Computers in Human Behavior 90 (2019) 181–187

Contents lists available at ScienceDirect

Computers in Human Behavior

journal homepage: www.elsevier.com/locate/comphumbeh

Full length article

A hybrid modeling approach for predicting the educational use of mobile T

cloud computing services in higher education
Ibrahim Arpaci
Tokat Gaziosmanpasa University, Department of Computer Education and Instructional Technology, Tokat, Turkey

A R T I C LE I N FO A B S T R A C T

Keywords: The decision to integrate mobile cloud computing (MCC) in education without determining optimal use sce-
Information management narios is a universal problem as the adoption of such services becomes widespread. Accordingly, this study
Cloud computing developed and validated a predictive model that explains the role of students' information management (i.e.
Artificial intelligence retrieve, store, share, and apply) practices in predicting their attitudes toward using the MCC services for
Machine learning
educational purposes. This study validated the model by the complementary use of machine learning algorithms
alongside a classical SEM-based approach based on data collected from 308 undergraduate students. The SEM
results indicated that the students' information management (i.e. retrieve, store, share, and apply) practices were
significantly associated with their attitudes, which were significantly associated with the behavioral intentions.
The structural model explained a significant portion of the variance in the behavioral intentions. Likewise, the
classifier model suggested that the students’ information management practices and attitudes predicted the
behavioral intentions. Further, the applied algorithms predicted the behavioral intentions with an accuracy of
more than 72% in most cases. Thereby, the study extended an original theory (TRA) into the MCC area by using a
multi-analytical approach. The findings implied that employing the MCC services for personal information
management should be supported and encouraged in the higher education by designing authentic learning
environments and scaffolding the students in using such services.

1. Introduction “infrastructure as a service” (IaaS) for information storage and man-

Mobile cloud computing (MCC) as a new distributed computing
paradigm can be defined as an infrastructure, application or process,
where the data storage and processing migrated from smart mobile
technologies to the distributed cloud servers (Dinh, Lee, Niyato, &
Wang, 2013). The MCC has gained a substantial attention of both the
organizations and individuals as a promising solution for the ubiquitous
environments, in which data storage and processing occur over a
“cloud” via the Internet (Park & Kim, 2014). Investing in the MCC,
organizations improved their capacity and capabilities without the cost
of installing new infrastructure or software (Subashini & Kavitha,
2011). Further, the MCC reduced the initial installation and main-
tenance costs, and thereby, promoted the efficiency and green tech-
nology (Aepona, 2018).
There are various cloud deployment-models such as private, public,
hybrid, and community cloud (Shon, Cho, Han, & Choi, 2014). The
study focused on the public cloud that allocates resources on a per-user
basis through applications such as Dropbox, SkyDrive, and Google
Drive. However, the MCC service models for individuals include
“platform as a service” (PaaS) for automatic content synchronizing and
agement. On the other hand, these services provide enterprises with the
IaaS for an enterprise infrastructure building and “software as a ser-
vice” (SaaS) for an enterprise solution provider (Arpaci, 2016).
The MCC provides users at the organizational and individual level
with several advantages and given these advantages the MCC services
are considered to be the growth engine of the industry 4.0 (Park & Kim,
2014). For example, the MCC increases data storage capacity and al-
lows an efficient data synchronization and information management in
a ubiquitous environment (Park & Kim, 2014). Besides, energy effi-
ciency is a critical issue since battery power of the mobile technologies
is limited. Migration of the complex processing from mobile technolo-
gies to the cloud servers extends the battery life (Cuervo et al., 2010).
Furthermore, service providers have certain security mechanisms and
backup systems to protect user data. Therefore, saving documents and
files on the distributed cloud servers may enhance the reliability by
reducing virus-like threats and data loss (Arpaci, 2016).
Smart mobile technologies change the way of information is man-
aged (Ogiela, 2017). For example, using mobile Internet on smart-
phones provides students with ubiquitous access to information and
learning materials. Further, social networking and communication

E-mail address: ibrahim.arpaci@gop.edu.tr.

https://doi.org/10.1016/j.chb.2018.09.005
Received 20 January 2018; Received in revised form 16 August 2018; Accepted 10 September 2018
Available online 11 September 2018
0747-5632/ © 2018 Elsevier Ltd. All rights reserved.
I. Arpaci
applications allow them to share information with peers. More im-
portantly, the integrated technologies enable them to use information
stored in the MCC services in problem solving and decision-making.
Considering the extensive use of mobile devices such as smartphones,
tablets, and smartwatches by undergraduate students, it is important to
find creative ways for an effective integration of these technologies into
the higher education. Such an integration may enable students to learn
by using the technologies, which they are familiar and confident with,
and highly motivated to use. It is important to note that providing an
effective management of personal information through information
retrieval, access, storage, sharing, and application is one of the main
advantages of the MCC services. However, the decision to integrate the
MCC in educational settings without determining optimal use scenarios
is a universal problem as the adoption of such services becomes wide-
spread. Thus, this study investigated adoption of the MCC among un-
dergraduate students by focusing on the use of these services for per-
sonal information management.
2. Literature review
The MCC is a recent research area and there are numerous studies
on the adoption of the MCC at the both organizational and individual
level. For example, Ul Amin, Inayat, Shahzad, Saleem, and Aijun (2017)
applied the “Unified Theory of Acceptance and Use of Technology”
(UTAUT) to identify key predictors of the MCC adoption by healthcare
professionals. Their results suggested the UTAUT constructs (i.e. per-
formance-expectancy, social influence, effort-expectancy) were sig-
nificant in determining the behavioral intentions. In another study,
Lian, Yen, and Wang (2014) investigated the factors affecting the cloud
computing adoption in Taiwan's hospital industry based on the HOT-fit
(Human-Organization-Technology fit) model and TOE (Technology-
Organization-Environment) framework. The results suggested that
technical competence, data security, cost, complexity, and top man-
agement support were critical factors in the adoption. Chen, Chen, and
Lee (2018) investigated key factors affecting the organizations' adop-
tion of the cloud services. Their results suggested that top-management-
support has a vital role in the adoption at the organizational level.
Lee (2016) identified the key factors explaining the adoption of
cloud services at the individual level based on the “Diffusion of In-
novations” (DOI) theory. The results suggested the relative advantage,
observability, and self-efficacy were significant antecedents of the
adoption. Park and Kim (2014) focused on key factors predicting the
individuals' acceptance of the MCC services. Results indicated that
connectedness, perceived mobility, service quality, security, and sa-
tisfaction were significant predictors of the user acceptance. Wang and
Huang (2016) reported that social influence was a key predictor of the
students’ intention to use cloud services. Further, they reported that
scaffolding in problem-solving and training were useful in familiarizing
the students with such services.
Arpaci (2016) proposed a theoretical structural model based on the
“Technology Acceptance Model” (TAM) and suggested that subjective
norm, trust, and perceived-usefulness play significant roles in pre-
dicting the students' attitudes towards using the MCC services. In an-
other study, Sharma, Al-Badi, Govindaluri, and Al-Kharusi (2016) de-
veloped a hybrid model based on the TAM to predict factors affecting
the adoption of the cloud computing by IT professionals. They em-
ployed linear regression along with neural network analysis and found
that job opportunity was the most significant predictor of the adoption.
Shiau and Chau (2016) developed a theoretical structural model based
on the TAM and motivational model to predict behavioral intentions to
join a cloud computing class. The results suggested subjective norms
and attitudes were positively associated with the behavioral intention.
Kim and Kim (2018) proposed a research model in order to identify the
antecedents of the MCC adoption. The results suggested that trust,
convenience, and perceived uncertainty were significant factors in the
individuals’ adoption decision.
182
Computers in Human Behavior 90 (2019) 181–187
3. Theoretical background
Most of the studies reviewed applied or extended a theory such as
TAM (Davis, 1989) or UTAUT (Venkatesh, Morris, Davis, & Davis,
2003) by conducting an explanatory statistical analysis. However, such
an approach has two main limitations. First, each theory was developed
and tested for a specific domain, and therefore, may not work well in
other domains or areas. Thus, this study extended the original “Theory
of Reasoned Action” (TRA) with additional constructs to better explain
the research context. Second, previous studies pointed out statistical
limitations of the technology adoption studies and emphasized the
importance of employing a multi-analytical approach by combining
predictive analytics and causal explanatory statistical analysis to vali-
date a predictive model (Sharma et al., 2016; Sharma, Joshi, & Sharma,
2016; Tan, Ooi, Leong, & Lin, 2014). The study therefore tested the
research model by using both the classical SEM approach and artificial
intelligence techniques (i.e. machine learning). The study employed the
SEM approach to understand causal relationships and the (com-
plementary) study applied the machine learning algorithms to predict
the behavioral intentions based on the proposed constructs. Further,
this study approaches the MCC adoption from an information man-
agement perspective. The fact that providing an effective information
management was one of the main advantages of the MCC services
supports the idea that individuals' information management practices
may have a critical role in their adoption decision. Accordingly, this
study investigated the role of students’ information management
practices on their attitudes toward using the MCC services in educa-
tional settings.
The TRA postulated that beliefs affect attitudes, which positively
influence behavioral intentions, whereas the intentions ultimately in-
fluence the actual behavior (Fishbein & Ajzen, 1975). Benbasat and
Barki (2007) suggested using the original theory (TRA) to provide a
strong theoretical grounding for a novel model by incorporating dif-
ferent antecedents relevant to the nature of ICT integration and use in
diverse settings. Likewise, Nistor (2014) argued that “educational
technology acceptance” studies should consider characteristics of the
educational context. Considering the shortcomings of the technology
acceptance models and analysis methods suggested by (Benbasat &
Barki, 2007), this study identified the key antecedents of the attitudes
towards using the MCC services and employed an innovative approach
comprising both machine learning algorithms and the SEM, and
thereby, strengthened the study design. Fig. 1 illustrates the research
model that suggests the behavioral intentions (continued use inten-
tions) are predicted by the attitudes, which are predicted by the stu-
dents’ information management practices such as information retrieval,
storage, sharing, and application.
Information
Retrieval
Information
Storage
.77**
Behavioral
Attitude
Intention
Information R2 = .33, e = .17 R2 = .59, e = .17
Sharing
Information
Application
*p< .01, **p< .001, Chi-Square= 393.25, DF= 235, Chi-Square/DF=1.67
Fig. 1. The structural model.
I. Arpaci Computers in Human Behavior 90 (2019) 181–187

Rogers's (2003, p.19) DOI theory suggested that the stages by which the use of such services.
an individual adopts a new technology or innovation, whereby diffu-
H4b. Information sharing would predict the behavioral intention to use
sion is accomplished include “awareness of the need for an innovation,
the MCC services for educational purposes.
decision to adopt (or reject) the innovation, initial use of the innovation
to test it, and continued use of the innovation.” Accordingly, a training H5a. An increase in the volume of information that's applied via the use
program introduced the participants to personal information manage- of the MCC services is positively associated with the attitude towards
ment, which refers to the practice of activities an individual performs to the use of such services.
create or acquire, store, retrieve, share, and apply the information re-
H5b. Information application would predict the behavioral intention to
quired to complete the tasks. The participants employed the mobile web
use the MCC services for educational purposes.
services (i.e. search, maps), mobile cloud applications (i.e. Google
Drive, iCloud, Dropbox) and social media applications to perform the
training tasks. 4. Method

3.1. Hypotheses 4.1. Sample

3.1.1. Behavioral intentions and attitudes A sample of 308 university students were recruited from a public
“Behavioral intention” was defined as “the degree of an individual's university in Turkey. The study included 186 females (60.4%) and 122
belief that he or she will continue to use a system” (Arpaci, 2017, p. (39.6%) males with a mean age of 21.88 years (SD = 2.37). Majority of
384). However, attitudes towards using a technology was defined as “an the students (42.5%) were seniors; 12.3% were freshmen, 27.6% were
individual's overall affective reaction to use a particular system” sophomore, and 17.5% were juniors.
(Arpaci, 2017, p. 384). This study therefore hypothesized that the at-
titude toward using the MCC services would be positively associated 4.2. Procedure
with the behavioral intention to use such services for educational
purposes. Further, the attitude would be a predictor of the behavioral IRB of the affiliated university approved the research and the pro-
intention. Therefore; cedures complied with the ethical standards of the institutional board
guidelines. All participants were informed about the purpose of the
H1a. Attitude towards using the MCC services would be positively
research after an informed consent was obtained. The study was con-
associated with the behavioral intention.
ducted in regularly scheduled IT classes (4 h per week) during 14
H1b. Attitude towards using the MCC services would predict the weeks. The participants performed practical implementations of the
behavioral intention. theoretical knowledge during the interactive tasks such as: 1) Search for
research articles in electronic databases (i.e. Scopus, WOS, and Google
Scholar), 2) Save an annotated bibliography of the articles to the MCC
3.1.2. Information management practices
services (i.e. OneDrive, Dropbox, and Google Drive), 3) Form a com-
Information is processed, categorized, organized or structured data
munities of practice and use the MCC services along with social net-
(Davenport & Prusak, 2000; Nonaka, 1994). However, knowledge is a
works to share files, communicate with others, and build a common
combination of skills, information, experience, and insights (Nonaka &
cognition to complete the tasks. Finally, the survey instrument with
Takeuchi, 1995). Knowledge includes organizational routines, values,
explicit assurance that anonymity is guaranteed was administered
and practices; therefore, this study preferred to use the term “in-
during the last week of the regular classes.
formation management” instead of “knowledge management.” This
study defined information management as practices that help retrieve
4.3. Instruments
and store information, share it with others, and apply it to solve pro-
blems and make better and timely decisions (Alavi & Leidner, 2001;
The measurement instrument for the variables “behavioral inten-
Mitchell, 2003). Turban, Sharda, and Delen (2011) suggested the use of
tions” and “attitudes” was a Likert scale adapted from Ajzen's theory
information technologies to enhance information management prac-
(1991). Moreover, the measurement instrument for the information
tices of retrieval and access, transfer, storage, and application. In the
management practices of retrieval, storage, sharing, and application
same vein, the present study suggested the integration of smart mobile
was adapted from a previously validated and reliable instrument
technologies into the higher education to support personal information
(Arpaci, 2017). The final instrument has 24 items with a five-point
management. The more information managed by using the MCC ser-
Likert-type scale ranging from “strongly disagree” to “strongly agree.”
vices the more favorable the attitude towards using such services for
educational purposes. Further, the students’ information management
5. Results
practices would be a predictor of the behavioral intention to use the
MCC services for educational purposes. Accordingly,
5.1. Internal reliability and validity
H2a. An increase in the volume of information that's retrieved via the
use of the MCC services is positively associated with the attitude IBM SPSS (ver. 25) and IBM AMOS (ver. 23) were used to screen and
towards the use of such services. analyze the data. Construct validity along with reliability of the scales
were assessed by conducting an EFA (See Table 1). Results revealed that
H2b. Information retrieval would predict the behavioral intention to
each item had a communality value and a factor loading greater than
use the MCC services for educational purposes.
the critical threshold (0.40) suggested by Field (2005). Cronbach's
H3a. An increase in the volume of information that's stored via the use alpha reliability analysis results indicated a good homogeneity and
of the MCC services is positively associated with the attitude towards reliability among the items. The AVE values were greater than 0.50,
the use of such services. indicating adequate convergent validity for all constructs (Hair,
Tatham, Anderson, & Black, 2006).
H3b. Information storage would predict the behavioral intention to use
the MCC services for educational purposes.
5.2. The structural model
H4a. An increase in the volume of information that's shared via the use
of the MCC services is positively associated with the attitude towards Kaiser's (1970) “KMO measure of sampling adequacy” and Bartlett's

183
 I. Arpaci

Table 1
Instrument reliability and validity.
Construct Item α Item-total correlation Factor load Communality Total variance CR AVE
explained

Behavioral intention CUI1. “I intend to continue to use mobile cloud services for educational purposes in the future.” .91 .76 .85 .72 74.34 .91 .68
CUI2. “I plan to continue to use mobile cloud services for personal information management in the future.” .78 .86 .75
CUI3. “I predict that I would frequently use mobile cloud services in the future.” .77 .86 .74
CUI4. “I predict that I would frequently use mobile cloud services for educational purposes.” .79 .87 .76
CUI5. “I predict that I would frequently use mobile cloud services for personal information management.” .79 .87 .76
Attitude A1. “Using mobile cloud services for educational purposes is a good idea.” .89 .73 .82 .67 65.30 .89 .59
A2. “Using mobile cloud services to create a retrievable archive of personal information is pleasant.” .78 .86 .74
A3. “Using social media applications on my smartphone to exchange information is fun.” .67 .78 .61
A4. “Using mobile cloud services for personal information management is a wise idea.” .75 .83 .69
A5. “Using mobile cloud services for educational purposes makes learning more interesting.” .68 .78 .61
A6. “I like the idea of using mobile cloud services for personal information management.” .67 .77 .60
Information retrieval .91 .75 .86 .74 78.43 .91 .51

184
IR1. “Using mobile Internet on my smartphone enables to retrieve learning materials and information anytime and
anywhere.”
IR2. “I access course materials and information using mobile cloud services on my smartphone.” .85 .92 .84
IR3. “Using mobile cloud services on my smartphone enables quick access to learning materials and information .84 .92 .84
resources.”
IR4. “Using mobile cloud services on my smartphone enables ubiquitous access to e-databases and e-journals.” .74 .85 .72
Information storage IS1. “Using mobile cloud services on my smartphone enables the retrievable storage of electronic information and .91 .81 .91 .83 84.06 .91 .52
documents.”
IS2. “I store course materials and documents using mobile cloud services on my smartphone.” .83 .93 .86
IS3. “Using mobile cloud services enables me to store learning materials and information with ubiquitous access.” .80 .91 .83
Information sharing ISh1. “Using social media applications on my smartphone enables me to exchange information and documents with .71 .51 .80 .64 64.61 .71 .51
classmates.”
ISh2. “I share course materials and information with classmates using mobile cloud services on my smartphone.” .62 .86 .74
ISh3. “I share learning materials and documents with classmates using mobile cloud services on my smartphone.” .48 .75 .57
Information application IA1: “I apply knowledge and experience gained by using mobile cloud services to complete learning tasks.” .83 .64 .83 .69 74.72 .83 .53
IA2. “I use knowledge obtained by using mobile cloud services in decision making processes.” .73 .89 .79
IA3. “I employ knowledge and intelligence gained from using mobile cloud services in problem-solving activities.” .71 .88 .67
Computers in Human Behavior 90 (2019) 181–187
I. Arpaci Computers in Human Behavior 90 (2019) 181–187

Table 2 attributes of the students’ information management practices and their

KMO and Bartlett's test results.
KMO Chi-Square
Behavioral intention .88 1022.27
Attitude .88 1001.80
Information retrieval .83 839.40
Information storage .75 594.58
Information sharing .65 198.79
Information application .71 357.46
(1951) “test of sphericity” suggested that conducting a CFA on the data
was appropriate (See Table 2). KMO values were all above 0.65 and
Bartlett's test values were all significant (p < .001). This suggested that
measures for each construct were interdependent (Leech, Barrett, &
Morgan, 2005).
The CFA was conducted using maximum-likelihood estimates in an
attempt to test the research model. Results suggested the structural
model demonstrates a “good fit” to data: [χ2/df = 1.67, GFI = 0.91,
NFI = 0.92, AGFI = 0.88, CFI = 0.96, TLI = 0.96, IFI = 0.96,
RMSEA = 0.05]. A GFI, NFI, TLI, CFI, and IFI equal or above 0.90 and a
RMSEA less than 0.08 indicate an acceptable model fit (Hair et al.,
2006). Further, Kline (2005) suggested a χ2/df less than 2.0 indicates a
good fit.
5.3. Hypothesis testing
A SEM approach was used to explore causal relationships among the
variables. The standardized-path-coefficients in the structural model
were all statistically significant (See Fig. 1). Results indicated that in-
formation retrieval, information storage, information sharing, and in-
formation application were significantly associated with the attitudes
toward using the MCC services. All these factors explained .33 of the
variance in the attitudes. Whereas, the information retrieval had the
largest impact on the attitudes with a standardized coefficient of 0.45.
On the other hand, the information storage had a relatively small im-
pact on the attitudes with a standardized coefficient of 0.16.
The results shown in Table 3 suggested the attitudes were sig-
nificantly associated with the behavioral intentions with a standardized
coefficient of 0.77. These factors accounted for a significant variance in
explaining the behavioral intentions (R2 = 0.59).
5.4. Testing the predictive model using the machine learning algorithms
Classification is one of the machine learning techniques that aims to
build a classifier model in order to predict a behavior through classi-
fying the data into a number of pre-defined classes based on a certain
criterion (Ngai, Xiu, & Chau, 2009). Common techniques employed for
the classification are Bayesian networks, decision trees, if-then-else
rules, and neural networks.
Machine learning, an artificial/computational intelligence tech-
nique, uses a set of attributes and searches for correlations between the
attributes and performance of the learning algorithms (Kotsiantis,
Zaharakis, & Pintelas, 2007). This study employed machine learning
classification algorithms to predict the behavioral intentions based on
attitudes. Behavioral intentions and attitudes have nominal values and
Sig. classified as low, medium or high by using the Mean ± SD combina-
tion. On the other hand, the attributes of information retrieval, storage,
.001 sharing, and application have numeric values.
.001
The predictive model was tested by using Weka (ver. 3.8.1) based
.001
.001
on the most common classification algorithms, which classify data into
.001 a number of pre-defined classes with different learning schemas. The
.001 results shown in Table 4 provided performance of the best performance
classifier based on two test modes; 10-fold cross-validation and per-
centage split (66.0%).
The study employed a linear logistic regression classifier (Logistic),
a meta AdaBoostM1, a Bayesian classifier (Naive bayes), a lazy LWL,
and two rule learners (OneR, JRip). The results showed that the OneR
rule learner performs slightly better than the other classifiers. OneR
classifier predicted the behavioral intentions with an accuracy of
73.70% and 71.43% for 10-fold cross-validation and percentage split,
respectively. Further, this algorithm had a better performance in terms
of true positive (TP) rate (0.74) and F-measure (0.73) for the 10-fold
test mode compared to the other classifiers (i.e. Naive bayes, Logistic,
JRip, AdaBoostM1, LWL). The results shown in Table 5 suggested that
attributes of the attitudes, information retrieval, information storage,
information sharing, and information application predicted the beha-
vioral intentions with an accuracy of 74%, and thereby, H1b, H2b, H3b,
H4b, and H5b were supported.
6. Discussion and conclusion
6.1. Research implications
Using the MCC services provides students with ubiquitous access to
instructional materials and information anytime and anywhere. In ad-
dition, these services provide students with a storage capacity to store
their reference materials. Furthermore, the MCC services enable the
students to share ideas, documents, and files with peers and classmates.
Eventually, the students can apply and transform that knowledge and
experience in problem-solving and decision-making. However, the de-
cision to integrate these services in educational settings requires de-
termining an optimal use scenario. This study therefore focused on the
use of the MCC services for personal information management.
Accordingly, the study employed a hybrid modeling approach for in-
vestigating the role of students’ information management (i.e. retrieve,
store, share, and apply) practices in their attitudes (toward use) and the
intentions to (continue to) use the MCC services.
The actuality of the ideas and employing an integrated multi-ana-
lytical approach were the novel contributions of the study. The study
developed a research model by extending the TRA with additional
constructs and tested the model by integrating the SEM and machine
learning approach. This study identified the causal relationships be-
tween the dependent and independent variables by employing the SEM
approach. The SEM results indicated that the students’ information
management practices (i.e. retrieve, store, share, and apply) and atti-
tudes were significantly associated with the behavioral intentions to use
the MCC services for educational purposes.
In a complementary study, machine learning algorithms were

Table 3
The SEM results.
Hypothesis Path Estimate Std. Estimate Std. Error Critical Ratio p-value Remarks

H1a A→BI .983 .776 .098 10.04 .001 Supported

H2a IR→A .412 .446 .059 6.976 .001 Supported
H3a ISt→A .084 .156 .030 2.802 .005 Supported
H4a ISh→A .125 .180 .043 2.918 .004 Supported
H5a IA→A .207 .279 .045 4.605 .012 Supported

Note: A = Attitude; BI= Behavioral Intention; IR= Information Retrieval; ISt = Information Storage; ISh = Information Sharing; IA= Information Application.

185
I. Arpaci Computers in Human Behavior 90 (2019) 181–187

Table 4
Detailed results for the OneR rule learner.
TP Rate FP Rate Precision Recall F-Measure MCC ROC Area PRC Area Class

10-fold cross-validation .648 .089 .687 .648 .667 .571 .780 .526 Low
.845 .430 .770 .845 .806 .434 .708 .748 Medium
.395 .042 .607 .395 .479 .427 .677 .324 High
Avg. .737 .297 .728 .737 .728 .465 .720 .638
Percentage split (66%) .650 .118 .565 .650 .605 .505 .766 .434 Low
.800 .429 .789 .800 .794 .374 .686 .764 Medium
.400 .056 .545 .400 .462 .394 .672 .304 High
Avg. .714 .316 .711 .714 .711 .402 .699 .636

Table 5 6.2. Practical implications

Performance results for the applied classifiers.
Algorithm CCIa (%) TPb-rate Precision F-Measure An effective management of information has a potential to improve
the students’ academic performance and efficiency of the learning tasks.
10-fold PSc 10-fold PS 10-fold PS 10-fold PS This implies that employing the MCC services for personal information
management should be supported and encouraged in the higher edu-
Naive Bayes 67.86 64.76 .679 .648 .664 .647 .666 .642
Logistic 72.08 71.43 .724 .714 .712 .720 .713 .696 cation by designing authentic learning environments and scaffolding
JRip 72.40 71.43 .714 .781 .711 .763 .711 .771 students in using such services. Universities should develop their own
AdaBoostM1 71.43 70.48 .705 .781 .604 .763 .650 .771 policies (i.e. bring-your-own-device policy) and cloud-based applica-
LWL 72.08 70.48 .721 .705 .706 .683 .689 .677 tions to maximize benefits of the such services. More importantly, in-
a structors should employ blended learning strategies that require in-
CCI: Correctly Classified Instances.
b formation management practices to promote educational use of the
TP: True Positive.
c
PS: Percentage Split. MCC services.

employed to explore whether the students' information management
practices and attitudes would predict the behavioral intentions. The
study tested a number of classification algorithms with different
methodologies, including bayesian networks, decision trees, if-then-else
rules, association rules, and neural networks. The results indicated that
the OneR rule learner performs slightly better than other classifiers. It is
worthy to mention that OneR (or one class) algorithm works well if one
of the attributes can predict the outcome (dependent) variable better
than the other attributes in the training data (Holte, 1993). The clas-
sifier model suggested that the attitude predicts the behavioral inten-
tion better than the other attributes (CCI = 0.74). In line with these
results, the structural model suggested that the attitude was a sig-
nificant factor in predicting the behavioral intention (β = 0.77). Fur-
ther, the SEM results suggested the students’ information management
practices and attitudes were significantly associated with the behavioral
intentions and accounted for 59% of the variance in the behavioral
intentions.
It is important to note that the relation between the attitudes and
the behavioral intentions explains a lot of variance in the model, which
is desirable, but not informative or useful. Conceptually, the attitudes
(toward use) and the intentions to (continue to) use are a (conceptual)
component of the later. However, this study overcomes this design flaw
by capitalizing on its two main strengths: 1) the complementary use of
parametric and nonparametric approaches, and 2) the extension of the
original theory (TRA) into the MCC area.
The SEM model included information management practices as
exogenous (independent) variables. In the model, both the attitudes and
behavioral intentions were endogenous (dependent) variables. The SEM
results identified the causal relationships among the endogenous vari-
ables and between the endogenous and exogenous variables. On the
other hand, the classifier model included the attitudes along with in-
formation management practices as predictors of the behavioral in-
tentions. Although, presentation of the two analyses (SEM vs. machine
learning) seems two separate results, considering the operationalization
of the same constructs in both model, indeed, this study presents a
“hybrid modeling” approach and not a mere dual-perspective approach.
6.3. Limitations and future directions
The SEM does not provide a complete solution to all statistical
problems of the prediction. On the other hand, it provides a powerful
means of testing a causal model based on a well-specified conceptual
theory. Accordingly, the complementary use of the SEM and the ma-
chine learning algorithms to develop a prediction model provided un-
ique information on the performance of the SEM-based prediction
compared to the other prediction models. Besides, using two ap-
proaches, one focused on contrasting/comparing and another on com-
pleting, help make robust predictions.
However, the study has a number of limitations. This study did not
account for the potential moderating role of cultural orientations in the
educational technology acceptance, as reported by (Arpaci, 2015;
Arpaci, Cetin Yardimci, & Turetken, 2015). Therefore, one should be
careful to generalize the findings to the samples with a different socio-
cultural background. Concerns reported by (Stergiou & Psannis, 2017)
may also affect the degree of adoption. Future studies may therefore
focus on the concerns of confidentiality, privacy, latency, performance
instability, data security, and lack of service-level-agreements. The fact
that an effective technology integration into the higher education re-
quires a techno-pedagogical approach and needs to consider the aspects
of pedagogy, curriculum, change management, competencies, and or-
ganizational readiness, implies that future studies should take into ac-
count all these aspects. Finally, a longitudinal study is recommended to
overcome limitations of the cross-sectional design.
References
Aepona (2018). Mobile cloud solutions. Retrieved from http://www.aepona.com/
solutions/mobile-cloud/, Accessed date: 16 July 2018.
Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human
Decision Processes, 50(2), 179–211.
Alavi, M., & Leidner, D. E. (2001). Review: Knowledge management and knowledge
management systems: Conceptual foundations and research issues. MIS Quarterly,
25(1), 107–136.
Arpaci, I. (2015). A comparative study of the effects of cultural differences on the
adoption of mobile learning. British Journal of Educational Technology, 46(4),
699–712. https://doi.org/10.1111/bjet.12160.
Arpaci, I. (2016). Understanding and predicting students' intention to use mobile cloud
storage services. Computers in Human Behavior, 58, 150–157. https://doi.org/10.
1016/j.chb.2015.12.067.

186
I. Arpaci
Arpaci, I. (2017). Antecedents and consequences of cloud computing adoption in edu-
cation to achieve knowledge management. Computers in Human Behavior, 70(5),
382–390. https://doi.org/10.1016/j.chb.2017.01.024.
Arpaci, I., Cetin Yardimci, Y., & Turetken, O. (2015). A cross-cultural analysis of smart-
phone adoption by Canadian and Turkish organizations. Journal of Global Information
Technology Management, 18(3), 214–238. https://doi.org/10.1080/1097198X.2015.
1080052.
Bartlett, M. S. (1951). The effect of standardization on a χ2 approximation in factor
analysis. Biometrika, 38(3/4), 337–344.
Benbasat, I., & Barki, H. (2007). Quo vadis TAM? Journal of the Association for Information
Systems, 8(4), 211–218.
Chen, S. L., Chen, J. H., & Lee, Y. H. (2018). A comparison of competing models for
understanding industrial organization's acceptance of cloud services. Sustainability,
10(3), 673.
Cuervo, E., Balasubramanian, A., Cho, D. K., Wolman, A., Saroiu, S., Chandra, R., et al.
(2010). MAUI: Making smartphones last longer with code offload. Proceedings of the
8th international conference on Mobile systems, applications, and services (pp. 49–62).
ACM.
Davenport, T. H., & Prusak, L. (2000). Working knowledge: How organizations manage what
they know. Boston, MA: Harvard Business School Press.
Davis, F. D. (1989). Perceived usefulness, perceived ease of use, and user acceptance of
information technology. MIS Quarterly, 13(3), 319–340.
Dinh, H. T., Lee, C., Niyato, D., & Wang, P. (2013). A survey of mobile cloud computing:
Architecture, applications, and approaches. Wireless Communications and Mobile
Computing, 13(18), 1587–1611.
Field, A. (2005). Discovering statistics using SPSS. Thousand Oaks, CA: SAGE.
Fishbein, M., & Ajzen, I. (1975). Belief, attitude, intention and behavior: An introduction to
theory and research. Reading, MA: Addison-Wesley.
Hair, J. F., Tatham, R. L., Anderson, R. E., & Black, W. (2006). Multivariate data analysis
(6th ed.). Upper Saddle River, NJ: Pearson Prentice Hall.
Holte, R. C. (1993). Very simple classification rules perform well on most commonly used
datasets. Machine Learning, 11(1), 63–90.
Kaiser, H. F. (1970). A second generation little jiffy. Psychometrika, 35(4), 401–415.
Kim, S. H., & Kim, J. K. (2018). Determinants of the adoption of mobile cloud computing
services: A principal-agent perspective. Information Development, 34(1), 44–63.
Kline, R. B. (2005). Principles and practice of structural equation modeling (2nd ed.). New
York: The Guilford Press.
Kotsiantis, S. B., Zaharakis, I., & Pintelas, P. (2007). Supervised machine learning: A re-
view of classification techniques. Emerging Artificial Intelligence Applications in
Computer Engineering, 160, 3–24.
Lee, Y. C. (2016). Why do people adopt cloud services? Gender differences. Social Science
Information, 55(1), 78–93.
Leech, N. L., Barrett, K. C., & Morgan, G. A. (2005). SPSS for intermediate statistics: Use and
interpretation (2nd ed.). Hillsdale, NJ: Lawrence Erlbaum.
Lian, J. W., Yen, D. C., & Wang, Y. T. (2014). An exploratory study to understand the
critical factors affecting the decision to adopt cloud computing in Taiwan hospital.
International Journal of Information Management, 34(1), 28–36.
Mitchell, H. J. (2003). Technology and knowledge management: Is technology just an
enabler or does it also add value? Knowledge management: Current issues and challenges
187
Computers in Human Behavior 90 (2019) 181–187
(pp. 66–78). Hershey: IRM Press.
Ngai, E. W., Xiu, L., & Chau, D. C. (2009). Application of data mining techniques in
customer relationship management: A literature review and classification. Expert
Systems with Applications, 36(2), 2592–2602.
Nistor, N. (2014). When technology acceptance models won't work: Non-significant in-
tention-behavior effects. Computers in Human Behavior, 34, 299–300. https://doi.org/
10.1016/j.chb.2014.02.052.
Nonaka, I. (1994). A dynamic theory of organizational knowledge creation. Organization
Science, 5(1), 14–37.
Nonaka, I., & Takeuchi, H. (1995). The knowledge-creating company: How Japanese com-
panies create the dynamics of innovation. New York: Oxford University Press.
Ogiela, M. R. (2017). Impact of ubiquitous computing technologies for modern in-
formation management. International Journal of Information Management, 1(37),
1517–1518.
Park, E., & Kim, K. J. (2014). An integrated adoption model of mobile cloud services:
Exploration of key determinants and extension of technology acceptance model.
Telematics and Informatics, 31(3), 376–385.
Rogers, E. M. (2003). Diffusion of innovations (5th ed.). New York, NY: Free Press.
Sharma, S. K., Al-Badi, A. H., Govindaluri, S. M., & Al-Kharusi, M. H. (2016a). Predicting
motivators of cloud computing adoption: A developing country perspective.
Computers in Human Behavior, 62, 61–69.
Sharma, S. K., Joshi, A., & Sharma, H. (2016b). A multi-analytical approach to predict the
Facebook usage in higher education. Computers in Human Behavior, 55, 340–353.
Shiau, W. L., & Chau, P. Y. (2016). Understanding behavioral intention to use a cloud
computing classroom: A multiple model comparison approach. Information &
Management, 53(3), 355–365.
Shon, T., Cho, J., Han, K., & Choi, H. (2014). Toward advanced mobile cloud computing
for the internet of things: Current issues and future direction. Mobile Networks and
Applications, 19(3), 404–413.
Stergiou, C., & Psannis, K. E. (2017). Recent advances delivered by mobile cloud com-
puting and internet of things for big data applications: A survey. International Journal
of Network Management, 27(3) e1930.
Subashini, S., & Kavitha, V. (2011). A survey on security issues in service delivery models
of cloud computing. Journal of Network and Computer Applications, 34(1), 1–11.
Tan, G. W. H., Ooi, K. B., Leong, L. Y., & Lin, B. (2014). Predicting the drivers of beha-
vioral intention to use mobile learning: A hybrid SEM-neural networks approach.
Computers in Human Behavior, 36, 198–213.
Turban, E., Sharda, R., & Delen, D. (2011). Decision support and business intelligence sys-
tems. Boston: Prentice Hall.
Ul Amin, R., Inayat, I., Shahzad, B., Saleem, K., & Aijun, L. (2017). An empirical study on
acceptance of secure healthcare service in Malaysia, Pakistan, and Saudi Arabia: A
mobile cloud computing perspective. Annals of Telecommunications, 72(5–6),
253–264.
Venkatesh, V., Morris, M. G., Davis, G. B., & Davis, F. D. (2003). User acceptance of
information technology: Toward a unified view. MIS Quarterly, 27(3), 425–478.
Wang, C. S., & Huang, Y. M. (2016). Acceptance of cloud services in face-to-face com-
puter-supported collaborative learning: A comparison between single-user mode and
multi-user mode. Innovations in Education & Teaching International, 53(6), 637–648.