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
cloud computing services in higher education
T
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:
Information management
Cloud computing
Artificial intelligence
Machine learning
The decision to integrate mobile cloud computing (MCC) in education without determining optimal use scenarios is a universal problem as the adoption of such services becomes widespread. Accordingly, this study
developed and validated a predictive model that explains the role of students' information management (i.e.
retrieve, store, share, and apply) practices in predicting their attitudes toward using the MCC services for
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
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 maintenance costs, and thereby, promoted the efficiency and green technology (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
“infrastructure as a service” (IaaS) for information storage and management. On the other hand, these services provide enterprises with the
IaaS for an enterprise infrastructure building and “software as a service” (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 allows an efficient data synchronization and information management in
a ubiquitous environment (Park & Kim, 2014). Besides, energy efficiency is a critical issue since battery power of the mobile technologies
is limited. Migration of the complex processing from mobile technologies 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 managed (Ogiela, 2017). For example, using mobile Internet on smartphones 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.
Computers in Human Behavior 90 (2019) 181–187
I. Arpaci
3. Theoretical background
applications allow them to share information with peers. More importantly, 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 widespread. Thus, this study investigated adoption of the MCC among undergraduate students by focusing on the use of these services for personal information management.
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 validate 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 (complementary) 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 management 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 educational settings.
The TRA postulated that beliefs affect attitudes, which positively
influence behavioral intentions, whereas the intentions ultimately influence 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 different 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 intentions) are predicted by the attitudes, which are predicted by the students’ information management practices such as information retrieval,
storage, sharing, and application.
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. performance-expectancy, social influence, effort-expectancy) were significant 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 (TechnologyOrganization-Environment) framework. The results suggested that
technical competence, data security, cost, complexity, and top management support were critical factors in the adoption. Chen, Chen, and
Lee (2018) investigated key factors affecting the organizations' adoption of the cloud services. Their results suggested that top-managementsupport 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 Innovations” (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 satisfaction 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 predicting the students' attitudes towards using the MCC services. In another study, Sharma, Al-Badi, Govindaluri, and Al-Kharusi (2016) developed a hybrid model based on the TAM to predict factors affecting
the adoption of the cloud computing by IT professionals. They employed 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.
Information
Retrieval
Information
Storage
Attitude
Information
Sharing
.77**
R2 = .33, e = .17
Behavioral
Intention
R2 = .59, e = .17
Information
Application
*p< .01, **p< .001, Chi-Square= 393.25, DF= 235, Chi-Square/DF=1.67
Fig. 1. The structural model.
182
Computers in Human Behavior 90 (2019) 181–187
I. Arpaci
Rogers's (2003, p.19) DOI theory suggested that the stages by which
an individual adopts a new technology or innovation, whereby diffusion is accomplished include “awareness of the need for an innovation,
decision to adopt (or reject) the innovation, initial use of the innovation
to test it, and continued use of the innovation.” Accordingly, a training
program introduced the participants to personal information management, which refers to the practice of activities an individual performs to
create or acquire, store, retrieve, share, and apply the information required to complete the tasks. The participants employed the mobile web
services (i.e. search, maps), mobile cloud applications (i.e. Google
Drive, iCloud, Dropbox) and social media applications to perform the
training tasks.
the use of such services.
3.1. Hypotheses
4.1. Sample
3.1.1. Behavioral intentions and attitudes
“Behavioral intention” was defined as “the degree of an individual's
belief that he or she will continue to use a system” (Arpaci, 2017, p.
384). However, attitudes towards using a technology was defined as “an
individual's overall affective reaction to use a particular system”
(Arpaci, 2017, p. 384). This study therefore hypothesized that the attitude toward using the MCC services would be positively associated
with the behavioral intention to use such services for educational
purposes. Further, the attitude would be a predictor of the behavioral
intention. Therefore;
A sample of 308 university students were recruited from a public
university in Turkey. The study included 186 females (60.4%) and 122
(39.6%) males with a mean age of 21.88 years (SD = 2.37). Majority of
the students (42.5%) were seniors; 12.3% were freshmen, 27.6% were
sophomore, and 17.5% were juniors.
H4b. Information sharing would predict the behavioral intention to use
the MCC services for educational purposes.
H5a. An increase in the volume of information that's applied via the use
of the MCC services is positively associated with the attitude towards
the use of such services.
H5b. Information application would predict the behavioral intention to
use the MCC services for educational purposes.
4. Method
4.2. Procedure
IRB of the affiliated university approved the research and the procedures complied with the ethical standards of the institutional board
guidelines. All participants were informed about the purpose of the
research after an informed consent was obtained. The study was conducted in regularly scheduled IT classes (4 h per week) during 14
weeks. The participants performed practical implementations of the
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
services (i.e. OneDrive, Dropbox, and Google Drive), 3) Form a communities of practice and use the MCC services along with social networks to share files, communicate with others, and build a common
cognition to complete the tasks. Finally, the survey instrument with
explicit assurance that anonymity is guaranteed was administered
during the last week of the regular classes.
H1a. Attitude towards using the MCC services would be positively
associated with the behavioral intention.
H1b. Attitude towards using the MCC services would predict the
behavioral intention.
3.1.2. Information management practices
Information is processed, categorized, organized or structured data
(Davenport & Prusak, 2000; Nonaka, 1994). However, knowledge is a
combination of skills, information, experience, and insights (Nonaka &
Takeuchi, 1995). Knowledge includes organizational routines, values,
and practices; therefore, this study preferred to use the term “information management” instead of “knowledge management.” This
study defined information management as practices that help retrieve
and store information, share it with others, and apply it to solve problems and make better and timely decisions (Alavi & Leidner, 2001;
Mitchell, 2003). Turban, Sharda, and Delen (2011) suggested the use of
information technologies to enhance information management practices of retrieval and access, transfer, storage, and application. In the
same vein, the present study suggested the integration of smart mobile
technologies into the higher education to support personal information
management. The more information managed by using the MCC services the more favorable the attitude towards using such services for
educational purposes. Further, the students’ information management
practices would be a predictor of the behavioral intention to use the
MCC services for educational purposes. Accordingly,
4.3. Instruments
The measurement instrument for the variables “behavioral intentions” and “attitudes” was a Likert scale adapted from Ajzen's theory
(1991). Moreover, the measurement instrument for the information
management practices of retrieval, storage, sharing, and application
was adapted from a previously validated and reliable instrument
(Arpaci, 2017). The final instrument has 24 items with a five-point
Likert-type scale ranging from “strongly disagree” to “strongly agree.”
5. Results
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
towards the use of such services.
IBM SPSS (ver. 25) and IBM AMOS (ver. 23) were used to screen and
analyze the data. Construct validity along with reliability of the scales
were assessed by conducting an EFA (See Table 1). Results revealed that
each item had a communality value and a factor loading greater than
the critical threshold (0.40) suggested by Field (2005). Cronbach's
alpha reliability analysis results indicated a good homogeneity and
reliability among the items. The AVE values were greater than 0.50,
indicating adequate convergent validity for all constructs (Hair,
Tatham, Anderson, & Black, 2006).
H2b. Information retrieval would predict the behavioral intention to
use the MCC services for educational purposes.
H3a. An increase in the volume of information that's stored via the use
of the MCC services is positively associated with the attitude towards
the use of such services.
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.
Item
α
Item-total correlation
Factor load
Communality
Total variance
explained
CR
AVE
Behavioral intention
CUI1. “I intend to continue to use mobile cloud services for educational purposes in the future.”
CUI2. “I plan to continue to use mobile cloud services for personal information management in the future.”
CUI3. “I predict that I would frequently use mobile cloud services in the future.”
CUI4. “I predict that I would frequently use mobile cloud services for educational purposes.”
CUI5. “I predict that I would frequently use mobile cloud services for personal information management.”
A1. “Using mobile cloud services for educational purposes is a good idea.”
A2. “Using mobile cloud services to create a retrievable archive of personal information is pleasant.”
A3. “Using social media applications on my smartphone to exchange information is fun.”
A4. “Using mobile cloud services for personal information management is a wise idea.”
A5. “Using mobile cloud services for educational purposes makes learning more interesting.”
A6. “I like the idea of using mobile cloud services for personal information management.”
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.”
IR3. “Using mobile cloud services on my smartphone enables quick access to learning materials and information
resources.”
IR4. “Using mobile cloud services on my smartphone enables ubiquitous access to e-databases and e-journals.”
IS1. “Using mobile cloud services on my smartphone enables the retrievable storage of electronic information and
documents.”
IS2. “I store course materials and documents using mobile cloud services on my smartphone.”
IS3. “Using mobile cloud services enables me to store learning materials and information with ubiquitous access.”
ISh1. “Using social media applications on my smartphone enables me to exchange information and documents with
classmates.”
ISh2. “I share course materials and information with classmates using mobile cloud services on my smartphone.”
ISh3. “I share learning materials and documents with classmates using mobile cloud services on my smartphone.”
IA1: “I apply knowledge and experience gained by using mobile cloud services to complete learning tasks.”
IA2. “I use knowledge obtained by using mobile cloud services in decision making processes.”
IA3. “I employ knowledge and intelligence gained from using mobile cloud services in problem-solving activities.”
.91
.76
.78
.77
.79
.79
.73
.78
.67
.75
.68
.67
.75
.85
.86
.86
.87
.87
.82
.86
.78
.83
.78
.77
.86
.72
.75
.74
.76
.76
.67
.74
.61
.69
.61
.60
.74
74.34
.91
.68
65.30
.89
.59
78.43
.91
.51
.85
.84
.92
.92
.84
.84
.91
.74
.81
.85
.91
.72
.83
84.06
.91
.52
.71
.83
.80
.51
.93
.91
.80
.86
.83
.64
64.61
.71
.51
.62
.48
.64
.73
.71
.86
.75
.83
.89
.88
.74
.57
.69
.79
.67
74.72
.83
.53
Attitude
184
Information retrieval
Information storage
Information sharing
Information application
.89
.91
.83
Computers in Human Behavior 90 (2019) 181–187
Construct
Computers in Human Behavior 90 (2019) 181–187
I. Arpaci
attributes of the students’ information management practices and their
attitudes. Behavioral intentions and attitudes have nominal values and
classified as low, medium or high by using the Mean ± SD combination. On the other hand, the attributes of information retrieval, storage,
sharing, and application have numeric values.
The predictive model was tested by using Weka (ver. 3.8.1) based
on the most common classification algorithms, which classify data into
a number of pre-defined classes with different learning schemas. The
results shown in Table 4 provided performance of the best performance
classifier based on two test modes; 10-fold cross-validation and percentage 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 behavioral intentions with an accuracy of 74%, and thereby, H1b, H2b, H3b,
H4b, and H5b were supported.
Table 2
KMO and Bartlett's test results.
Behavioral intention
Attitude
Information retrieval
Information storage
Information sharing
Information application
KMO
Chi-Square
Sig.
.88
.88
.83
.75
.65
.71
1022.27
1001.80
839.40
594.58
198.79
357.46
.001
.001
.001
.001
.001
.001
(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
6. Discussion and conclusion
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 information retrieval, information storage, information sharing, and information 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 impact on the attitudes with a standardized coefficient of 0.16.
The results shown in Table 3 suggested the attitudes were significantly 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).
6.1. Research implications
Using the MCC services provides students with ubiquitous access to
instructional materials and information anytime and anywhere. In addition, 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 decision to integrate these services in educational settings requires determining 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 investigating 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-analytical 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 between 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 attitudes were significantly associated with the behavioral intentions to use
the MCC services for educational purposes.
In a complementary study, machine learning algorithms were
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 classifying 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 technique, 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
Table 3
The SEM results.
Hypothesis
Path
Estimate
Std. Estimate
Std. Error
Critical Ratio
p-value
Remarks
H1a
H2a
H3a
H4a
H5a
A→BI
IR→A
ISt→A
ISh→A
IA→A
.983
.412
.084
.125
.207
.776
.446
.156
.180
.279
.098
.059
.030
.043
.045
10.04
6.976
2.802
2.918
4.605
.001
.001
.005
.004
.012
Supported
Supported
Supported
Supported
Supported
Note: A = Attitude; BI= Behavioral Intention; IR= Information Retrieval; ISt = Information Storage; ISh = Information Sharing; IA= Information Application.
185
Computers in Human Behavior 90 (2019) 181–187
I. Arpaci
Table 4
Detailed results for the OneR rule learner.
10-fold cross-validation
Avg.
Percentage split (66%)
Avg.
TP Rate
FP Rate
Precision
Recall
F-Measure
MCC
ROC Area
PRC Area
Class
.648
.845
.395
.737
.650
.800
.400
.714
.089
.430
.042
.297
.118
.429
.056
.316
.687
.770
.607
.728
.565
.789
.545
.711
.648
.845
.395
.737
.650
.800
.400
.714
.667
.806
.479
.728
.605
.794
.462
.711
.571
.434
.427
.465
.505
.374
.394
.402
.780
.708
.677
.720
.766
.686
.672
.699
.526
.748
.324
.638
.434
.764
.304
.636
Low
Medium
High
6.2. Practical implications
Table 5
Performance results for the applied classifiers.
Algorithm
Naive Bayes
Logistic
JRip
AdaBoostM1
LWL
a
b
c
CCIa (%)
TPb-rate
Low
Medium
High
Precision
An effective management of information has a potential to improve
the students’ academic performance and efficiency of the learning tasks.
This implies 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
students in using such services. Universities should develop their own
policies (i.e. bring-your-own-device policy) and cloud-based applications to maximize benefits of the such services. More importantly, instructors should employ blended learning strategies that require information management practices to promote educational use of the
MCC services.
F-Measure
10-fold
PSc
10-fold
PS
10-fold
PS
10-fold
PS
67.86
72.08
72.40
71.43
72.08
64.76
71.43
71.43
70.48
70.48
.679
.724
.714
.705
.721
.648
.714
.781
.781
.705
.664
.712
.711
.604
.706
.647
.720
.763
.763
.683
.666
.713
.711
.650
.689
.642
.696
.771
.771
.677
CCI: Correctly Classified Instances.
TP: True Positive.
PS: Percentage Split.
6.3. Limitations and future directions
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 classifier model suggested that the attitude predicts the behavioral intention better than the other attributes (CCI = 0.74). In line with these
results, the structural model suggested that the attitude was a significant factor in predicting the behavioral intention (β = 0.77). Further, 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 variables and between the endogenous and exogenous variables. On the
other hand, the classifier model included the attitudes along with information management practices as predictors of the behavioral intentions. 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.
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 machine learning algorithms to develop a prediction model provided unique information on the performance of the SEM-based prediction
compared to the other prediction models. Besides, using two approaches, one focused on contrasting/comparing and another on completing, 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 sociocultural 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 requires a techno-pedagogical approach and needs to consider the aspects
of pedagogy, curriculum, change management, competencies, and organizational readiness, implies that future studies should take into account 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
Computers in Human Behavior 90 (2019) 181–187
I. Arpaci
(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 intention-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 companies create the dynamics of innovation. New York: Oxford University Press.
Ogiela, M. R. (2017). Impact of ubiquitous computing technologies for modern information 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 computing 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 behavioral 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 systems. 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 computer-supported collaborative learning: A comparison between single-user mode and
multi-user mode. Innovations in Education & Teaching International, 53(6), 637–648.
Arpaci, I. (2017). Antecedents and consequences of cloud computing adoption in education 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 smartphone 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 review 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