006) 541–550
Table 1 Summary of hypothesis and supporting literature Hypothesis
Statement
Literature support
H1 and H2 H3 and H4
Training improves computer and Internet self-efficacy Computer anxiety influences training outcome in terms of pattern of change in computer self-efficacy and Internet self-efficacy Computer user attitude influences training outcome in terms of pattern of change in computer self-efficacy and Internet self-efficacy The interaction between computer anxiety and user attitude influences training outcome in terms of pattern of change in computer self-efficacy and Internet self-efficacy
[10,12,13,15,16,26,31,39] [9,17,22,7]
H5 and H6 H7 and H8
valid and reliable measures that can be used with confidence to accumulate findings and facilitate substantive hypotheses testing. We have not developed a sufficient pool of research instruments in this domain and this has hampered research on computer self-efficacy. Two broad perspectives have emerged for measures of computer self-efficacy. One encourages developing task specific measures, while another encourages developing general measures. Works of Murphy et al. [29] and Compeau et al. [11] are examples of general and task specific measures, respectively. Marakas et al. suggested the need for specific computer self-efficacy measures while Torkzadeh et al. validated and used general measures. The argument for use of specific measures of computer self-efficacy suggested that it was important to align computer self-efficacy with the particular task domain. Such a strong alignment would be likely to improve the exploratory power of the instruments. This perspective served to motivate the development of measures that assessed the individual’s level of selfefficacy with specific task such as the use of a spreadsheet, suggesting that a new measure is needed every time a new area is explored. The argument for general measures of computer self-efficacy is that it is difficult to establish a research tradition if new measures are to be developed every time research is attempted: task specific measures may not be stable over time for use in longitudinal or follow up studies. This perspective has served to motivate the development of general measures that are likely to be stable over time and applicable for longitudinal studies. The inherent nature of computer technology involves constant and often rapid change, most of which is not accurately predicted. This unpredictable nature has posed challenges for research in IS in general and instrument development in particular. This does not mean that existing measures should not be used, but that they must be used with caution and their reliability must be reported each time they are used.
[7,39,40,45] [6,18,23]
4.1. Computer self-efficacy We used an instrument developed by Murphy et al. based on the work of Bandura, in which he conceptualized self-efficacy as individualized selfperception varying across activities and situational circumstances, rather than as a goal disposition that can be assessed by an omnibus test. Domain specific measures, such as computer self-efficacy, allow researchers to be more accurate in evaluating computer training outcome than general instruments. Owen [33] also suggested that self-efficacy could be reliably measured and that such measures might be used to assess a composite of affect, cognition, and performance in the attainment of program and course objectives. Murphy et al. suggested that their instrument could be used to evaluate skill attainment at both preand post-training. Later, Harrison and Rainer [19] used this instrument to measure the respondent’s perceptions regarding specific computer-related knowledge and skills and reported an overall reliability coefficient of 0.95 for the instrument. In a more recent study, Torkzadeh et al. used it in a study of training and computer self-efficacy and also reported a reliability of 0.95. Our study used 25 of the original 32-item instrument developed by Murphy et al. (see Appendix A). Items related to mainframe system were deemed irrelevant here, and were dropped. 4.2. Internet self-efficacy The computer self-efficacy instrument pre-dated the rise and importance of Internet related skills. In a more recent study, Torkzadeh and Van Dyke [40] developed a 17-item instrument for measuring the individual’s self-perception and self-competency in interacting with the Internet. They based their measurement development on Bandura’s conceptualization of self-efficacy and other studies of social and cognitive psychology. Later, they used this 17-item instrument in a study of training and Internet self-
G. Torkzadeh et al. / Information & Management 43 (2006) 541–550
efficacy and reported a reliability of 0.96. We used 15 items of the instrument developed by Torkzadeh and Van Dyke for measuring Internet self-efficacy. Two items in the original instrument were deemed redundant and were dropped. Both computer selfefficacy and Internet self-efficacy instruments used a five point Likert-type scale where 1 = strongly disagree and 5 = strongly agree. 4.3. Computer anxiety and user attitude Our study used measures of computer anxiety developed by Heinssen et al. [21] and used by Compeau et al. to study computer self-efficacy and system use. It also used measures of user attitude developed by Loyd and Gressard [25] and used by Compeau et al. In their study, Compeau et al. used five items for measuring user attitude or ‘‘affective response’’ and four for measuring computer anxiety. Affect or user attitude, they argued, represented the positive side (where a person enjoyed interacting with a computer) and anxiety represented the negative side (where a person experienced feelings of apprehension or anxiety when interacting). Compeau et al. used partial least square tests (PLS) to assess their measurement model and reported internal consistency greater than 0.7 for measures of computer anxiety and user attitude. We used the five items of user attitude and four items of computer anxiety of Compeau et al. Both computer anxiety and user attitude instruments used a five point Likert-type scale, where 1 = strongly disagree and 5 = strongly agree. 5. The study A survey instrument containing measures of computer self-efficacy, Internet self-efficacy, user attitude, and computer anxiety was administered to business undergraduates at a large state university in the southwest region of the United States. Students in multiple sections of an introductory course participated in the study. The course is required for all business students. Class sizes ranged from 45 to 60. Course content and grade distribution were similar for all sections. The course covered information technology infrastructure, management and organizational support systems, information systems in the enterprise, electronic business, telecommunications and networks, management of hardware and software, file and database management, systems development, decision support systems, and aspects of a computer career. The measures were administered at both the beginning and end of the course. The time between
545
pre- and post-test ranged from 11 to 12 weeks. Respondents had no prior knowledge that they would be asked to repeat the survey at the end and were asked to respond to questions focusing on their current belief. Participation in the study was voluntary; student names were used to match pre- and post-training responses. A total of 347 usable matching responses consisting of 201 male (58%) and 146 female students were obtained. Incomplete responses and those with matching problems were omitted. The survey also asked respondents about their educational interest and age. The sample was diverse in terms of respondents’ educational interest. They were majoring in accounting (19.7%), MIS (18.3%), management (16.8%), finance (15.7%), marketing (13.0%), economics (2.3%), and other. The respondents’ age distribution was: under 20 years (10.1%), 20–29 (77.8%), 30–39 (9.5%), and greater than 40 (2.6%). This sample was considered large enough to confirm the measures and test the research hypotheses. 6. Data analysis and results The reliability (internal consistency) of items in each scale was examined using Cronbach’s alpha to confirm the adequacy of the measures for testing the hypotheses (see Table 2). These coefficients confirmed results of earlier studies and provided confidence in testing the hypotheses. The reliability results are based on pretraining data. However, post-training data had very similar results. Respondents were grouped as people with ‘high’ or ‘low’ computer anxiety using the mean score of 1.95 (S.D. = 0.83) on the pre-test computer anxiety scale. Similarly, respondents were grouped as people with ‘favorable’ or ‘unfavorable’ attitude toward computers using the mean score of 3.98 (S.D. = 0.74) on the pre-test computer user attitude scale. Overall, computer user attitude scores were high for all respondents. This is consistent with recent studies and different from what was reported in earlier studies [34], suggesting an upward change in user’s attitude toward computers over the years. Table 2 Reliability of measures Factors
No. of items
Alpha
Computer self-efficacy Internet self-efficacy Attitude Anxiety
25 15 5 4
0.95 0.91 0.81 0.81
546
G. Torkzadeh et al. / Information & Management 43 (2006) 541–550
Table 3 The effect of training on computer and Internet self-efficacy
Computer self-efficacy Internet self-efficacy
Pre, mean (S.D.)
Post, mean (S.D.)
4.24 (0.57) 3.59 (0.81)
4.34 (0.58) 3.77 (0.79)
Source
Table 3 provides pre- and post-training scores for computer self-efficacy and Internet self-efficacy for all respondents. The score for each scale was equal to the total points divided by the number of items. The difference in mean scores was examined using a paired t-test procedure. Both differences in mean scores were significant at p < 0.001. The results suggested that respondents entered training programs with a relatively high level of computer and Internet self-efficacy but that the levels improved significantly after the course. These findings supported Hypotheses 1 and 2. Data were also analyzed to examine the influence of computer anxiety and user attitude on computer and Internet self-efficacy as students went through the training program. The ANOVA method was used to examine the difference between subject responses as well as possible interaction effects of computer anxiety and user attitude on computer and Internet selfefficacy. Pre- and post-training data were used to examine the pattern of change in computer and Internet self-efficacy. Table 4 presents the ANOVA results; they suggested computer and Internet self-efficacy scores were significantly different for respondents with ‘high’ and ‘low’ computer anxiety. In the between subjects tests, there were significant main effects of computer anxiety on both computer and Internet selfefficacy. In the within-subjects tests, there was a significant main effect for computer and Internet selfefficacy; consistent with the t-test results. The interaction effect for within-subjects is not applicable Table 4 The effect of computer anxiety on computer and Internet self-efficacy
Within subject effects Training Training anxiety Between subject effects Anxiety
Sig. 4.25 5.22
0.000 0.000
Table 5 The effect of user attitude on computer and Internet self-efficacy
6.1. Individual differences and self-efficacy development
Source
t
Computer S.E.
Internet S.E.
F-value
Sig.
F-value
Sig.
16.16 5.17
0.000 0.024
27.47 0.29
0.000 0.593
146.80
0.000
59.93
0.000
Within subject effects Training Training attitude Between subject effects Attitude
Computer S.E.
Internet S.E.
F-value
Sig.
F-value
Sig.
20.82 5.84
0.000 0.016
25.63 0.61
0.000 0.436
125.20
0.000
55.00
0.000
here; we used the interaction effect for betweensubjects throughout the paper. These results supported Hypotheses 3 and 4. Table 5 presents the ANOVA results that suggested that computer and Internet self-efficacy scores were significantly different for respondents with ‘favorable’ and ‘unfavorable’ attitudes toward computers. In the within-subject tests, there was a significant main effect for both computer self-efficacy and Internet selfefficacy. In the between-subjects tests, there were significant main effects of user attitude on both computer self-efficacy and Internet self-efficacy. These results supported Hypotheses 5 and 6. 6.2. Interaction effects Table 6 presents the ANOVA results; they suggested that the interaction between user attitude and
Table 6 The interaction effect on computer and Internet self-efficacy training Source
Within subject effects Training Training attitude Training anxiety Training attitude anxiety Between subject effects Anxiety Attitude Attitude anxiety
Computer S.E.
Internet S.E.
F-value
Sig.
F-value
Sig.
14.19 2.45 2.19 0.06
0.000 0.119 0.140 0.811
17.88 0.64 0.00 0.61
0.000 0.423 0.970 0.435
88.25 53.42 4.08
0.000 0.000 0.044
33.22 20.02 2.89
0.000 0.000 0.090
G. Torkzadeh et al. / Information & Management 43 (2006) 541–550
547
Table 7 Scenarios for individual differences
computer anxiety influenced the outcome of training for computer self-efficacy (at 0.05) but not for Internet self-efficacy. These results supported Hypotheses 7 but not 8. Results supported earlier analyses. Data were analyzed to further examine the interaction effects of possible scenarios for individual differences. Respondents were grouped as having ‘high’ or ‘low’ computer anxiety as well as having ‘favorable’ or ‘unfavorable’ attitude toward computers. This enabled us to analyze four possible scenarios of individual differences. Table 7 depicts these scenarios and shows the sample size for each. The sample size is sufficiently large to warrant test of difference in mean scores for the pattern of change in computer and Internet self-efficacy as individuals go through training. Both Tukey and Scheffe multiple comparisons procedures were used to test differences between possible mean scores; they produced similar results. Among the four possible situations, scenarios 2 and 3 were expected to have the most influence on selfefficacy development, in an opposite direction. Results suggested significant differences (at 0.01) between mean score for scenario 2 and for the other three scenarios for both computer and Internet self-efficacy. Results also suggested significant differences (at 0.01) between mean score for scenario 3 and for the other three scenarios for computer self-efficacy and scenarios 1 and 2 of Internet self-efficacy; there was no significant difference between mean scores for scenario 3 and 4 for Internet self-efficacy. 7. Discussion User acceptance and the effective use of information technology have been considered essential success factors for technology management [43]. In order to manage information technology better, we need to
understand why individuals readily accept and use computer applications, persist at improving their computer skills, select challenging projects, search for innovative and new ways of using computer systems, and eventually have high rates of success. The construct of computer self-efficacy is thus important to our understanding of computer user behavior: it helps predict user perceptions and subsequent acceptance and use of computer systems [42]. There has been increased research to help understand computer self-efficacy construct and its correlates. Studies have examined the role of training in computer self-efficacy development. The results of our study support earlier findings that training programs influence self-efficacy. More specifically, we examined the role of training in computer and Internet selfefficacy. We examined the influence of individual differences on the outcome of self-efficacy development. User attitude and computer anxiety were shown to influence the pattern of change in computer and Internet self-efficacy as students went through a computer course. The results suggested that individuals with ‘favorable’ attitudes toward computers improved their computer and Internet self-efficacy significantly more than individuals with ‘unfavorable’ attitudes. The results also suggested that individuals with ‘low’ computer anxiety improved their computer and Internet self-efficacy significantly more than individuals with ‘high’ computer anxiety. The results support the hypothesis that the interaction between user attitude and computer anxiety influence the level of improvement in computer and Internet self-efficacy. Individuals with ‘low’ computer anxiety and ‘favorable’ attitudes toward computers improved their computer and Internet self-efficacy significantly more than any other group. In contrast, individuals with ‘high’ computer anxiety
548
G. Torkzadeh et al. / Information & Management 43 (2006) 541–550
Table 8 Interaction effect by gender Source
Computer self-efficacy Male (n = 201) F-value
Within subject effects Training Training anxiety Training attitude Training attitude anxiety Between subject effects Anxiety Attitude Attitude anxiety
Sig.
Internet self-efficacy Female (n = 146)
Male (n = 201)
Female (n = 146)
F-value
Sig.
F-value
Sig.
F-value
Sig.
6.24 0.37 5.76 0.21
0.013 0.545 0.017 0.645
5.65 2.89 0.37 0.13
0.019 0.091 0.542 0.724
10.09 0.00 0.14 0.07
0.002 0.963 0.707 0.798
5.73 0.01 3.19 1.59
0.018 0.906 0.076 0.209
58.47 59.54 3.22
0.000 0.000 0.074
30.00 5.52 2.24
0.000 0.020 0.137
21.71 33.50 1.39
0.000 0.000 0.239
11.99 0.05 3.33
0.001 0.817 0.070
and ‘unfavorable’ attitudes toward computers improved their computer and Internet self-efficacy less than any other group. Analyses of scenarios for individual differences suggested that computer anxiety exerted more influence on self-efficacy development than did user attitude. Studies have also examined gender issues in selfefficacy development. Earlier findings suggested that males reported higher self-efficacy than females [8,24,28] or males feel more in control when interacting with computers [20]. More recent studies suggested that there were no significant differences in responses for males and females or that the gender difference in computer skill is diminishing [35]. It seemed appropriate to examine the relationship between training and self-efficacy for males and females here, especially since the sample included sufficient responses for each sex. Analysis suggested that the pattern of change in computer and Internet self-efficacy is similar for males and females (see Table 8): both improved significantly for all students. Computer anxiety significantly influenced computer and Internet self-efficacy for both male and female respondents. User attitude significantly influenced computer and Internet self-efficacy for male respondents only. User attitude significantly influenced computer self-efficacy, but not Internet self-efficacy, for female respondents. The results are mixed and nonsignificant when data are split. Results of our study represent several implications. They underlined the importance of individual differences and their effects on computer training and selfefficacy development. Evaluating individuals’ attitude toward computers may be necessary at the beginning of a training program, given the eventual influence of
attitude on the outcome. Favorable attitudes may be reinforced through user involvement in system development activities or through job enrichment and career development programs. User involvement in the design and development of IT applications is broadly accepted as an essential factor influencing user acceptance and system use. Self-efficacy development may also be influenced by other conditions, such as required versus voluntary involvement in training programs. Individuals may not want to participate if they do not feel that it will help their career objectives. College students show greater interest in courses and programs that they feel will help them with job opportunities or desired career paths. As with other such studies, there is a limitation in using student responses. 8. Conclusion We examined the effect of computer training on computer self-efficacy and Internet self-efficacy and the influence of user attitude and computer anxiety on training outcome in terms of pattern of change in computer and Internet self-efficacy. Results suggested that computer training significantly influenced computer and Internet self-efficacy development and further suggested that user attitude and computer anxiety significantly influenced computer and Internet self-efficacy development. Data analyses suggested that there was an interaction between user attitude and computer anxiety and the effect of that interaction on computer self-efficacy but not on Internet selfefficacy. These findings have implications for the design as well as the evaluation of computer training programs.
G. Torkzadeh et al. / Information & Management 43 (2006) 541–550
Appendix A. Self-efficacy, user attitude, and computer anxiety measures Computer self-efficacy I feel confident. . . working on a personal computer (microcomputer) getting software up and running using the user’s guide when help is needed entering and saving data (numbers or words) into a file escaping/exiting from a program or software calling up a data file to view on the monitor screen understanding terms/words relating to computer hardware understanding terms/words relating to computer software handling a disk correctly learning to use a variety of programs (software) making selections from an on-screen menu using a printer to make a ‘‘hardcopy’’ of my work copying a disk copying an individual file adding and deleting information from a data file moving the cursor around the monitor screen writing simple programs for the computers using the computer to write a letter or essay describing the function of computer hardware (keyboard, monitor, disk drives, computer processing unit) understanding the three stages of data processing: input, processing, output getting help for problems in the computer systems using the computer to organize information getting rid of files when they are no longer needed organizing and managing files troubleshooting computer problems Internet self-efficacy I feel confident. . . browsing the World Wide Web (WWW) surfing the World Wide Web (WWW) encrypting my email messages encrypting my email messages before sending them over the Internet decrypting email messages that I receive decrypting email messages creating a home page for the World Wide Web (WWW) making changes on a home page downloading from another computer scanning pictures to save on the computer sending a fax via the computer receiving a fax on my computer recovering a file I accidentally deleted editing (size, color) a scanned picture finding information on the World Wide Web (WWW) User attitude I like working with computers. I look forward to those aspects of my job that require me to use a computer Once I start working on the computer, I find it hard to stop Using a computer is frustrating for me I get bored quickly when working on a computer
549
Appendix A (Continued ) Computer anxiety I feel apprehensive about using computers It scares me to think that I could cause the computer to destroy a large amount of information by hitting the wrong key I hesitate to use a computer for fear of making mistakes I cannot correct Computers are somewhat intimidating to me
References [1] A. Bandura, Toward a unifying theory of behavioral change, Psychological Review 84, 1977, pp. 191–215. [2] A. Bandura, The self system in reciprocal determinism, American Psychologist 33, 1978, pp. 344–358. [3] A. Bandura, Self-efficacy in mechanism human agency, American Psychologist 1982, pp. 122–147. [4] A. Bandura, Social Foundations of Thought and Action: A Social Cognitive Theory, Prentice-Hall, Englewood Cliffs, NJ, 1986. [5] A. Bandura, Regulation of cognitive processes through perceived self-efficacy, Developmental Psychology 25, 1989, pp. 729–735. [6] J.J. Beckers, H.G. Schmidt, The structure of computer anxiety: a six-factor model, Computers in Human Behavior 17, 2001, pp. 35–49. [7] G.G. Bitter, S.J. Davis, Measuring the development of computer literacy among teachers, AEDS Journal 18, 1985, pp. 243–253. [8] R.D. Carlson, B.L. Grabowski, The effects of computer selfefficacy on direction-following behavior in computer assisted instruction, Journal of Computer-Based Instruction 16, 1992, pp. 6–11. [9] H.W. Chou, Effects of training method and computer anxiety on learning performance and self-efficacy, Computers in Human Behavior 17, 2001, pp. 51–69. [10] D.R. Compeau, C.A. Higgins, Computer self-efficacy: development of a measure and initial test, MIS Quarterly 19, 1995, pp. 189–211. [11] D.R. Compeau, C.A. Higgins, S. Huff, Social cognitive theory and individual reactions to computing technology: a longitudinal study, MIS Quarterly 23, 1999, pp. 145–158. [12] P.C. Earley, Self or Group? Cultural effects of training on selfefficacy and performance Administrative Science Quarterly 39, 1994, pp. 89–117. [13] C.A. Frayne, G.P. Latham, Application of social learning theory to employee self-management of attendance, Journal of Applied Psychology 72, 1987, pp. 387–392. [14] M. Gist, Self-efficacy: implications for organizational behavior and human resource management, Academy of Management Review 2, 1987, pp. 472–486. [15] M. Gist, The influence of training method on self-efficacy and idea generation among managers, Personnel Psychology 42, 1989, pp. 787–805. [16] M. Gist, C. Schwoerer, B. Rosen, Effects of alternative training methods on self-efficacy and performance in computer software training, Journal of Applied Psychology 74, 1989, pp. 884–891. [17] K.V. Harrington, J.C. McElroy, P.A. Morrow, Computer anxiety and computer based training: a laboratory experiment, Journal of Educational Computing Research 6, 1990, pp. 343–358. [18] R.W. Harris, Attitudes towards end-user-computing: a structural equation model, Behaviour and Information Technology 18, 1999, pp. 109–125.
550
G. Torkzadeh et al. / Information & Management 43 (2006) 541–550
[19] A. Harrison, K. Rainer, The influence of individual differences on skill in end user-computing, Journal of Management Information Systems 9, 1992, pp. 93–111. [20] J. Hattie, The computer and control over learning, Education 110, 1990, pp. 414–417. [21] R.K. Heinssen, C.R. Glass, L.A. Knight, Assessing computer anxiety: development and validation of the computer anxiety rating scale, Computers and Human Behavior 3, 1987, pp. 49–59. [22] C.M. Henry, B.D. Holtan, Introduction novices to the computer: hands-on vs. demonstration, School Science and Mathematics 87, 1987, pp. 598–607. [23] M. Igbaria, S. Parasuraman, A path analytic study of individual characteristics, computer anxiety and attitudes toward microcomputers, Journal of Management 15, 1989, pp. 373–388. [24] P. Jorde-Bloom, Self-efficacy expectations as a predictor of computer use: a look at early childhood administrator, Computers in the Schools 5, 1988, pp. 45–63. [25] B.H. Loyd, C. Gressard, Reliability and factorial validity of computer attitude scales, Educational and Psychological Measurement 44, 1984, pp. 5001–5505. [26] G.M. Marakas, Y.Y. Mun, R.D. Johnson, The multilevel and multifaceted character of computer self-efficacy: toward clarification of the construct and an integrative framework for research, Information Systems Research 9, 1998, pp. 126–163. [27] G.A. Marcoulides, The relationship between computer anxiety and computer achievement, Journal of Educational Computer Research 4, 1988, pp. 151–157. [28] I.T. Miura, The relationship of self-efficacy expectations to computer interest and course enrollment in college, Sex Roles 16, 1987, pp. 303–311. [29] C. Murphy, D. Coover, S. Owen, Development and validation of the computer self-efficacy scale, Educational and Psychological Measurement 49, 1989, pp. 893–899. [30] T.A. Oliver, F. Shapiro, Self-efficacy and computers, Journal of Computer-Based Instruction 20, 1993, pp. 81–85. [31] M. O’Malley, T. Kelleher, Papayas and pedagogy: geographically dispersed teams and Internet self-efficacy, Public Relations Review 28, 2002, pp. 175–184. [32] C. Ong, J. Lai, Y. Wang, Factors affecting engineers’ acceptance of asynchronous e-learning systems in high-tech companies, Information and Management 41, 2004, pp. 795–804. [33] S.V. Owen, Using self-efficacy in program evaluation, in: Proceedings of the paper presented at the Annual Meeting of the American Educational Research Association, San Francisco, CA, 1986. [34] P. Popovich, K. Hyde, T. Kakrajsek, C. Blumer, The development of the attitudes toward computer usage scale, Educational and Psychological Measurement 47, 1987, pp. 261–269. [35] P. Schumacher, J. Morahan-Martin, Gender, Internet and computer attitudes and experiences, Computers in Human Behavior 17, 2001, pp. 95–110. [36] D.H. Schunk, Self-efficacy perspective on achievement behavior, Educational Psychologist 19, 1984, pp. 48–58. [37] D.H. Schunk, Self-efficacy, Cognitive skill learning, in: C. Ames, R. Aims (Eds.), Research on Motivation in Education, (vol. 3), Academic Press, San Diego, CA, 1989. [38] G. Torkzadeh, I. Angulo, The concept and correlates of computer anxiety, Behaviour and Information Technology 11, 1992, pp. 99– 108. [39] G. Torkzadeh, K. Pflughoeft, L. Hall, Computer self-efficacy, training effectiveness and user attitudes: an empirical study, Behaviour and Information Technology 18, 1999, pp. 299–309.
[40] G. Torkzadeh, T.P. Van Dyke, Development and validation of an Internet self-efficacy scale, Behaviour and Information Technology 20, 2001, pp. 275–280. [41] G. Torkzadeh, T.P. Van Dyke, Effects of training on Internet selfefficacy and computer user attitudes, Computers in Human Behavior 18, 2002, pp. 479–494. [42] V. Venkatesh, F.D. Davis, A model of the antecedents of perceived ease of use: development and test, Decision Sciences 27, 1996, pp. 451–481. [43] V. Venkatesh, M.G. Morris, G.B. Davis, F.D. Davis, User acceptance of information technology: toward a unified view, MIS Quarterly 27, 2003, pp. 425–478. [44] R.E. Wood, A. Bandura, Social cognitive theory of organizational management, Academy of Management Review 14, 1989, pp. 361–384. [45] E. Zoltan, A. Chapanis, What do professional persons think about computers? Behaviour and Information Technology 1, 1982, pp. 55–68. Gholamreza Torkzadeh is Professor and Chair of MIS at the University of Nevada, Las Vegas. He has published on management information systems issues in academic and professional journals including Management Science, Information Systems Research, MIS Quarterly, Communications of the ACM, Decision Sciences, Journal of MIS, Omega, Journal of Operational Research, Information & Management, Journal of Knowledge Engineering, Educational and Psychological Measurement, Behaviour & Information Technology, Long Range Planning, and others. His current research interests include the impact of information technology, measuring e-commerce success, computer self-efficacy, and information systems security. He holds a Ph.D. in Operations Research from The University of Lancaster, England and is a member of The Institute for Operations Research and the Management Science, Association for Information Systems, and Decision Sciences Institute. Jerry Cha-Jan Chang is Assistant Professor in the Department of MIS, College of Business, University of Nevada, Las Vegas. He has a BS in Oceanography from National Ocean University (Taiwan), an MS in Computer Science from Central Michigan University, an MBA from Texas A&M University, an MS in MIS and Ph.D. in MIS from the University of Pittsburgh. His research interests include performance measurement, management of IS, computer self-efficacy, software piracy, IS offshoring, and organizational learning. His articles have appeared in MIS Quarterly, Journal of MIS, Information & Management, Decision Support Systems, DATA BASE, Communications of the ACM, and others. Didem Demirhan is Assistant Professor in the Department of MIS, College of Business, University of Nevada, Las Vegas. She has a BA in Business Administration from Bogazici University (Turkey) and Ph.D. in MS and IS from the University of Texas at Dallas. Her research interests include IT investment and technology cost. Her research articles have appeared in Journal of MIS and JITTA.