In the past decades, computer and Internet technologies offer educators
and learners an innovative learning environment that can stimulate and
enhance the teaching and learning process (Reader and Hammond, 1994).
In an attempt to make the instructional resources more efficient and meet
the diverse needs of learners, many organizations and higher education
institutions around the world, have been integrated and utilized learning
objects into their e-learning systems. It is an idea to decompose existing
course materials into smaller (relative to the size of an entire course),
self-contained and modular pieces of instructional components that can
be reused a number of times in different learning contexts (Parrish, 2004).
Now, this concept has gained such broad acceptance and has filtered into
the fields of education. Many learning object repositories, such as Multimedia
Educational Resource for Learning and Online Teaching (http://www.merlot.org/),
Campus Alberta Repository of Educational Objects (http://aloha.netera.ca/),
Educational Object Economy (http://www.eoe.org.uk/)
and Wisconsin Online Resource Centre (http://www.wisc-online.com)
have been developed to cater for a variety of knowledge domains.
Like any information systems (Doll and Torkzadeh, 1994), the success
of learning object technology also depends on user satisfaction and acceptance.
A high level of user satisfaction reflects the users` willingness to accept
and continue using the technology (Stokes, 2001). The measurement of the
user perception (McMahon et al., 1999) and understand the factors
that promote the effective use of systems (Yi and Hwang, 2003) become
increasingly important to enhance our understanding and prediction of
the acceptance and utilization of educational technologies. Thus, learners`
behavioral intentions and acceptance of learning objects need to be explored.
As the Management Information Systems (MIS) discipline has much empirical
research of information technology acceptance in the management area,
it would be beneficial to study information technology acceptance in educational
contexts by building upon the foundations in both the education and MIS
||Technology acceptance model
Several intention-based models such as the Technology Acceptance
Model (TAM) (Davis, 1989) and the Theory of Planned Behavior (TPB) (Ajzen,
1991) have been proposed and empirically tested in the last decade in
understanding user adoption and usage of IT innovations. Those frameworks
have been applied to a variety of information technologies in different
contexts and populations (Ong et al., 2004; Saade and Bahli, 2005).
Among them, the TAM is one of the most influential and frequently tested
models and widely applied to explain general information technology adoption
in the MIS literature.
The TAM as shown in Fig. 1 is a specific model developed
to explain and predict users` computer usage behavior. It predicts user
acceptance based on the influence of two use beliefs: Perceived Usefulness
(PU) and Perceived Ease of Use (PEU). Both PU and PEU are posited as having
significant impact on a user`s attitude (AT) toward using the system.
Behavioral Intentions (BI) to use is jointly determined by a person`s
attitude toward using the system and its perceived usefulness. BI then
determines the Actual Use (AU) of the system. Davis (1993) had also suggested
that external factors might be important determinants in order to gain
more specific information that can more accurately assess the adoption
of the information systems.
The aim of this study was to examine and identify the underlying factors
and hypothesized causal relationship that determined learners` behavioral
intentions to adopt learning objects in higher education.
MATERIALS AND METHODS
Research model and hypotheses: The study model is formulated with
the constructs and variables gleaned from the literature of education
and MIS research to determine underlying factors and causal relationships
in predicting learners` acceptance of learning objects. A total of five
external variables from two perspectives (learning object characteristics
and individual characteristics) and four constructs were identified.
Based on the review of MIS research studies, within this study framework,
three learning object characteristics (technical quality, content quality
and pedagogical quality) were proposed as they are considered to be important
attributes for the development of digital content resources (Nesbit and
Han, 2005). Technical quality refers to technical attributes, such as
ease of use, turnaround time, accessibility and flexibility of the learning
objects. It is an implicit expectation that is important in supporting
and raising learners` confidence in learning object usability. Content
quality is related to how well the content is tailored to the needs of
the intended users. Learning objects should come with accurate, complete
and sufficient depth of content for a particular curriculum activity in
order to be meaningful. Pedagogical quality is related to whether the
learning object`s potential effectiveness as a teaching and learning tool
fits the context(s) in which it will be used to support the learning goal
that it claims. They are antecedent to perceive usefulness and perceived
ease of use of learning objects. Thus, the interrelationship of learning
object characteristics to perceived usefulness and perceived ease of use
will be used to evaluate the beliefs learners have toward the adoption
of learning objects.
Venkatesh and Davis (1996) theorized that perceptions about a new system`s
usefulness and ease of use are anchored on an individual`s general computer
self-efficacy. Thus an analysis from the learner perspective, such as
user general characteristics and specific entry competencies must be conducted
for the educational technology to be used effectively. Davis (1993) suggested
user characteristics to be mediated by the TAM that have an impact upon
behavioral intention to use. On the other hand, there have been numerous
studies involving the experience and attitude-behavior relationship. Ventatesh
and Davis (2000) found that experience directly and indirectly influences
system usage behavior through perceived usefulness and perceived ease
of use. This study posited self-efficacy and Internet/computer experience
as being mediated by perception of usefulness and ease of use to influence
behavioral intentions. Building upon prior related research foundations,
Fig. 2 portrays the preliminary Learning Object Acceptance
Model (LOAM) for this study which integrates not only the core determinants
of TAM, but also two external variables was studied.
This research model involves testing four sets of hypotheses as follows:
||Perceived usefulness of learning object is positively
influenced by the learning object characteristics of technical quality
(H1a), content quality (H1b) and pedagogical
quality (H1c), the individual characteristics of self-efficacy
(H1d) and Internet experience (H1e) and perceived
ease of use (H1f).
||Learning object acceptance model
||Perceived ease of use of learning object is positively
influenced by the learning object characteristics of technical quality
(H2a), content quality (H2b) and pedagogical
quality (H2c) and the individual characteristics of self-efficacy
(H2d) and Internet experience (H2e).
||Behavioral intention to use learning object is positively affected
by perceived usefulness (H3a) and perceived ease of use
||Actual use of the learning object is positively influenced by behavioral
intention to use.
Research design and procedures: This study utilized a web-based
survey to collect data for quantitative testing of the research model.
A review of the MIS literature was used to identify existing measures
for constructs, which had been used in previous research studies. The
scales for SE, IE, PU, PEU, BI and AU were adapted from MIS literature
studies (Compeau et al., 1999; Henry and Stone, 1995; Venkatesh
and Davis, 2000). However, for the learning object characteristics, no
previously validated items that matched our constructs of interest were
located. Therefore, items were developed based on features considered
to be important for learning objects as cited in the literature (Nesbit
and Han, 2005). Items were rewritten as necessary to fit the context of
The target population for the study consists of undergraduate IT students
who were enrolled in Faculty of Information Science and Technology (FIST),
Multimedia University. This study sought experience online learning users
who are familiar with web technologies in general sense and had the basic
ability using online learning system, in this regard the in house develops
online Multimedia Learning System (MMLS) so that they could evaluate the
learning objects based on their current online learning experience. All
students enrolled in Digital Systems course (sessions 2005/06 and 2006/07)
agreed to participate in this study, resulting in a sample of 601 potential
users of learning objects.
In this study, relevant learning objects for this course were retrieved
from various general repositories (e.g., Connexions-http://cnx.org/,
Multimedia Educational Resource for Learning-http://www.merlot.org/
and Online Teaching and Wisconsin Online Resource Centre- http://www.wisc-online.com),
which provide higher education level learning objects. These repositories
where selected for being among the few learning objects repositories that
permits public access, which made the study possible. In order to produce
cohesive and pedagogically sound learning materials and to effectively
search for relevant learning objects, the researcher designed a generic
structure of the Digital Systems course consisting of a series of electronic
folders, similar to the traditional course hierarchy (chapters, lessons
and topics) to hold the retrieved learning objects. Relevant learning
objects were linked into the syllabus each week from the lecture notes
with the aim of helping students to understand the more abstract and complex
aspects of learning content.
At the beginning of the semester, e-mails were sent to the instructors
to seek permission and to arrange time for their class students to participate
in the study. For both samples, the instructors provided a brief in-class
introduction to learning objects, describing the benefits of learning
objects and their relevance to the curriculum. Following a demonstration
of learning objects session, respondents completed the first survey which
consists of the demographics questionnaires. After 12 weeks, the second
survey questionnaire was conducted to evaluate their post-usage of learning
objects. Responses from the two surveys for sample 1 and sample 2 were
matched to create a single record for each respondent.
DATA ANALYSIS AND RESULTS
For data analysis, the two-step approach to model construction and testing
was adapted (Gerbing and Anderson, 1988), by using the computer program
AMOS 4.0. Sample 1 was used to derive a final structural model while sample
2 was used to further ensure the consistency of the structural model from
Demographic data: In all, 660 respondents participated in the
study. There were 312 respondents from a total of 342 students in Sample
1 with a response rate of 91.2% and 289 respondents from a total of 318
students in sample 2 with a response rate of 90.8%. Of the 660 students,
a total of 601 respondents completed the surveys with a response rate
of 91.1%. Majority of the respondents have 2 to 4 years of computer experiences
and spend about 2 to 4 h on the Internet everyday.
The measurement model: The analysis of the measurement model was
to refine the LOAM by eliminating measured variables or latent constructs
that did not fit in well with the initial Confirmatory Factor Analysis
(CFA) using Sample 1 observations. The overall fit for the initial measurement
model as presented in Table 1 was reasonable fit. All
the fit criteria were within the acceptance level except Goodness of Fix
Index (GFI) (0.873) which was below the 0.90 acceptance level. Based on
the standardized residuals and modification indices, the initial measurement
model could be further improved.
As a result, the final measurement model indicated a good model fit.
The ratio of Chi-square statistics and its degree of freedom (χ2/df)
(1.030) measure was less than 3.0, the Root Mean Squared Error of
Approximation (RMSEA) (0.010) was less than 0.05 indicating a close fit,
the Goodness of Fix Index (GFI) (0.919), Normed Fit Index (NFI) (0.963)
and Comparative Fit Index (CFI) (0.999) were all above the 0.90 acceptable
levels and the Adjusted Goodness of Fix Index (AGFI) (0.900) was also
above its 0.80 threshold value. In summary, results from the final measurement
model showed a good fit.
The next step assessed the reliability and validity of constructs and
indicators. Convergent validity of scale items was assessed using three
criteria (reliability, composite reliability and average variance extracted)
as recommended by Fornell and Larcker (1981). The standardized CFA loadings
for all scale items exceeded the minimum loading criterion of 0.70. Furthermore,
the composite reliabilities of all factors also exceeded the required
minimum of 0.80. The average variance extracted values of all constructs
exceeded the threshold value of 0.50. Hence all three conditions for convergent
validity were met as shown in Table 2.
For discriminant validity testing, this was obtained by comparing the
square roots of the average variances extracted from each latent construct
with the correlations between factors (Segars and Grover, 1998). As shown
in Table 3, all the square roots of the average variances
extracted were greater than the correlations between factors. Hence the
discriminant validity criterion was also met for CFA models, giving further
confidence in the adequacy of the measurement scales. The results of the
confirmatory factor analysis indicated that the best fitting measurement
model was acceptable. Therefore, the derived measurement model was incorporated
into the structural equation model analysis with latent variables.
The structural model: Building upon the best fitting measurement
model, a path analysis for the Structure Equation Model (SEM) with latent
variables was performed to evaluate the hypothesized causal relationships
that predict learners` behavioral intention to use and actual use of learning
objects. As shown in Table 4, this initial SEM model
indicated a reasonable fit to the data with χ2/df = 1.957,
RMSEA = 0.055, GFI = 0.854, AGFI = 0.830, NFI = 0.926 and CFI = 0.962.
Further inspection of the modification indices suggested the addition
of three correlation paths for technical quality, content quality and
pedagogical quality. We respecified the initial SEM model. As a result,
the initial SEM model after modification performed satisfactorily with
GFI, NFI and CFI exceeding 0.90 and AGFI of 0.899. The values of RMSEA
and χ2/df were within acceptable thresholds.
||Measurement model fit comparison
|N = 312, a: Recommended values
||Convergent validity for best fitting measurement model
||Inter constructs correlations
|Diagonals represent the square roots of average variances
extracted and the other matrix entries are the factor correlations
||Structural model fit comparison-between initial and
final SEM model
|N = 312, a: Recommended values
Finally, in order to test the consistency of the structural model in
predicting the adoption of learning object, the final SEM model was tested
using sample 2 data collected from learners who underwent a similar process.
Application of the final SEM model to the 289 subjects in sample 2 generated
acceptable fit indices (χ2/df = 1.151, RMSEA = 0.023,
GFI = 0.900, NFI = 0.958, CFI = 0.994, AGFI = 0.883) as shown in Table
5. The same significant paths were once again significant and the
non-significant paths remained non-significant. This result provided further
evidence of the consistency of the final SEM model.
Hypotheses testing: The proposed LOAM hypothesized fourteen relationships.
The results of the analysis of the final structural model, including standardized
direct (path coefficients), indirect, total effects, path significances
and variance explained (R2-values) for each dependent variable
are shown in Table 6.
||Comparison of fit indices for sample 1 and 2
|a: Recommended values
||Standardized causal effects for the final structural
|N = 312, *p< 0.001, **p<0 01
Overall, none of the individual
characteristics variables had significant effect on the users` beliefs.
However, all of the remaining hypothesized effects were positive and statistically
significant, indicating that the three learning objects characteristics
and two users` beliefs were important determinants, more so than the individual
Starting from the perceived usefulness of learning objects, pedagogical
quality (β = 0.374, p<0.001) had significant positive effects
on it. The total effect of this determinant on perceived usefulness was
0.426, primarily due to its significant direct effects. As expected, perceived
ease of use had significant positive effects on perceived usefulness (β
= 0.347, p<0.001). These determinants explained about 64% of the variance
of perceived usefulness of learning objects. Therefore, hypotheses H1c
and H1f were supported. The total effects of self-efficacy,
Internet experience technical quality and content quality were insignificant.
As to perceived ease of use, the major determinant of perceives ease
of use was technical quality (β = 0.381, p<0.001), followed by
content quality (β = 0.355, p<0.001) and pedagogical quality (β
= 0.151, p<0.01). All total effects were statistically positive significant
and solely due to direct effects. The total effect of self-efficacy and
Internet experience were insignificant. Therefore, hypothesized H2a,
H2b and H2c were supported. These determinants accounted
for approximately 65% of the variance of perceived ease of use.
With regard to behavioral intention to use learning objects, about 58%
of the variance in behavioral intention could be explained by perceived
ease of use (β = 0.228, p<0.01) and perceived usefulness (β
= 0.564, p<0.001). The major determinant was perceived usefulness with
a total effect of 0.564, solely due to the direct effect followed by perceived
ease of use with a total effect of 0.424, mainly due to direct effect
(0.228) and partly due to indirect effect (0.196). Therefore, hypotheses
H3a and H3b were supported.
Finally, behavioral intention to use had a significant positive effect
on actual use (β = 0.544, p<0.001). Therefore, hypotheses H4
was supported. The model accounted for approximately 30% of the variance
of actual use of learning objects.
||Revised LOAM; *p<0.001, **p<0.01
The proposed Learning Object Adoption Model (LOAM) hypothesized fourteen
relationships. Most of the parameter estimates exhibit correct signs,
appropriate standard errors and significant critical ratios. The paths
between IE/PEU, SE/PEU, TQ/PU, CQ/PU, IE/PU and SE/PU were insignificant
across all data sets. Given that the model behaved consistently between
data sets, all insignificant six paths (IE/PEU, SE/PEU, TQ/PU, CQ/PU,
IE/PU and SE/PU) dropped and two constructs (IE and SE) deleted. The remaining
coefficients were all significant. The revised LOAM is shown in
A test of the proposed model indicated that the learning objects characteristics
had significant effect on the users` belief constructs which, was consistent
with the influence of more general system characteristics reported in
studies of other information technologies (Jackson et al., 1997).
On the other hand, individual characteristics were found to have no significant
effect on users` belief constructs.
After taking into account the learning object characteristics and users` beliefs,
individual characteristics (Internet experience and self-efficacy) was found
to have no statistically significant causal relationships on the belief constructs
and behavioral intention. This means that individual with computer efficacy
and Internet experiences do not guarantee the usage of learning objects. This
is especially true for learning objects where the contents are the essence and
the technical and pedagogy aspects play supportive roles in promoting the use
of learning objects. Thus, educators and instructional designers of learning
objects should ensure the compatibility between learning objects and users`
needs in order to enhance learning objects` adoption for individual learning.
The importance of the learning objects characteristics and user`s beliefs
in influencing the behavioral intention and actual use of learning objects
have several implications for researchers and practitioners. First, it
highlights the importance of attending to learning objects characteristics,
especially in usefulness and ease of use when learning objects are designed
and developed. Thus, educators and instructional designers of learning
objects should carefully consider the needs and values of learning object
users. For example, learners who perceived that the learning objects had
better turnaround time and flexible which allowed for a better feeling
of control over course content would indicated that the learning objects
were easier to use. Moreover, learners who indicated that the learning
objects fitted in with their learning contexts with comprehensive, up-to-date,
easy to comprehend contents together with appropriate pedagogy features
to support their learning goals helped them to become committed towards
Thought should be given to the functionality of the learning objects
and any interactions to be used should be carefully designed as the interactivity
is influenced by the degree of the availability of learners` control (Robertson,
1998; Stoney and Wild, 1998) as well as the availability of the functional
features that encourage users to actively learn. Several recommendations
serve as useful guidelines for the design of learning objects which stem
from the research literature and study cited above, supplemented by the
personal experience of the researchers are described here.
Learners` control: Learners` control plays an important part in
creating a learning environment to fulfill students` needs. By giving
learners greater control over various aspects of instruction, such as
pacing and sequencing, they can tailor the instructions to their own style
of learning, thereby enhancing the efficacy and efficiency of learning.
The following recommendations may provide greater freedom and should motivate
students, thus enhancing learning even more.
||Sequence control: Learning objects should have
the ability for non-linear instruction which allow users to branch
off and explore different content through links may be used to foster
control of sequence such as to skip and revisit topics.
||Content control: Learning objects should provide learners
with the opportunity to control over the learning activities they
wish to explore such as viewing content, examples and practicing problems.
||Pace control: Learning objects should allow learners to cover
the learning content and activities at the level that they are comfortable
with (e.g., slower, faster or take a break and return at a later date
Graphics and animations: Students are particularly receptive to
non-textual elements such as graphics and animations because these can
motivate them to focus and help recall information to assist in the development
of the learning concept (Rieber, 1996). Rieber (1996) also pointed that
students under animation-based instruction required less time to retrieve
the information they learned. The following recommendations may encourage
students to make appropriate use of learning objects:
||Analyze the relevance of graphics or animations to the
particular learning objectives. It is important that they should be
used for instructional and motivational purposes to achieve the learning
||Maintain a balance between text and graphical information to avoid
overloading the learners` working memory. When graphics are used,
there should be a text alternative to the image to enable the student
to relate the two simultaneous representations. But text-intensive
content should be avoided to overcome the split attention effect as
suggested by Kalyuga et al. (2000).
||Complex animations may not be optimal for beginners (Rieber, 1996).
It is better to opt for graphics or animations which are simpler in
design to provide learners with experience of each of the learning
components separately before presenting an entire interactive complex
Audio assets: Three types of audio assets commonly used are music,
narration voice-overs and sound effects. They can be used for both effects
and information and together with print to form a useful alternative and
aid to reading alone. The following recommendations are seen as important
and may encourage students to make appropriate use of learning objects:
||Audio assets can be toggled on and off with ability
for user to control the volume.
||Narration voice-overs must occur simultaneously with the relevant
animation and accessible transcripts should always be provided in
conjunction with audio or animations. All the key steps in the learning
activities should be emphasized by speaking important words or phrases
in a louder and deeper voice to direct the attention of the learner.
||Audio effects should be free from extraneous information (e.g.,
background noise) which can distract some groups of learners and impede
Feedback: Feedback has been shown to play an important role in
the students` learning. In general, instructional feedbacks provide information
to students about the correctness of their learning activities. Because
of the importance of feedback in online environments, the following recommendations
will increase the effectiveness of learning objects:
||Learning object should design with feedback to explain
what makes the right answer correct and what`s wrong about the learner`s
answer in order to lead to more meaningful learning for completing
the assignments (Moreno and Mayer, 2005).
||The feedback should focus on improving the skills needed to achieve
the learning objectives.
Given the increasing use of learning objects, a better understanding
and implementation of effective learning objects will enhance the use
and educational value of such educational technology. This study contributes
to the understanding of user acceptance of learning objects by identify
the underlying factors and causal relationships that predicted learners`
behavioral intention and subsequent actual use of learning objects. The
proposed study model, LOAM demonstrated that the learning objects characteristics
(pedagogical quality, technical quality and content quality) were all
important determinants of users` belief constructs. Specific characteristics
of the learners (self-efficacy and Internet experience) had no influence
upon users` belief constructs. This indicated that learning object characteristics
were important external stimuli for learners as they formed the perception
and intention to use learning objects. This study provides new insight
on the determinants of perceptions and user acceptance of learning objects.
However, it is merely a stepping stone in the realm of learning object
diffusion; no single research is conclusive of facts. More studies in
the future are needed to verify and refine the findings of this study
to expand the knowledge base on important determinants of learning objects
use that will assist educators to understand the factors leading to an
effective and efficient adoption of learning objects.