Você está na página 1de 10

Computer Standards & Interfaces 33 (2011) 422–431

Contents lists available at ScienceDirect

Computer Standards & Interfaces


j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / c s i

Dimensions of self-efficacy in the study of smart phone acceptance


Kuanchin Chen a,1, Jengchung V. Chen b,2, David C. Yen c,⁎
a
Department of Business Information Systems, Haworth College of Business, Western Michigan University, Kalamazoo, MI 49008, United States
b
Institute of International Management, National Cheng Kung University, Tainan, Taiwan, ROC
c
Department of DSC & MIS, Miami University, Oxford, OH 45056, United States

a r t i c l e i n f o a b s t r a c t

Article history: This study focuses on smart phone acceptance in a major delivery service company in Taiwan. Technology
Received 8 February 2010 Acceptance Model (TAM) was tested in two different forms, the original and with self-efficacy added. A two-
Accepted 25 January 2011 dimensional improvement to the theory of self-efficacy was investigated in this research. Only assisted self-
Available online 2 February 2011
efficacy was related to both perceived ease of use and perceived usefulness, but Individual Self-efficacy was
only a predictor of perceived ease of use. This result offers new evidence to the debate of the role of self-
Keywords:
efficacy on TAM constructs. Interesting findings including a strong influence of perceived ease of use on
Technology Acceptance Model
Self-efficacy
perceived usefulness and behavioral intention were compared to prior studies to add additional insights for
Smart phone future research.
Delivery service industry © 2011 Elsevier B.V. All rights reserved.

1. Introduction technology. In addition, the same report predicts smart phones to


have a compound annual growth rate of 29% through 2009.
Technology advances have made personal, technological gadgets The growth of smart phone sales seems inevitable, but this
smaller, more powerful, much easier to use, and conveniently technology has not yet offered a converged set of features. Chang,
portable. Personal communication devices have evolved from dedi- Chen and Zhou [7] suggested eleven “must-have” and eight
cated landlines for the whole household to personalized cell phone “desirable-to-have” features. Even their “must-haves” are not univer-
devices for individuals. As more of these personal technology devices sally available on all smart phones today, pointing to current market
become available and affordable for the general consumers, conve- situation that there is no one smart phone standard that is agreed
nience and needs for a more integrated solution for all these devices upon across smart phone manufacturers. Much interest from various
may very likely shape the future trend. Consequently, researchers and types of technology manufacturers (e.g., traditional cell phone
practitioners have focused on the promising development on this vendors, PDA manufacturers and even computer software/hardware
technology. One great example is the emergence of smart phones that companies) has contributed to the divergence of smart phone
integrates cell phone and personal digital assistant (PDA) technolo- features. Some of these manufacturers just build smart phones on
gies into one single device. top of their existing technological base and hence, making the
Such a mixture of two or more technologies opens new challenges products very diverse in the different features offered. As a rather
and opportunities for businesses as well as individuals. With smart recent example, Apple's iPhone includes features such as adding a
phones as an integrated solution, consumers no longer have to carry sensor to rotate contents when the device is rotated and abandoning
multiple technology devices where each device offers only a handful of keyboard/stylus user interface in favor of the use of bare fingers.
of limited functions to accomplish certain tasks. One report from The success of Apple's touch keyboard and screen rotation as a
Jupiter Research [18] indicates that 62% of consumers prefer to carry a bundled feature of their iPhone and iPad has attracted interest of
single technology device (such as smart phones) even if the features other device manufacturers. Similar features are becoming available
of the device may compromise advanced functions (e.g., size and on cell phones, tablet computers and other personal devices.
battery life). Around 74% said the key mobile feature — telephony, Another reason is that business use cases for such converged
should remain as one core component of the aforementioned device have not yet been expanded beyond basic emailing and
scheduling. A compact smart phone with loaded functions like
wireless transmission (Infrared, Bluetooth, WiFi, and WiMax), email,
camera and barcode/image recognition, voice command and recog-
⁎ Corresponding author. Tel.: +1 513 529 4826 (office); fax: +1 513 529 9689. nition, Power Point Presentation, scheduling, Web browsing, and
E-mail addresses: kc.chen@wmich.edu (K. Chen), victor@mail.ncku.edu.tw
(J.V. Chen), yendc@muohio.edu (D.C. Yen).
desktop publication tools is deserved to be seamlessly integrated and
1
Tel.: +1 269 387 5795 (office); fax: +1 269 387 5710. used professionally. A killer application to business areas has not
2
Tel.: +886 6 275 7575xt53238 (office); fax: +886 6 275 3882. been available to satisfy a broader business audience. As companies

0920-5489/$ – see front matter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.csi.2011.01.003
K. Chen et al. / Computer Standards & Interfaces 33 (2011) 422–431 423

continue to uncover their new technological integrations into the 3. Literature review
smart phones area, it is becoming increasingly difficult to precisely
define what a smart phone really is. Nonetheless, the trend of turning 3.1. Technology Acceptance Model (TAM)
it into an “all-in-one” personal communication device may reshape
the future of business communications in general and information Davis' [13,14] Technology Acceptance Model (TAM) was originat-
technology area in particular. ed from several psychological theories, including Theory of Reasoned
Action (TRA) and Theory of Planned Behavior (TPB). Both TRA and
TAM theories agree that beliefs affect a person's attitude toward a
2. Research purposes target behavior. Attitude, in turn, may influence behavioral intention,
and eventually affects the actual behavior.
Although the adoption of smart phones is gaining momentum, In TAM (Fig. 1) the actual behavior is affected by behavioral
smart phone's unique position as an integrated communication device intention (BI), and behavioral intention is directly influenced by the
with sufficient possibilities for different hardware and software perceived usefulness (PU) and attitude towards the target system.
combinations has made it a viable choice for many uses. Although Attitude is affected by both PU and perceived ease of use (PEOU).
many features of smart phones may easily attract younger generations Further, PU is also influenced by PEOU.
or people who are technology savvy, it may be an up-hill battle if it is Although many scholars have proven that TAM works well in
introduced to a totally different audience. Few businesses have a predicting user intention to accept a system, many studies applied
majority of their employees who are knowledgeable or even TAM in the areas where the use of a target information system is
comfortable with technology. For most others, technology readiness coerced by the administration of an organization. Such a coercion to
for their employees may stay on the low-end. Therefore, business use a system can sometimes have an effect on the relationship of PU
values of smart phones may diminish when employees are not ready and PEOU onto attitude and use intention.
for the business use of this technology. Venkatesh and Davis [51] expand the original TAM model and
This study examines the perceptions of Taiwanese cargo drivers inserted Social Influence Processes and Cognitive Instrumental Processes
on smart phone adoption factors. They were given smart phones as additional factors. Both new factors contain additional sub-factors for
with pre-installed network capabilities and embedded camera for example, subjective Norm, Voluntariness, Image are under social
barcode scanning, Outlook for scheduling, GPS for optimized influence processes. Job relevance, Output Quality, Result Demonstra-
delivering routes, and GPRS for headquarter to trace the parcel/ bility, and Perceived Ease of use are under cognitive instrumental
package delivery. processes.
As these cargo drivers have the first contact with customers when Gefen and Keil [19] consider IS development as an example of
delivering the goods, many delivery service companies are now Social Exchange Theory (SET) and therefore extended into TAM with
turning it into profitable sales venues. Instead of the traditional role of an inclusion of SET. Developer responsiveness was found to influence
goods delivery, truck drivers are now called “sales drivers” to initiate PU and PEOU. Kwasi and Salam [33] found that trust in new technology
sales with their customers. The embedded software in smart phones and mandated training and communication greatly influenced the PU
can be a great business tool for these sales drivers. In most cases, these and PEOU during the implementation stage of systems.
cargo sales drivers were neither young nor technology savvy, but are Grefen and Straub [21] extended TAM with gender and other
willing to react to new ways to facilitate their tasks. To this end, the human efficiency factors and confirmed their effect on behavioral
main purpose of this research is to examine the acceptance level of intention. Mathieson, Peacock, and Chin [37] supported that the
the smart phone technology and the factors that affect such an feeling of abundant resources can encourage or restrict certain specific
acceptance. Since smart phones are equipped with features from its actions. Moon and Kim [38] discovered individual's experience was
two ascendants (PDAs and cell phones), issues may well arise when influential to a person's intention to use WWW. Chau [8] separated
applying existing technology acceptance theories to predict the smart perceive usefulness into short term usefulness and long term
phone acceptance. In addition, business functions (delivery industry usefulness and both attributes had a similar effect on attitude and
in our case) may impose different requirements on the usability of intention to use a system.
smart phones. This could easily turn small annoyances into large Trust has appeared to play a significant role in TAM for repeated
hindrances for adoption and/or acceptance for implementation. For purchases [20] and for these companies to keep first-time on-line
example, the small screen of many smart phones is somewhat consumers [32]. Koufaris and Sosa also found that perceived company
inconvenient but can be reasonably acceptable for normal use. Using reputation, intended customer-friendly products, and service could
smart phones for PIM (personal information management) or services engender initial trust. Shih [41] further developed a rational behavior
other than making calls may be difficult if the user is in motion, such theory and expanded TAM to forecast consumers on internet
as driving. There has not been enough guidance from the existing purchasing. The aforementioned studies confirmed the role of PEOU
literature as to how acceptance of such a complicated technology is and PU on behavioral intention. Other factors, such as satisfaction
modeled. Nor is there enough evidence from the literature that could with WWW, perceptions on information content, and system services
offer an empirical foundation to predict smart phone acceptance. all had an effect on system acceptance.

Fig. 1. Technology Acceptance Model (TAM) by F. D. Davis, R. P. Bagozzi, and P. R. Warshaw [3].
424 K. Chen et al. / Computer Standards & Interfaces 33 (2011) 422–431

3.2. Applications of TAM to non-IT systems Hypothesis 2. Perceived ease of use of smart phones is directly
related to perceived usefulness.
TAM has also been used outside of the area of information
technology with some success. Examples include acceptance of medical Hypothesis 3. Perceived usefulness of smart phones is directly
technology [9,25], dairy growth technology [17], games [27], trust [41], related to behavioral intention.
search engines [35], online learning [42], and virtual socialization [45].
The study of Simon and Paper [43] included subjective norm (SN) in 4.2. Model 2 — TAM with self-efficacy
TAM to study the acceptance of Naval Voice Interactive Device, a voice
recognition system for voice activated data input. PEOU, PU and SN all 4.2.1. Self-efficacy — the definition
played a confirmed positive effect on behavioral intention, with PU and Bandura [2,4] defined self-efficacy as the individual's decision
SN being the strongest out of three aforementioned factors in terms of made to accomplish a specific goal through organization and
their effect towards attitude. PEOU had no effect on PU. In James et al.'s execution. Self-efficacy is composed of (a) prediction of result and
study [29] of biometric device adoption, more factors were added to (b) prediction of efficacy. Prediction of result refers to individual's
TAM (such as perceived need for privacy) and all main relationships prediction that his or her action could lead to a certain result.
among PU, PEOU and BI were confirmed. This includes the positive effect Prediction of efficacy refers to an individual's prediction on whether
of PEOU on PU. One major difference between the two aforementioned or not he or she is able to successfully execute a set of procedures to
studies lies in the respondents' experience in domain knowledge. This is achieve the desired result. Therefore, self-efficacy does not measure
what Hu, Chau, Sheng, and Tam [25] concerned that some studies may one's existing skill set, and instead, it measures one's perception on
use respondents from a “nonprofessional” context, thereby limiting the how he or she is able to complete a task at hand.
explanatory power of TAM. In their study, Hu et al. tested TAM model Compeau and Higgins [12] defined computer efficacy as the
using a group of physicians who had or would be involved in personal evaluation of one's capability of using computer. It emphases
telemedicine programs. PU was the one of strong factors that affected on the evaluation of what an individual can achieve in the future, but
attitude and intention. Simon and Paper's study [43] surveyed Navy not what an individual has done in the past. Torkzadeh and Koufteros
sailors, while James' [29] respondents were college faculty and students. [49] pointed out that computer self-efficacy should concentrate on the
In studying smart card adoption, Plouffe, Hulland and Vandenbosh [40] individual differences and the ability to discover computer related
surveyed merchants who were involved in a Smart Card trial. The results environment. Levine [34] indicated faith is a deciding factor on
showed a strong effect of PU on BI and PEOU on PU, but a weak PEOU behavior and confidence in computer was also greatly affecting the
on BI. amount of investment in learning process.
To follow up this line of TAM research, the role of PU on BI was
confirmed in most prior studies. Findings on the effect of PEOU on BI 4.2.2. Dimensions of self-efficacy
or attitude were, however, mixed, no matter respondents had Self-efficacy has long been treated by many researchers as a
experience in the domain knowledge (“professional”) or not. The unitary construct with no individual dimensions. However, this may
same finding was proved to be true for the relationship between PEOU vary depending on how the construct is operationalized and where it
and PU. However, one observation seems to suggest that if the is applied to. Even long-time self-efficacy researchers (e.g., Marakas et
respondents were “pragmatic” or under time pressure to get things al. [36] and Bandura [3]) advised us to develop new or revise/
done, PU's role may overshadow the PEOU on BI. More research is revalidate existing measures for each study. Nonetheless, one of the
definitely needed in this area to clarify this contradicting result. most used self-efficacy measures is a ten-item instrument developed
by Compeau and Higgins [12]. Although the aforementioned ten self-
4. Research hypotheses and research models efficacy items again formed a unitary construct in Compeau and
Higgins, the authors acknowledged that self-efficacy varies on three
4.1. Model 1 — TAM in smart phone acceptance distinct but interrelated dimensions/aspects and they are magnitude,
strength and generalizability. The magnitude dimension of self-
In this paper, the first model is developed to closely follow the efficacy “refers to the level of task difficulty one believes is attainable.”
original form of TAM to examine smart phone acceptance. Based on (p. 192). The strength dimension actually refers to how strong one's
existing studies on TAM [8,13,24,25,26,50,51,53], construct relation- conviction about his judgment. Generalizability is referred to as the
ships are hypothesized as follows. The model is graphically depicted extent to which the self-efficacy perception is constrained to
in Fig. 2 with all proposed hypotheses labeled. Note that this is a particular situations. In their later study, Thompson, Compeau and
simplified TAM model since prior studies have either suggested to Higgins [47] discovered two dimensions out of their original 10-item
remove the attitude construct from the model or directly adopted the construct. They chose to retain the first factor (self-efficacy variables
simplified form of the TAM model (e.g., [1,10,52]). SE1, SE2, SE6, and SE7) from the principal component analysis
because the variation in the first factor was greater.
Hypothesis 1. Perceived ease of use of smart phones is directly By visually inspecting the ten self-efficacy items proposed by
related to behavioral intention. Compeau and Higgins, there seems to exist some potential grouping

Fig. 2. Model 1 — TAM for smart phone acceptance.


K. Chen et al. / Computer Standards & Interfaces 33 (2011) 422–431 425

(or dimensions) of the construct. For example, the first four self- Hypothesis 5. Smart phone self-efficacy is directly related to
efficacy variables (SE1, SE2, SE3, and SE4) refer to one's self-efficacy perceived ease of use.
belief in general, while the remaining items refer to more or less one's
self-efficacy perception when some form of external resources (e.g., Hypothesis 6. Smart phone self-efficacy is directly related to
availability of a helper, availability of built-in support, and availability perceived usefulness.
of past experience with a similar task) is available. Furthermore, these
external resources may be available in the two following types: 5. Research method
human and non-human. From the discussion above, it comes with no
surprise that studies exploring these self-efficacy dimensions are A survey instrument was developed based on prior TAM and self-
emerging. Thatcher et al. [46] found two distinct self-efficacy efficacy literature. The questionnaire was reviewed by a panel of
dimensions and they named these dimensions human assisted content experts to ensure clarity and some good face validity. The
computer self-efficacy (self-efficacy variables SE5, SE6 and SE9) and survey was then, conducted at seven major operating locations of a
individual computer self-efficacy (self-efficacy SE1, SE2 and SE3). In large delivery service provider in Taiwan called Hsin Chu Trans (HCT).
their pretest, SE4, SE7, SE8 and SE10 were dropped out from the It is one of the first and is still one of the largest delivery service
analysis because they were neither human assisted nor individual companies in Taiwan. Target respondents were employees who
computer self-efficacy. These dropped self-efficacy items in fact, were directly involved with the task of the delivery of goods. The
characterize some other form of resource availability and may offer a respondents were first trained to use smart phones with respect to the
rationale for including another self-efficacy dimension or blending core features including calendar, barcode scanner, GPS, maps, delivery
with other existing ones. Availability of human assistance (as route analysis, SMS, and phone calls. To ensure uniformity of the
discussed in Thatcher et al. [46]) may also be considered as a form operating condition (hardware and software), all respondents were
of external resource which is more critical to one's self-efficacy trained to use the same model of smart phones.
perception. A total of 376 surveys were distributed and 334 were successfully
returned, which represents a response rate of 88.83%. The collected
4.2.3. The role of self-efficacy in TAM data were reexamined to exclude the records with incomplete
Through perceived ease of use, perceived usefulness, and per- answers and possible response set bias (i.e., respondents who
ceived test of accuracy, computer self-efficacy on behavior complian- answered all questions with the same response). The resulting data
cy has the effect onto decision making [54]. Together with TAM set consists of 215 usable questionnaires, making the final valid
factors, self-efficacy is also used to explain the willingness to use response rate to be 59.59%. A description of the sample demographic
online learning [39], online purchase intention [50], and decision attributes is provided in Table 1.
making systems [58]. As these prior studies indicated, self-efficacy is Out of the 215 respondents, 210 are males, and only 5 are females.
related to some constructs in TAM and should also be included to 30.80% of the subjects are younger than 31 years of age, 45.65% are
predict attitude and use intention. Therefore, our second model aims between ages 31 and 40, 18.67% are between ages 41 and 50, and the
at offering a composite version of TAM by including self-efficacy into remaining 7.91% are older than 51. In terms of their educational level,
the model (Fig. 3). Hypothesized relationships were, consequently, 5.11% of the drivers have no formal education other than elementary
developed based on previous studies including the studies of Hong et school, 28.00% have graduated from junior high, 57.70% have finished
al. [24]; Thong et al. [48]; Wang [54]; Hu et al. 26]; Yi and Hwang [58]; high school, and only 12.21% have graduated from college or received
Ong et al. [39]; Vijayasarathy [50] and Chen et al. [11]. other advanced degrees. About 20.51% of the respondents have
worked less than 5 years, 20.11% have worked 5 to 9 years, 45.65%
Hypothesis 1. Perceived ease of use of smart phones is directly have worked 11 to 20 years, and 17.74% have over 21 years of work
related to behavioral intention. experience.

Hypothesis 2. Perceived ease of use of smart phones is directly 6. Scale development


related to perceived usefulness.
Items in the questionnaire were derived from existing TAM and
Hypothesis 3. Perceived usefulness of smart phones is directly self-efficacy literature. A 5-point Likert-type scale was employed.
related to behavioral intention. The scale ranges from 1 (strongly disagree) to 5 (strongly agree).
Constructs were first analyzed with an exploratory factory analysis to
Hypothesis 4. Smart phone self-efficacy is directly related to verify their dimensionality. Sampling adequacy was then, measured
behavioral intention. by running the Kaiser–Meyer–Olkin (KMO) test, which resulted in a

Fig. 3. Model 2 — TAM with self-efficacy.


426 K. Chen et al. / Computer Standards & Interfaces 33 (2011) 422–431

Table 1 are all unidimentional and items are loaded into their respective
Samples demographics (n = 215). constructs.
n Min Max Mean S.D.

Gender Male: 210


7. Results
Female: 5
Age 18 57 34.81 845 7.1. Omnibus run for the entire sample
Educational level 1 4 2.74 .72
Elementary school: 11 (5.10%)
7.1.1. Model 1: TAM
Junior high school: 58 (27.00%)
High school: 122 (56.70%) In the first model, the relationship was tested out among TAM
College or higher: 24 (11.20%) constructs. As with several existing TAM-based studies, the result
Work years with the company 0 24 4.15 4.43 supports similar path relationships among TAM constructs (see
Total work years 0 35 11.30 7.27 Fig. 4). The chi-square statistic for this model is χ2 (196) = 428.29,
p b .001. Although the chi-square is significant, several prior studies
(such as Kline [31]) indicate that this statistic is extremely sensitive to
value of .92. Since this value exceeds the recommended value of .90, sample size. The value of Chi-square divided by degrees of freedom is
the data are considered to be appropriate for factor analysis. Bartlett's approximately 2.20, which is below the cut off value of five suggested
test of sphericity is significant (p b .01), thereby providing some by Wheaton et al. [55] as indicative of model fit. While the GFI
evidence that pursuing a factor model is feasible. Principal component (GFI = .81) was marginally acceptable (see for example Hair et al.
analysis was then performed with Varimax rotation. All constructs [23]), this statistic is biased downward when degrees of freedom are
entered the analysis and only those factors with an eigenvalue greater large relative to sample size [44]. As a result, Steiger proposed the
than one were selected. The result indicated five factors as predicted concept of adjusted GFI (also referred to as GFI-hat). The GFI-hat
(See Table 2). Total variance explained is 74.84%. To improve the calculated for this model is 0.97, which falls in the range of good
instrument, items were eliminated if their corrected item-total model fit. Both CFI (.93) and TLI (.91) fit indices are above .90,
correlation were below .5 [16]. As a result, the only item deleted in providing some additional evidence of a fitting model. RMSEA is .07.
this study was the last item of self-efficacy, which measures “I will use RMSEA values that range from 0.05 to 0.08 are considered as
a smart phone for the current task if I have used something similar to acceptable [5,23]). As a whole, the above fit indices suggest a good
perform the same task before.” The two self-efficacy dimensions fit of the model to the underlying data. Fig. 4 shows that both PU and
(factors 4 and 5) are labeled (a) Individual Self-Efficacy (ISE) as it PEOU have a positive effect on BI. However, PEOU's effect on BI
measures an individual's self-efficacy perceptions without referencing outweighs the effect coming from PU. PEOU is also a predictor of PU.
the availability of external resources, and (b) Assisted Self-Efficacy
(ASE) as it measures one's self-efficacy perception contingent upon 7.1.2. Model 2: TAM with self-efficacy
the availability of some external assistance or resources. Further from The second model is developed to model the traditional TAM
Table 2, reliability of constructs ranges from .86 to .94 and all have constructs with the two dimensions of the self-efficacy construct. The
exceeded the minimum acceptable threshold of .70. The self-efficacy chi-square for this model is χ2 (262) = 551.81, p b .001. The chi-
construct is split into two dimensions as predicted. Findings in Table 2 square statistic is significant (p b .01). The chi-square value divided by
are consistent with the existing literature in that TAM constructs the degree of freedom is below five (551.81/262 = 2.12). GFI is .83,
which again indicates the model is marginally acceptable. Steiger's
GFI-hat is calculated as .96, which falls in the range of good model
Table 2 fit. Both CFI (.94) and TLI (.93) fit indices are above .90, providing
Factor loadings of the individual items.
additional evidence of a fitting model. RMSEA is 0.07, again, lower
Component than 0.8 for an acceptable fit. As a whole, the above fit indices suggest
1 2 3 4 5 an acceptable fit of the model to the underlying data. As Fig. 5 shows,
path coefficients are similar to those of Fig. 4. PEOU still has a stronger
Perceived ease of use 3 .81
Perceived ease of use 5 .78 effect on BI than PU. ASE is a predictor of both PU and PEOU, but ISE is
Perceived ease of use 2 .78 only related to PEOU.
Perceived ease of use 4 .74
Perceived ease of use 1 .74 7.2. Self-efficacy clusters
Perceived ease of use 6 .74
Perceived usefulness 2 .77
Perceived usefulness 1 .77 To further examine the effect of the two types of efficacy
Perceived usefulness 4 .74 dimensions (ISE versus ASE), factor scores were saved for further
Perceived usefulness 3 .72 analyses. A hierarchical cluster analysis (using the Ward's method to
Perceived usefulness 5 .72
minimize the within-cluster differences) was then conducted on the
Perceived usefulness 6 .70
Behavioral intention 1 .84 saved factor scores to explore the groupings of respondents. By
Behavioral intention 3 .80 examining the agglomeration coefficients across different stages of
Behavioral intention 2 .79 the hierarchical process, a single highest increase was found between
Behavioral intention 4 .77 stages 213 (agglomeration coefficient = 292.65) and 214 (agglomer-
Self-efficacy 7 .77
ation coefficient = 428.00). Therefore, this result of the hierarchical
Self-efficacy 9 .74
Self-efficacy 6 .71 cluster analysis offers evidence to favor the two-cluster solution.
Self-efficacy 5 .66 Cluster one consists of one hundred and fifty-eight cases, while cluster
Self-efficacy 8 .65 two consists of fifty-seven cases. Cluster membership was again,
Self-efficacy 2 .82
saved for further analyses. A profile analysis using ANOVA was
Self-efficacy 4 .76
Self-efficacy 1 .74 conducted on the two-cluster solution to examine whether or not the
Self-efficacy 3 .70 two clusters are truly distinctive between each other in terms of self-
Cronbach's alpha .94 .93 .94 .87 .86 efficacy. As per Table 3, the two clusters differ from each other on
Variance explained 49.12% 9.27% 6.42% 5.84% 4.19% nearly all self-efficacy variables. The only item that falls in the border
Total variance explained 74.84%
line (p = .05) is the fifth self-efficacy variable. As predicted, the first
K. Chen et al. / Computer Standards & Interfaces 33 (2011) 422–431 427

Fig. 4. Omnibus run — TAM.

cluster includes respondents who scored higher on the first four self- significant at .05 level as predicted. Both PEOU and PU are strong
efficacy variables (i.e., SE1, SE2, SE3 and SE4). Characterized by these predictors for BI, while PEOU strongly influences the perception on
four self-efficacy variables, the respondents in this group perceived PU. One area that differs slightly from several existing TAM studies is
themselves more capable of working with smart phones without that the path coefficients between PEOU and BI appear to be heavier
needing much of external help. The second cluster, however, includes than the path between PU and BI. Since a good portion of PU is also
respondents who scored lower on the first four self-efficacy variables, predicted by PEOU (B = .71), PEOU seems to play the most significant
but higher on the remaining five self-efficacy variables. As these five role in this model for this group of respondents. It significantly affects
self-efficacy variables measure one's self-efficacy perception assuming the other two variables in the model.
the availability of some external help or assistance, these respondents For the respondents of the ASE type, the path that goes from PU to
in the second cluster appear to perceive higher on self-efficacy when BI was not significant (S.E. = .19, C.R. = .59, p N .05). PEOU appears to
some form of external help is available. be the only strong predictor for BI (Fig. 7).
Further profiling of demographics (see Table 4) shows that the two To examine the true difference among the above sub-models, all
respondent clusters differ primarily on their education attainment. factor loadings and structural paths were constrained to be equal in the
Respondents in the second cluster possess higher educational training multi-group structural model. The chi-square for this constrained model
or experience than that of respondents in the first cluster. Although is χ2 (215) = 466.05, p b .001. (CFI = .92, TLI = .91, RMSEA = .07). The
there is a slight age difference between the two clusters, the difference constrained model shows an increase of chi-square of 38.77 and an
is not that significant (p = .05) to be of great interest. The two groups increase of degree of freedom of 20. Also, it shows a slightly lower CFI
do not seem to be quite different on other demographic variables, and GFI and higher RMSEA. These changes are significant (pb .001).
such as gender, seniority at the current company, seniority in terms of Therefore, the relationship among TAM constructs apparently differs
total work experience, and job location. between the two self-efficacy groups.

7.3. Multi-group structural models for two types of self-efficacy 7.4. Discussion and conclusion

The multi-group structural model utilizes the traditional TAM This study investigates TAM in its original form and in a revised
constructs to investigate the difference between the two self-efficacy version with self-efficacy added as an additional construct. The
groups. For respondents of the ISE type, all paths in Fig. 6 are rationale for the added self-efficacy is based on the concept that

Fig. 5. Omnibus run: TAM with self-efficacy.


428 K. Chen et al. / Computer Standards & Interfaces 33 (2011) 422–431

Table 3
Profile analysis — self-efficacy.

n Mean S.D. Sum of squares df Mean square F Sig.

Self-efficacy 1 1 158 3.75 .91 Between groups 6.99 1 6.99 7.53 .01
2 57 3.35 1.00 Within groups 107.84 213 .93
Total 215 3.65 .98 Total 204.84 214
Self-efficacy 2 1 158 3.80 .89 Between groups 15.03 1 15.03 18.31 .00
2 57 3.19 1.01 Within groups 181.80 213 .85
Total 215 3.64 .96 Total 197.42 214
Self-efficacy 3 1 158 3.76 .93 Between groups 11.05 1 11.06 12.99 .00
2 57 3.25 .89 Within groups 181.42 213 .85
Total 215 3.62 .95 Total 192.48 214
Self-efficacy 4 1 158 3.83 .98 Between groups 31.28 1 31.28 37.79 .00
2 57 2.95 1.00 Within groups 176.32 213 .83
Total 215 3.60 .99 Total 207.60 214
Self-efficacy 5 1 158 3.89 .74 Between groups .26 1 .26 .46 .50
2 57 3.95 .73 Within groups 119.68 213 .56
Total 215 3.91 .75 Total 120.14 214
Self-efficacy 6 1 158 3.87 .80 Between groups 5.45 1 5.40 10.00 .00
2 57 4.23 .54 Within groups 116.24 213 .55
Total 215 3.05 .75 Total 121.70 214
Self-efficacy 7 1 158 3.01 .87 Between groups 15.63 1 15.63 24.45 .00
2 57 4.42 .57 Within groups 136.20 213 .04
Total 215 3.97 .84 Total 151.83 214
Self-efficacy 8 1 158 3.05 .78 Between groups 22.87 1 22.87 40.95 .00
2 57 4.40 .65 Within groups 118.94 213 .56
Total 215 3.85 .81 Total 141.81 214
Self-efficacy 9 1 158 3.72 .77 Between groups 20.87 1 20.87 40.39 .00
2 57 4.42 .57 Within groups 110.09 213 .52
Total 215 3.90 .78 Total 130.95 214

perceptions only affect final behavior to a limited extent unless one technology savvy (e.g., [57]). However, research about respondents
also considers himself or herself to possess the capability of who represent the other end of educational level has been lacking.
performing the behavior. This is the reason that triggered the In the realm of self-efficacy, this study found a closer map between
evolution of TRA into TPB with the added perceived behavioral the conceptual meanings of Compeau and Higgins' ten-item self-
control construct. Although there are still some differences between efficacy measure and the dimensionality of the construct. The two
self-efficacy and perceived behavioral control in terms of how they dimensions are named individual self-efficacy (ISE) and assisted self-
are operationalized, the importance of perception of one's capability efficacy (ASE) in the current study, respectively. The former
to perform an action is considered to be important in both theories. dimension captures one's general belief on their self-efficacy without
Although TAM is studied extensively elsewhere, this study differs referencing much of external sources, while the latter dimension
from these other studies in several ways. First, most TAM studies depends heavily on the existence or availability of external resources
concern most about the acceptance of some kind of software systems (such as human helper, past experience, and availability of time). This
(e.g., [15,56]) but our study focuses on acceptance of software and finding adds empirical evidence to the dimensionality of self-efficacy
hardware as a whole. Second, smart phone use in the delivery service by allowing most Compeau and Higgins' ten-item measure to be
industry has not been extensively studied. Third, interests of TAM mapped to the two self-efficacy dimensions. In the past, not all ten
have been in respondents who are either highly educated or self-efficacy items were mapped to the two dimensions because of the

Table 4
Profile analysis — demographic data.

Cluster n Mean S.D. Sum of squares df Mean square F Sig.

Gender 1 158 1.01 .112 Between groups .07 1 .07 2.96 .09
2 57 1.05 .225 Within groups 4.82 213 .02
Total 215 1.02 .151 Total 4.88 214
Age 1 158 35.49 9.104 Between groups 270.24 1 270.24 3.84 .05
2 57 32.95 5.947 Within groups 14,992.32 213 70.39
Total 215 34.81 8.445 Total 15,262.56 214
Education 1 158 2.67 .727 Between groups 2.81 1 2.81 5.51 .02
2 57 2.93 .678 Within groups 108.81 213 .51
Total 215 2.74 .722 Total 111.41 214
Location of job 1 158 2.63 1.342 Between groups 3.27 1 3.27 1.64 .20
2 57 2.91 1.596 Within groups 425.27 213 2.00
Total 215 2.71 1.415 Total 428.54 214
Seniority at the current company 1 158 4.48 4.717 Between groups 62.93 1 62.93 3.24 .07
2 57 3.25 3.377 Within groups 4131.85 213 19.40
Total 215 4.15 4.427 Total 4194.78 214
Total seniority (current and previous companies) 1 158 11.39 7.741 Between groups 4.10 1 4.10 .08 .78
2 57 11.07 5.805 Within groups 11,294.51 213 53.03
Total 215 11.30 7.200 Total 11,298.02 214
K. Chen et al. / Computer Standards & Interfaces 33 (2011) 422–431 429

Fig. 6. Respondents who scored high on individual self-efficacy.

variability in variance in the two factors (e.g., Thompson, Compeau respondents (cargo delivery) and their educational level, this finding
and Higgins [47]) and difference in classification of types of self- may not be an unexpected surprise. Delivery service requires a high
efficacy (e.g., Thatcher et al. [46]). As a result, Thompson, Compeau level of accuracy and speed to get cargos or goods from the source to
and Higgins chose to drop the second dimension in their study, the destination. Therefore, the same expectations may well likely be
retaining only the first factor (which is characterized by self-efficacy placed upon the technology used to carry out the job or enhance the
variables SE1, SE2, SE6 and SE7). Thatcher et al. identified two forms business process. If the drivers cannot easily use the device to quickly
of self-efficacy (i.e., individual versus human assisted self-efficacy), and accurately get their job done, most likely they will abandon it
excluding four self-efficacy variables (i.e., SE4, SE7, SE8 and SE10).Our despite the usefulness aspect of the device.
finding not only supports the two-dimensional characteristic of the Another way to interpret this result is to think from the usability
original self-efficacy construct, but also suggests that there are two point of view. If we conceptually categorize the features of smart
logical groupings of self-efficacy variables (individual versus assisted). phones into three main areas, “must-have,” “more-is-better” and
These two dimensions differ from those of Thatcher et al. in that the “attractive” usability, it is likely the “must-have” and ‘more-is-better”
second dimension captures both human assisted self-efficacy and self- factors were more relevant to the respondents in our study. According
efficacy assisted through other forms of resources (such as computer to Jokela [30], the “must-have” factors are the quality the consumers
manual). expect the product to have. They are not conscious customer
The majority of TAM studies examine some form of information requirements since they take these features for granted. To utilize
systems, but TAM's explanatory power is sometimes also confirmed in smart phones for communication, we considered that the telephony
non-IT systems. Our study focuses on smart phone adoption, which function to be a must-have, while PIM features to be another one. The
examines both software (i.e., the operating system and applications “more-is-better” aspect extends the core functionality with more
installed in smart phone) and hardware (i.e., hardware of smart features, but these features may be known to mostly exist in other
phone) components. Despite the many variations of TAM versions, types of devices. Full-featured Internet browsers, high resolution
both of our models, nonetheless, conform to the original version of camera and high capacity memory are likely to be the “more-is-
TAM in that PEOU and PU both had a direct and positive effect on better” factors in smart phones. The “attractive” factors offer the
behavioral intention. PEOU's positive role on PU was also confirmed. WOW effect that meet the unspoken or unexpected needs of users.
Interestingly, PEOU in both models had a stronger effect on behavioral Vendor specific features, such as picture rotation sensors in Apple's
intention than PU did. In addition, the effect of PEOU on PU has been iPhone when it first came out, fall well in this category. This
very strong for both models, indicating a good amount of usefulness categorization of three types of usability is not permanent. Today's
perception is largely related to how respondents perceived the device “attractive” factors may be tomorrow's “more-is-better” feature. In
to be easy to use. Considering the nature of the tasks of our terms of smart phones for business use, most “relevant” and “useful”

Fig. 7. Respondents who scored high on assisted self-efficacy.


430 K. Chen et al. / Computer Standards & Interfaces 33 (2011) 422–431

features are more likely on the “must-have” and “more-is-better” additional training or effectively demonstrating. Once employees
sides. Therefore, our finding of PEOU's influential role on both PU and are comfortable with “must-have” features, then the training may
behavioral intention does offer a managerial implication in that the proceed to the “more-is-better” or other advanced features.
“must-have” and “more-is-better” features perhaps are the important
focus to affect employees' perceptions on certain technology devices. References
Self-efficacy concerns about one's perception on his or her own
[1] D.A. Adams, R.R. Nelson, P.A. Todd, Perceived usefulness, ease of use and usage of
ability to perform the intended task. Our second model adds self-efficacy information technology: a replication, MIS Quarterly 17 (1992) 227–247.
into the first model to predict the overall effect of the expanded model [2] A. Bandura, Self-efficacy mechanism in human agency, American Psychologist 28
on smart phone use. Existing relationships of TAM factors are once again (1982) 122–147.
[3] A. Bandura, Guide for Constructing Self-Efficacy Scales, 2001, pp. 7–38.
confirmed with both PEOU and PU as the antecedents for behavioral [4] A. Bandura, Social Foundations of Thought and Action, Prentice Hall, Englewood
intention. Only Assisted Self-efficacy is a strong antecedent for PU, Cliffs, NJ, 1986.
but both individual and assisted self-efficacy are predictors of PEOU. [5] M.W. Brown, R. Cudeck, Alternative ways of assessing model fit, in: K.A. Bollen, S.
Long (Eds.), Testing Structural Equation Models, Sage Publications, Newbury Park,
Existing self-efficacy and TAM research does not seem to agree on the
CA, 1993, pp. 136–162.
role of self-efficacy. For example, self-efficacy was shown to be the [6] S.C. Chan, M.T. Lu, Understanding Internet banking adoption and use behavior: a
antecedent of both PEOU and PU [22], the antecedent of only PEOU Hong Kong perspective, Journal of Global Information Management 13 (2004)
[6,28,52] or no (or negative) effect on PU/PEOU [8]. Chau [8] speculated 22–44.
[7] Y.F. Chang, C.S. Chen, H. Zhou, Smart phone for mobile commerce, Computer
that the difference between his findings and Igbaria and Iivari's [28] was Standards and Interfaces 31 (2009) 740–747.
that Chau's study targeted a specific program use (i.e., MS Word), while [8] P.Y.K. Chau, An empirical assessment of a modified technology acceptance model,
Igbaria and Iivari aimed at measuring general computer use. Journal of Management Information Systems 14 (1996) 185–204.
[9] P.Y.K. Chau, P.J.H. Hu, Investigating healthcare professionals' decisions to accept
This speculation is not justified without its merit, but our results telemedicine technology: an empirical test of competing theories, Information &
offer some empirical evidence that deserves further attention. First, Management 40 (2002) 294–311.
our findings add a more fine-grained view to the debate of self- [10] P.Y.K. Chau, Influence of computer attitude and self-efficacy on IT usage behavior,
Journal of End-User Computing 14 (2001) 27–34.
efficacy and its role in TAM. Rather than asking if self-efficacy affects [11] J.V. Chen, D.C. Yen, K. Chen, The acceptance and diffusion of the innovative smart
TAM constructs, perhaps it might be more appropriate to ask what phone use: a case study of a delivery service company in logistics, Information &
self-efficacy dimension has an effect on what TAM constructs. Results Management 47 (2009) 241–248.
[12] D.R. Compeau, C.A. Higgins, Computer self-efficacy: development of a measure
from our study confirm that self-efficacy plays a positive role on PEOU,
and initial test, MIS Quarterly 20 (1995) 189–211.
while it only partially (i.e., individual self-efficacy, but not assisted [13] F.D. Davis, Perceived usefulness, perceived ease of use and user acceptance of
self-efficacy) affects PU. Second, we did not ask about conceptual information technology, MIS Quarterly 14 (1989) 319–340.
[14] F.D. Davis, R.P. Bagozzi, P.R. Warshaw, User acceptance of two theoretical models,
impression on smart phones in general. The respondents were trained
Management Science 36 (1989) 982–1003.
in one particular smart phone model. They were also asked to bring [15] F.D. Davis, V. Venkatesh, Toward preprototype user acceptance testing of new
the smart phones for a trial use to see how it fits their daily routine information systems: implications for software project management, IEEE
job. Their impressions on the smart phones were then surveyed. Such Transactions on Engineering Management 52 (2004) 32–47.
[16] W.J. Doll, G. Torkzadeh, The measurement of end user computing satisfaction, MIS
design may be more accurate in offering relevancy between the Quarterly 13 (1988) 259–274.
subjects' jobs and the technology device. To this end, the use context [17] R. Flett, F. Alpass, S. Humphries, C. Massey, S. Morriss, N. Long, The technology
lean more towards the Chau's side of the story. In contrast to Chau' acceptance model and use of technology in New Zealand dairy farming,
Agricultural Systems 80 (2004) 199–211.
study, our findings still confirm the relationship between PEOU and [18] M. Gartenberg, A. Wood, Integrated handsets: Balancing device functionality with
PU. Secondly, apart from the majority of TAM studies that concern consumer desires, 2005, available through Jupiter Research.
about computer and software systems; our study is an example that [19] D. Gefen, M. Keil, The impact of developer responsiveness on perceptions of
usefulness and ease of use: an extension of the technology acceptance model,
applies TAM to model perceptions on both hardware and software. As ACM SIGMIS Database 30 (1998) 36–50.
stated in the review of literature on TAM's applications to non-IT [20] D. Gefen, E. Karahanna, D.W. Straub, Trust and TAM in online shopping: an
systems, the findings on the role of PEOU on PU are still mixed. Our integrated model, MIS Quarterly 28 (2003) 52–90.
[21] D. Gefen, D.W. Straub, Gender differences in the perception and use of e-mail: an
confirmation of the existence of such a relationship in our study adds
extension to the technology acceptance model, MIS Quarterly 22 (1997) 389–400.
more evidence and insight to the original TAM framework. [22] P. Guriting, N.O. Ndubisi, Borneo online banking: evaluating customer perceptions
This study also shows that educational attainment is one of the and behavioral intention, Management Research News 30 (2006) 6–16.
[23] J. Hair, R.E. Anderson, R.L. Tatham, W.C. Black, Multivariate Data Analysis, Upper
major demographic variables that the two respondent groups mainly
Saddle River, NJ, 1998.
differ. Respondents in the assisted self-efficacy group have completed [24] W. Hong, J.Y.L. Thong, W.M. Wong, K.Y. Tam, Determinants of user acceptance of
more formal education than respondents in the individual self- digital libraries: an empirical examination of individual differences and systems
efficacy group. Samples from many TAM studies consist of respon- characteristics, Journal of Management Information Systems 19 (2001) 97–124.
[25] P.J. Hu, P.Y.K. Chau, O.R. Liu Sheng, K.Y. Tam, Examining the technology
dents with a medium to high level of formal educational training. As a acceptance model using physician acceptance of telemedicine technology, Journal
result, their technology readiness and self-efficacy may likely be of Management Information Systems 17 (1999) 91–112.
higher compared to the one of the respondents in our study. The [26] P.J.H. Hu, T.H.K. Clark, W.W. Ma, Examining technology acceptance by school
teachers: a longitudinal study, Information & Management 42 (2003) 227–241.
highest degree of approximately 88.80% of our subjects is high school [27] C.L. Hsu, H.P. Lu, Why do people play on-line games? An extended TAM with social
or below. This somewhat limits them in taking advantage of new influences and flow experience, Information & Management 42 (2004) 853–868.
technology, unless such technology use is mandated by the company [28] M. Igbaria, J. Liravi, The effects of self-efficacy on computer usage, OMEGA
International Journal of Management Science 24 (1995) 587–605.
with a good training. This finding offers a very useful managerial [29] T. James, T. Pirim, K. Boswell, B. Reithel, R. Barkhi, Determining the intention to use
implication in that corporate training for technology utilization needs biometric devices: an application and extension of the technology acceptance
to have a balanced focus on affecting one's attitude and perception model, Journal of Organizational and End User Computing 19 (2006) 1–25.
[30] T. Jokela, When good things happen to bad products: where are the benefits of
toward one's ability to use the technology. Self-efficacy has an
usability in the consumer appliance market? Interactions 11 (2004) 28–35.
important role on BI and PEOU as per the findings of our study. Efforts [31] R.B. Kline, Principles and Practice of Structural Equation Modeling, Guilford, New
should be made to affect the employees' perception on new York, 1998.
[32] M. Koufaris, W.H. Sosa, The development of initial trust in an online company by
technology. One plausible way to make that happen is not just offer
new customers, Information & Management 42 (2004) 377–397.
training, but to offer training in stages on targeted features of smart [33] A.G. Kwasi, A.F. Salam, An extension of the technology acceptance model in an ERP
phones with also supplemental resources to assist those who prefer implementation environment, Information & Management 42 (2004) 731–745.
help in order to do well (assisted self-efficacy). For example, “must- [34] T. Levine, Commitment to learning: effects of computer experience, confidence
and attitudes, Journal of Research on Computing in Education 17 (1997) 83–105.
have” features tend to be well known in user community despite the [35] S.S. Liaw, H.M. Huang, An investigation of user attitudes toward search engines as
experience level, but their creative use for business activities may need an information retrieval tool, Computers in Human Behavior 20 (2003) 751–765.
K. Chen et al. / Computer Standards & Interfaces 33 (2011) 422–431 431

[36] G.M. Marakas, Y. Yi, R.D. Johnson, The multilevel and multifaceted character of [57] C.-S. Wu, F.-F. Cheng, D.C. Yen, Y.-W. Huang, User acceptance of wireless
computer self-efficacy: toward clarification of the construct and an integrative technology in organizations: a comparison of alternative models, Computer
framework for research, Information Systems Research 9 (1998) 126–163. Standards and Interfaces 33 (2011) 50–58.
[37] K. Mathieson, E. Peacock, W.W. Chin, Extending the technology acceptance [58] M.Y. Yi, Y. Hwang, Predicting the use of web-based information systems: self-
model: the influence of perceived user resources, ACM SIGMIS Database 33 efficacy, enjoyment, learning goal orientation and the technology acceptance
(2001) 86–112. model, International Journal of Human–Computer Studies 59 (2003) 431–449.
[38] J.W. Moon, Y.G. Kim, Extending the TAM for a world-wide-web context,
Information & Management 39 (2001) 217–230.
[39] C.S. Ong, J.Y. Lai, Y.S. Wang, Factors affecting engineers' acceptance of asynchronous
e-learning systems in high-tech companies, Information & Management 42 (2004) Kuanchin Chen is an Associate Professor of Computer
795–804. Information Systems at Department of Business Information
[40] C.R. Plouffe, J.S. Hulland, M. Vandenbosch, Research report: richness versus Systems, Western Michigan University. Dr. Chen's research
parsimony in modeling technology adoption decisions — understanding merchant interests include electronic business, privacy & security,
adoption of a smart card-based payment system, Information Systems Research 13 online behavioral issues (e.g., interactivity, dependency, and
(2001) 208–222. tracking/protection), Internet frauds, usability, data mining,
[41] H.P. Shih, An empirical study on predicting user acceptance of e-shopping on the and human computer interactions. He has published articles
Web, Information & Management 42 (2004) 351–368. in journals and other academic publication outlets, includ-
[42] H.M. Selim, An empirical investigation of student acceptance of course websites, ing Decision Support Systems, IEEE Transactions on Systems,
Computers & Education 41 (2003) 343–360. Man, and Cybernetics, Information & Management, Commu-
[43] S.J. Simon, D. Paper, User acceptance of voice recognition technology: an empirical nications of the Association for Information Systems (CAIS),
extension of the technology acceptance model, Journal of Organizational and End IEEE Transactions on Education, Journal of Database Manage-
User Computing 20 (2007) 25–51. ment, Journal of Computer Information Systems and many
[44] J.H. Steiger, Structural model equation and modification: an interval estimation others. He currently serves on the editorial or advisory boards of Information Resources
approach, Multivariate Behavioral Research 26 (1990) 173–180. Management journal, International Journal of Information Systems and Change Manage-
[45] H.H. Teo, H.C. Chan, K.K. Wei, Z. Zhang, Evaluating information accessibility and ment, Journal of Website Promotion, Communications of the ICISA, IGI Global (formerly Idea
community adaptivity features for sustaining virtual learning communities, Group), eWeek, and CMP.
International Journal of Human–Computer Studies 59 (2003) 671–697.
[46] J.B. Thatcher, M.J. Gundlach, D.H. McKnight, M. Srite, Individual and human-
Jengchung Victor Chen is an Associate Professor in the
assisted computer self-efficacy: an empirical examination, WI 2007: 8th
Institute of International Management at National Cheng
International Conference Wirtschaftsinformatik, Karlsruhe, Germany, 2007,
Kung University, Tainan, Taiwan. Since graduated from the
pp. 841–858.
CIS doctoral program at the University of Hawaii in 2002, he
[47] R. Thompson, D. Compeau, C. Higgins, Intentions to use information technologies:
has 30 articles published or accepted in refereed journals
an integrative model, Journal of Organizational and End User Computing 18
such as Information and Management, Decision Support
(2006) 25–46.
Systems, and CyberPsychology and Behavior. His research
[48] J.Y.L. Thong, W. Hong, K.Y. Tam, Understanding user acceptance of digital
interests are Information Ethics, Electronic Commerce,
libraries: what are the roles of interface characteristics, organizational context,
Service Quality, and Project Management.
and individual differences? International Journal of Human–Computer Studies 58
(2002) 215–242.
[49] G. Torkzadeh, X. Koufteros, Factorial validity of a computer self-efficacy scale and
the impact of computer training, Educational and Psychological Measurement 55
(1994) 813–821.
[50] L.R. Vijayasarathy, Predicting consumer intentions to use on-line shopping: the
case for an augmented technology acceptance model, Information & Management David C. Yen is currently a Raymond E. Glos Professor in
42 (2004) 747–762. Business and a Professor of MIS of the Department of
[51] V. Venkatesh, F.D. Davis, A theoretical extension of the technology acceptance Decision Sciences and Management Information Systems at
model: four longitudinal field studies, Management Science 47 (2000) 186–204. Miami University. Professor Yen is active in research and has
[52] V. Venkatesh, F.D. Davis, A model of the antecedents of perceived ease of use: published books and articles which have appeared in
development and test, Decision Science 28 (1996) 451–481. Communications of the ACM, Decision Support Systems,
[53] V. Venkatesh, M.G. Morris, Why don't men ever stop to ask for directions? Gender, Information & Management, Information Sciences, Computer
social influence, and their role in technology acceptance and usage behavior, MIS Standards and Interfaces, Government Information Quarterly,
Quarterly 25 (2000) 115–139. Information Society, Omega, International Journal of Organiza-
[54] Y.S. Wang, The adoption of electronic tax filing systems: an empirical study, tional Computing and Electronic Commerce, and Communica-
Government Information Quarterly 21 (2002) 333–352. tions of AIS among others. Professor Yen’s research interests
[55] B.B. Wheaton, B. Muthen, D.F. Alwin, G.F. Summers, Assessing reliability and include data communications, electronic/mobile commerce,
stability in panel models, in: D.R. Heise (Ed.), Sociological Methodology, Jossey- and systems analysis and design.
Bass, San Francisco, CA, 1977.
[56] I.L. Wu, K.W. Wu, A hybrid technology acceptance approach for exploring e-CRM
adoption in organizations, Behavior & Information Technology 25 (2005)
303–316.

Você também pode gostar