1 Introduction

The continuous development of Information and Communication Technologies (ICTs) has created new opportunities within the domain of education. ICTs are becoming vital tools in enhancing the quality of learning and teaching (Lin 2007; Kent et al. 2016). As such, there has been an increase in the uptake of eLearning technologies such as learning management systems (LMS) throughout the world in order to create a novel learning strategy and enhance a collaborative and effective learning environment for both students and instructors (Kim and Park 2018). LMS are web-based applications used to administer courses. They provide the ability to track students’ progress and can also serve as content management systems that facilitate access to resources required for courses. Since an LMS is web-based, access to course resources is available for any internet-ready device from anywhere and at any time.

Developing countries are not left alone as countries such as Indonesia (Kim and Park 2018), Egypt (Abdel-Wahab 2008), Jordan (Abbad et al. 2009) and Tanzania (Maleko Munguatosha et al. 2011) have invested significant resources in implementing LMS. The reason for this uptake could be as a result of the benefits of LMS which include enhancing the teaching and learning process (Salloum and Shaalan 2018), providing access to the education curriculum, expanding educational opportunities and reducing the long-term costs of learning (Lwoga 2014). Other studies have shown that LMS motivate students’ interaction with other students and their instructors and ease communication (Arkorful and Abaidoo 2015). Additionally, an LMS is an effective tool for instruction and has been shown to increase in knowledge (Salter et al. 2014).

In Nigeria, the National Policy on ICT recognizes the importance of ICT in the education sector (NITDA 2012). Many stakeholders in the Nigeria higher education sector have started to appreciate the effectiveness of integrating technology in teaching and learning (Yakubu et al. 2019). However, just like many developing countries, the uptake of the use of ICTs has been very slow with few universities that have incorporated the use of technology, specifically LMS, for teaching and learning (Yakubu et al. 2019). Also, the few institutions that have adopted the use of LMS have opted for proprietary software; this is especially true for public (government-owned) universities. These proprietary systems are designed without taking into consideration, neither the pedagogical aspects of eLearning nor the student requirements for their learning needs (Eberendu 2015). This result in the presence of anomalies and a system void of the expected benefits of LMS (Adebayo and Abdulhamid 2014).

Furthermore, prior research has shown that the implementation and use of LMS in institutions of higher education in developing countries have continued to fail (Eberendu 2015). Salloum et al. (2018) argue that many students do not persist with learning how to use LMS after the initial experience with using the system. Thus, there is a need to understand the factors that contribute to the students’ acceptance of LMS in developing countries after deployment. While there is little existing research investigating this phenomenon in Nigerian universities (Adewole-Odeshi 2014; Ayodele et al. 2016; Ogunbase 2014; Olatunbosun et al. 2015), none of these studies have taken into account certain significant and influential factors in the context of eLearning, such as instructor quality, learning value and course quality. Research has shown that these factors contribute to the student’s acceptance of eLearning systems (Cheng 2012; Lwoga 2014; Ain et al. 2016). The main aim of this study is to identify the factors that facilitate Nigerian students’ acceptance of learning management systems, based on a conceptual model we developed. The model adopts constructs drawn from renowned technology acceptance theories as well as prior research on the acceptance of eLearning systems.

This study contributes to the literature by identifying the key factors that influence the acceptance of LMS by Nigerian students in tertiary institutions. Practically, findings of the study will guide LMS developers to develop systems that are beneficial to the students and instructors. This research is structured as follows:

  1. 1.

    Literature review section presents a background on the research and reviews previous studies that use established theories to examine the factors that influence LMS acceptance by students.

  2. 2.

    Section 3 discusses the research model, constructs, and hypotheses to be tested by this study.

  3. 3.

    Section 4 presents the methods and materials used to direct the study.

  4. 4.

    Section 5 presents the results of the data analysis.

  5. 5.

    Section 6 provides the discussion, research implications, and limitations of the study.

  6. 6.

    The study concludes with section 7, where the key findings are summarized.

2 Literature review

There is a wide range of literature examining the acceptance of e-learning systems such as LMS. Several theories have been postulated to explain why users accept or adopt technology. Theories such as the technology acceptance theory (TAM) (Davis 1989), unified theory of acceptance and use of technology (UTAUT) (Venkatesh et al. 2003), and the DeLone and McLean IS success model (DeLone and McLean 2003) are the most frequently used theories in explaining the acceptance of technology by users especially in the domain of eLearning systems use.

Technology acceptance model (TAM), by Davis (1989), is one of the most popular models used to explain and predict user acceptance of technology by examining individuals’ attitudes and perceptions of technologies. In the context of eLearning, several studies have used the TAM model to investigate the acceptance of eLearning systems, such as (Mohammadi 2015; Tarhini et al. 2017; Al-Azawei et al. 2017; Ibrahim et al. 2017; Rabaa'i 2016). Another popular model used to explain the acceptance of technology, due to the limitations of TAM in explaining an individuals acceptance of technology, is the unified theory of acceptance and use of technology (UTAUT) (Venkatesh et al. 2003). With regard to eLearning applications, UTAUT has also been used extensively in the investigation of students acceptance of eLearning systems (Alshehri et al. 2019; Ain et al. 2016; Salloum and Shaalan 2018; Yakubu and Dasuki 2018; Raman et al. 2014). While the UTAUT model has been empirically validated in various contexts, the model was extended to explain consumers’ acceptance and use of technology.

In the extended UTAUT model (Venkatesh et al. 2012), three constructs namely hedonic motivation, habit, and price value were added to the original UTAUT model (Venkatesh et al. 2003) in order to further explain the adoption of technology by individuals. Ain et al. (2016) used the extended UTAUT model to explain students’ acceptance of an LMS by replacing the price value construct. Learning value replaced the price value construct because a monetary value could not be attributed to the use of LMS by students. Instead, the value obtained by students includes the time and effort invested in using the LMS.

The DeLone and McLean information systems success model is another frequently used model that attempts to evaluate the success of information systems via different perspectives. Similar to the UTAUT and TAM models, the DeLone and McLean information systems success model has also been applied to the domain eLearning (Lwoga 2014; Cheng 2012; Wang et al. 2007; Lin 2007). Some scholars have further modified the information quality construct in the DeLone and McLean model to course quality in order to explain the influence of quality of the course design and content on the acceptance of LMS (Choi et al. 2007; Lee et al. 2009; Liu et al. 2010). Another quality factor that has been added to both TAM and the DeLone and McLean IS success model is the instructor quality. Several scholars have postulated that the quality of the instructor plays a significant role in influencing the students’ acceptance of an eLearning system (Lee et al. 2009; Cheng 2012; Lwoga 2014).

Within the context of developing countries, understanding the acceptance of e-learning technologies has been recognized to be complex and critical (Kim and Park 2018). Factors such as system quality, social influence, performance expectancy, effort expectancy, facilitating conditions, perceived ease of use, perceived usefulness, instructor quality, information quality, and learning value have constrained the widespread adoption and acceptance of LMS (Ain et al. 2016; Ayodele et al. 2016; Fathema et al. 2015; Yakubu and Dasuki 2018; Yakubu et al. 2019; Tarhini et al. 2013a; Lwoga 2014). In the sub-Saharan African context and Nigeria, where this study situates itself, there have been few studies that have examined the acceptance of LMS. Some have adopted the TAM model to highlight the influence of perceived ease of use and perceived usefulness on students’ behavioral intention to use LMS (Adewole-Odeshi 2014; Ogunbase 2014; Ayodele et al. 2016). Others have adopted the UTAUT to illustrate the influence of technology culturation, power, self-efficacy, and anxiety on students’ acceptance of LMS (Nicholas-Omoregbe et al. 2017; Olatunbosun et al. 2015).

Table 1 outlines selected studies that investigate students’ acceptance of eLearning systems in developing countries using the 3 research models described above.

Table 1 Previous studies on students’ acceptance of eLearning systems in developing countries
Full size table

To the best of our knowledge, none of the eLearning acceptance studies within the Nigerian context have examined the effect of instructor quality, learning value, and course quality on the acceptance of LMS in Nigeria. Hence, this study aims at addressing this gap by proposing a model that captures these significant constructs that have been shown to contribute to the acceptance of LMS by students.

3 Research model and hypotheses

TAM and UTAUT are among the popular theories used to investigate the propensity of technology acceptance among students. In the context of mobile/online banking, the theories overlooked critical factors such as “trust” and “security” (Aboelmaged and Gebba 2013; Aderonke 2010; Alalwan et al. 2015; Alsajjan and Dennis 2006). In the context of LMS, the present paper strives to bridge a similar gap by incorporating constructs such as learning value (Ain et al. 2016), instructor quality (Cheng 2012; Lwoga 2014), and course quality (Cheng 2012). The conceptual model is shown below in Fig. 1.

Fig. 1
figure 1

Research model

Full size image

3.1 Instructor quality (IQ)

According to Ozkan and Koseler (2009), the quality of the instructor dictates the learners’ attitudes towards an eLearning system; this statement is supported by Cheng who states that the instructor “is the key person that is important to learners’ behaviors in the e-learning process” (Cheng 2012, p. 369). Prior research has shown that learners’ acceptance of eLearning is influenced by the instructors’ attitude (Cheng 2012; Lwoga 2014; Ozkan and Koseler 2009). The instructor’s attitude towards the eLearning system is made up of the “instructor’s response timeliness, teaching style, and explanation/help towards learners via the eLearning system” (Cheng 2012, p. 369). This means that if an instructor can respond to students promptly and is adept at using the eLearning system to aid students to learn, then the students will be more inclined to accept the eLearning system. All 4 universities participating in this study have a considerable amount of experience in the use of LMS for teaching and learning, and as a result, we expect that the instructors will be skilled in using the LMS to teach and also, they will help the students learn how to use the LMS. Thus, the following proposition is derived:

  • H1: The influence of instructor quality on behavioral intention to use the LMS will be positive and significant.

3.2 Course quality (CQ)

An important measure of the quality of an information system is the output or information that can be obtained from the system in the form of reports (DeLone and McLean 1992, 2003). The quality of information obtained from an IS system is measured based on “dimensions such as accuracy, completeness, currency, efficiency, relevance, scope, and timeliness of information” (Cheng 2012, p. 365). In the context of eLearning, several authors (Lee et al. 2009; Liu et al. 2010; Cheng 2012) all agree that these dimensions (accuracy, completeness, currency, efficiency, relevance, scope, and timeliness of information), can also be used to measure the quality of a course, i.e., the content and design of a course. The content of a course includes the relevance, resource quality and how recent the resources are, while the design of the course ensures that the learners’ needs are met at different levels and facilitates access to the learning resources. If the course content and design meet the learners’ expectations, it is assumed that the learners are more likely to use the eLearning system. Once more, based on the experience in the use of LMS for instruction by the universities in this study, it is expected that the course quality would have been improved upon significantly over time so as to meet the learning needs of the students. Thus this study hypothesizes that:

  • H2: The influence of the course quality on behavioral intention to use the LMS will be positive and significant.

3.3 Learning value (LV)

We adopt the definition of learning value from Ain et al. (2016), which states that learning value is “the students’ positive perceptions about learning from the LMS influencing their intention to devote more time and effort to explore and obtain the required knowledge from the LMS” (Ain et al. 2016, p. 6). We also adopted the relationship between LV and the students’ behavioral intention to use the LMS as it was observed that learning value has a positive and significant influence on the students’ behavioral intention to use the LMS (Ain et al. 2016). Thus we propose that:

  • H3: The influence of learning value on behavioral intention to use the LMS will be positive and significant.

3.4 Social influence (SI)

Social influence is derived from the UTAUT framework and “is defined as the degree to which an individual perceives that important others believe he or she should use the new system” (Venkatesh et al. 2003, p. 451). SI is similar to the social norm construct derived from the TAM model. In the context of eLearning systems, such as an LMS, SI measures the influence of people considered to be important to the students on the students’ behavioral intention to use the LMS. The important people include the students’ classmates, instructors, and school administration. It is assumed that if the students are encouraged to use the LMS by their classmates, teachers, and the university administration, then they will be more likely to use the LMS. This study adopts the relationship between SI and behavioral intention from the UTAUT model as well as prior research (Raman et al. 2014; Ain et al. 2016). Therefore we propose that:

  • H4: The influence of social influence on behavioral intention to use the LMS will be positive and significant.

3.5 Facilitating conditions (FC)

Facilitating conditions are an individual’s collective perception that the organizational and technological resources required to use a particular technology are available (Venkatesh et al. 2003). Students using the LMS require infrastructures such as internet availability, access to devices, and technical support in using the system. If this infrastructure is not available, then the students will be less inclined to use the LMS. We adopt the relationship between FC and behavioral intention from the UTAUT model which has been corroborated by prior research (Jong 2009; Raman et al. 2014), where it was observed that FC significantly influences students’ behavioral intention to use LMS. Therefore we propose that:

  • H5: The influence of facilitating conditions on behavioral intention to use the LMS will be positive and significant.

3.6 System quality (SQ)

System quality was derived from the DeLone, and McLean IS success model (DeLone and McLean 1992; DeLone and McLean 2003). According to Jalote (2008), the quality of software (in this case, the LMS) can be measured by the following factors: usability, functionality, portability, reliability, maintainability, and efficiency. With regard to eLearning systems, the system quality construct measures the quality of the LMS, such as ease of use, functionality, reliability, and efficiency. Prior studies on the acceptance of eLearning systems have shown that system quality has a causal effect on behavioral intention to use an eLearning system (Ramayah et al. 2010; Mohammadi 2015). Therefore, if the LMS is easy to use, reliable, functional, and efficient, then the students are more likely to be encouraged to use the system. On this premise, we pose the following proposition:

  • H6: The influence of system quality on behavioral intention to use the LMS will be positive and significant.

3.7 Perceived usefulness (PU)

According to Davis (1989), perceived usefulness is defined as the “degree to which a person believes that using a particular system would enhance his/her job performance” (Davis 1989, p. 320). PU is derived from TAM and is similar to the performance expectancy construct in the UTAUT framework. In the context of LMS, it is the degree to which students believe that if they use the LMS, then their learning performance will be enhanced, hence resulting in better grades (Wang et al. 2009). In the context of LMS, prior studies have shown that PU positively and significantly influences students’ behavioral intention to use LMS (Lee et al. 2009; Venter et al. 2012; Cheng 2012). Students will be more inclined to accept and use the LMS if they believe that the LMS will aid in achieving their academic aspirations. In this study, we believe that PU will significantly influence the students’ behavioral intention to use the LMS as the students will find the LMS useful to their studies. Therefore, we propose that:

  • H7: The influence of perceived usefulness on behavioral intention to use the LMS will be positive and significant.

3.8 Perceived ease of use (PEOU)

Similar to PU, PEOU is also derived from TAM and is defined as “the degree to which a person believes that using a particular system would be free of physical and mental effort” (Davis 1989, p. 320). In the context of eLearning systems such as LMS, PEOU is the degree to which a student believes that the eLearning system will require minimal effort to learn how to use. The system should be intuitive and should not require specialized training before using the system. Cheng (2012) observed that PEOU has a positive and significant influence on behavioral intentions. Thus, this study hypothesizes that:

  • H8: The influence of perceived ease of use on behavioral intention to use LMS will be positive and significant.

3.9 Behavioral intentions (BI)

Behavioral intention is an individuals’ intention to use specific technology for several undertakings. In the context of this study, BI captures students’ behavioral intention to use the LMS to perform their learning activities. We propose that once the intention to use the LMS is formed, it would translate to the actual usage of the system. The relationship between BI and actual usage has been established in the following theories: UTAUT (Venkatesh et al. 2003), TAM (Davis 1989), the theory of reasoned action (TRA) (Ajzen and Fishbein 1980) and the theory of planned behavior (TPB) (Ajzen 1991). Studies based on these theories and in the context of eLearning systems have also supported this significant relationship (Tarhini et al. 2013b). Therefore, this study proposes that:

  • H9: The influence of behavioral intention on actual usage to use the LMS will be positive and significant.

4 Material and methods

Participants were drawn from two public and two private-owned universities in the Northern part of Nigeria by using a convenience sampling technique. This sampling approach has the capacity to facilitate data collection in a short duration of time and is the cheapest to implement. In collecting the data required for this study, two methods were used to administer the questionnaires. In the case of the public universities a paper-based survey was distributed to the students because the universities do not have a mailing list of their students. Access to the LMS is via a portal where the students log into the portal with a user ID and password. All communication is done while the students are logged into the portal. The distribution of the questionnaires was done by selected administrative staff of the universities who were trained in administering the surveys by the researcher. The training was done in order to meet the regulations stipulated by the research ethics board. The staff were informed of the purpose of the research, and they also ensured that the respondents read and kept a copy of the consent form. Collection of the completed questionnaires was via a ballot styled box to ensure anonymity. A total of 1000 questionnaires were distributed, and 738 usable responses werereceived.

An online survey was designed exactly like the paper-based version using the Survey Monkey online tool, and a link to the survey was emailed to the students of the two private universities via the student’s affairs office and the student government associations. The questionnaire was configured using the online tool to ensure confidentiality by ensuring that the emails and IP addresses of the responses were not captured. A total of 378 usable responses were received from the online questionnaire. Both sets of questionnaires were structured into three sections. Section 1, the consent form, introduced the study to the respondents stating the purpose of the study and stating what is required of the respondents from the study. It also mentioned the risks and benefits associated with participating in the study. The respondents were informed that participation was voluntary and that the responses would be confidential. Finally, the researcher’s contact details were listed in the event the respondents needed further clarification or questions about the study.

The second section of the questionnaire was used to capture demographic data about the respondents. The demographic data captured included the respondent’s age, their level of education (i.e., undergraduate or postgraduate), the length of time they have used the LMS and whether they were trained on using the LMS. Section 3 of the questionnaire employed a 5-point Likert scale to capture the responses of the students’ perceptions of the ten constructs used in the conceptual model. The scale ranged from 1 (strongly agree) to 5 (strongly disagree). Table 2 below shows the demographic information of the respondents. A total of 1116 responses were gathered from both sets of questionnaires.

Table 2 Demographic information of the respondents (n = 1116)
Full size table

5 Data analysis and results

5.1 Structural equation modeling versus artificial intelligence technique

According to Henseler et al. (2009), the statistical choice should be made based on 3 criteria. One, for theory or hypotheses testing, covariance-based structural equation modeling (CB-SEM) is a better option as it can measure the goodness-of-fit of the data with the model based on parametric statistical calculation. Two, for theory building or exploration with a relatively weak theoretical foundation, partial least square structural equation modeling (PLS-SEM) is a better option. Three, for relational prediction artificial intelligence technique (ANN), would be the better approach. Figure 2 depicts the overlap between the statistical methods. Based on the study’s aim, PLS-SEM and ANN seem to be better options.

Fig. 2
figure 2

CB-SEM (e.g. LISREL), VB-SEM (e.g. PLS) and ANN (Henseler et al. 2009)

Full size image

5.2 PLS-SEM analysis

SmartPLS (Version 3) software was utilized for data analysis, more specifically, PLS-SEM technique. The author(s) ran the measurement model and assessed the coefficients of retained items outer loadings, construct reliability, convergent, and divergent validity. Scale items with low factor loadings were discarded, the retained items outer loadings exceeded the threshold >.50 (See Fig. 3) and are statistically significant (see Fig. 4). The constructs Cronbach’s alpha and composite reliability (CRs) exceeded the threshold value of 0.7; the average variance extracted (AVE) exceeded the threshold value of 0.5 (Hair et al. 2010). Except for learning value, however, when AVE value is less than .50, Fornell and Larcker (1981) asserted that validity can still be inferred if the CR is above .70. Drawing on this evidence, we concluded that scale reliability and convergent validity for the constructs have been established. As for discriminant validity, HTMT-ratio and The Fornell–Larcker criterion was assessed; Heterotrait-monotrait (HTMT) ratio of correlation was acceptable below 1 (Henseler et al. 2015) and AVEs were higher than the squared inter-construct correlations (Fornell and Larcker 1981). See Table 3.

Fig. 3
figure 3

Model with factor loadings, beta coefficients, and R2 values

Full size image
Fig. 4
figure 4

Model with t-values and beta coefficients

Full size image
Table 3 Reliability, convergent and divergent validity
Full size table

Results from PLS-SEM analysis revealed that instructor quality, course quality and learning value did not exert significant impacts on behavioral intentions respectively (β = −.009, ρ = .741), (β = −.021, ρ = .564), (β = −.015, ρ = .674). Thus, H1, H2 and H3 did not receive empirical support. Social influence and facilitating conditions exerted significant impacts on behavioral intentions respectively (β = .265, ρ = .000), (β = .117, ρ = .000), which lend empirical support to H4 and H5. Contrary to the initial prediction system quality exerted a significant negative impact on behavioral intentions (β = −.056, ρ = .031). Thus, H6 did not receive empirical support. Perceived usefulness and perceived ease of use exerted significant impacts on behavioral intentions (β = .130, ρ = .000), (β = .358, ρ = .000). Finally, behavioral intentions exerted a significant impact on actual usage (β = .194, ρ = .000). Thus, H7, H8 and H9 gained empirical support. See Figs. 3 and 4 for further details.

5.3 Predictive analytics with ANN

PLS-SEM highlighted important links and significant associations between the variables. The associations between instructor quality, course quality, learning value, and behavioral intention were ignored due to lack of significance. Artificial neural networks (ANN) are information processing systems that comprise of information processing units (also known as “cells,” “neurons”). These units are divided into three layers, namely: input layer that accepts input data, the hidden layer where data are processed and the output layer for outcome generation. ANN has the capability to model complex interactions in comparison to traditional methods such as regression, CB-SEM and PLS-SEM (Hew et al. 2017), by detecting non-linear relationships that are hidden in case data and can further apply these relationships to a new dataset (de la Paz-Marín et al. 2012).

ANN is lenient in nature with no mandate for factor loadings, normality assumptions, linearity, homoscedasticity, and sample size (Abubakar 2018; de la Paz-Marín et al. 2012). ANN has high predictive accuracy and validity (Abubakar et al. 2017), fast learning and accurate predictions - ANN can acquire new learning and store this memory (Abubakar et al. 2019), fault tolerance - ANN can accommodate samples with variations (Li et al. 2015), robust against noisy or missing data (Göçken et al. 2016). These attractive features of ANN largely insulate the method from traditional statistical flaws and myths. A (3:2 hidden nodes) Multi-Layer Perceptron ANN model based on Resilient Backpropagation with Weight Backtracking algorithm was developed using neuralnet package in R studio. Logistic function was used as the activation function for both the hidden and the output layer, while Sum Squared Errors (SSE) was used as the differentiable error function.

The algorithm can minimize error until the ANN learns through the learning process. During the training process, random synaptic weights were assigned to the connections, and the aim is to adjust them to obtain the minimal error. 75% of the data was used for training, and 25% for testing. The synaptic weights of the input nodes on the hidden and output nodes are shown in Figs. 5 and 6.

First, the impact of social influence, facilitating conditions, system quality, perceived ease of use and perceived usefulness on behavioral intention was modeled (See Fig. 5). The Mean Square Error (MSE) for linear modeling is .27, and for cross-validation is .29, while the value for the neural network model seems better at .05. The training process needed 3484 steps until all absolute partial derivatives of the error function were smaller than 0.010.

Fig. 5
figure 5

Neural network model 1

Full size image
Fig. 6
figure 6

Neural network model 2

Full size image

Accordingly, Alice (2016) suggested that the distribution of the generalized weights is easy and useful in interpreting the nature of the effects. Figure 7 presents the distribution of the generalized weights. The estimated weights show that the predictor variables exerted non-linear effects on the response variables. The present outcome did not only provide support for PLS-SEM findings regarding the influence of social influence, facilitating conditions, perceived ease of use and perceived usefulness on behavioral intention, but also contradicts the findings in PLS-SEM which shows that system quality has a negative influence on behavioral intention. ANN shows that system quality can have both negative and positive impact on behavioral intention. Finally, the influence of behavioral intention on actual usage of LMS was confirmed.

Fig. 7
figure 7

Distributed weights of neural network 1 and 2

Full size image

To avoid over-fitting, a 10-fold cross-validation modeling with a ratio of 75:25 data for training and testing was used to examine the accuracy of the model. The MSE of the networks reported in Table 4 is closer to .000, which signals model reliability in terms of predictive validity. In sum, H4, H5, H6, H7, H8, and H9 received empirical support.

Table 4 Neural network 1 and 2 - MSEs
Full size table

6 Discussion

In this study, a total of nine hypotheses were tested out of which we found support for six of the relationships between the constructs. Figure 1 indicates that there are 8 independent variables which are hypothesized to influence the students’ behavioral intentions to use an LMS. 5 of these relationships were supported. The strongest influence on behavioral intentions was the perceived ease of using the LMS (PEOU). The statistically significant relationship between PEOU and BI supports similar findings, also in similar contexts (Mbengo 2014; Al-Azawei et al. 2017; Ibrahim et al. 2017). Nigerian students, therefore, attribute their intention to use the LMS mainly to the ease of using the system. Scholars have claimed PEOU, and similar constructs would be more salient for inexperienced users (Venkatesh et al. 2003). This corroborates our findings as most of the users claim to have used the LMS for under 1 year. Our findings on the relationship between PEOU and BI indicate that the LMS is easy to use.

The next most significant factor in influencing Nigerian students’ intention to use the LMS is the social influence on the students. The students believe that people they regard as important encourage them to use the LMS. Similar to the PEOU construct, SI has been shown to be more salient to inexperienced users (Venkatesh et al. 2003), which is in agreement with our findings where the higher percentage of the student sample used in this study have used the LMS for under 1 year. However, this relationship disagrees with a similar study by Yakubu and Dasuki (2018) where an insignificant association was observed between SI and BI, the authors attributed this to the mandatory use of LMS in the university by all instructors claiming that the usage of the LMS was not dependant on the referral by the student’s colleagues or instructors. The differing results could be due to the fact that 2 out of the 4 universities in this study allow instructors to choose if they want to use the LMS or not (i.e., voluntary use of the LMS). Also, in one of the universities in this study, there are 2 LMS for the instructors to choose from. Thus instructors will use the LMS they feel comfortable with and likewise encourage their students also to use the LMS.

Perceived usefulness was observed to positively predict the student’s behavioral intention to use the LMS. Thus, the students believe that the LMS is useful for their studies and this encourages them to use the LMS; this finding is consistent with prior research (Cheng 2012; Alharbi and Drew 2014). Functionalities such as submitting assignments and access to learning resources make the LMS a useful tool for achieving learning outcomes. The students also find the tracking features of the LMS beneficial as they can use it to know where and how they can improve their scores to attain their desired grades and also to be aware of modules that need to be completed or revised.

Our findings also show that the quality of the system has a positive and significant influence on students’ behavioral intention to use the LMS. The system quality captures the functionality, reliability, and efficiency of the LMS. Thus students attribute the use of the LMS to these features. If the system were unreliable, nonfunctional, or inefficient, students would be less inclined to use the LMS as a lot of effort will be required to use the system. This is in agreement with Ramayah et al. (2010) who states that a good system is essential to sustain the use of eLearning applications.

The final positive and significant relationship with BI was observed by facilitating conditions. This means that the students believe that the existence of organizational and technological infrastructure plays an important role in influencing the student’s intention to use the LMS. Facilities such the technical help desk support, internet access, and the availability of devices encourage the students to use the LMS. Our finding is supported by similar studies that examine the acceptance of eLearning systems by students (Raman et al. 2014; Salloum and Shaalan 2018).

An unexpected result was the insignificant relationship between learning value (LV) and students’ behavioral intention to use the LMS. Our result conflicts with the observed relationship in the study by Ain et al. (2016). Based on the measurement items for LV used in this study, and adapted from Ain et al. (2016), the students in our study do not believe that using the LMS enables them to decide their learning pace nor to increase their knowledge or control their success in the course. They also do not believe that the LMS facilitates the sharing of knowledge with others using the discussion forums. The insignificant relationship between LV and BI is possibly due to the fact that 2 of the universities in this study only use the LMS to host the learning resources for the courses. The LMS does not really offer a student-centered learning approach. Therefore, they do not really see the value of the LMS, especially for students who are used to using physical learning resources such as books.

Course quality (CQ), which should be an important factor that influences the students to use the LMS, was observed to have an insignificant relationship with the students’ behavioral intention to use the LMS, indicating that the students perceive that they do not have access to the updated resources, flexible and learner-centered courses (Cheng 2012). Similar to the insignificant relationship between LV and BI, this finding is attributed to the fact that the LMS is used as a content management system where learning resources placed on the LMS for students to access without screening for the appropriateness of the content.

The third insignificant relationship was observed between instructor quality (IQ) and BI. Cheng (2012) attributes this to the teaching style and timely response to the learners using the eLearning system. We believe this is based on the stagnated style of teaching witnessed in Nigerian higher education institutions. This is supported by claims made by Gengle et al. (2017) that teachers in Nigeria need to adopt different teaching strategies to improve students performance. Higher education institutions need to concentrate on training instructors on how to use LMS. This will enable the instructors to engage their students by using the LMS to provide timely responses to assist students (Cheng 2012; Lee 2010), and to provide an interactive learning atmosphere for students through the use of various multimedia resources and interactive teaching styles.

6.1 Implications for practice

The results of this study provide a few managerial implications for university administration, instructors, eLearning managers, and software developers. These implications will aid in motivating the students to use LMS in Nigerian higher education institutions. Firstly, the results show that software quality, perceived ease of use, and perceived usefulness of the LMS were significant in influencing the students’ behavioral intention to use the LMS. This would suggest that the eLearning system designers should take into consideration the functionality, the user interface as well as the students’ interactivity on the system when designing or updating LMS. Also, regular surveys should be carried out to improve on the current designs as the students learning needs will change over time. Instructors should be aware of, and learn to use, the majority of the functionality of the LMS and also be able to integrate some of these functionalities into their curriculum so as to motivate the students’ use of the LMS.

Secondly, the findings from this study indicate that both the instructor quality and course quality have an insignificant relationship with the students’ behavioral intention to use the LMS suggesting that the students do not agree that the instructors’ response via the LMS and attitude towards the LMS does not contribute to their usage of the system. The university administration, therefore, need to ensure that the instructors are trained properly on using the LMS. The training should be done by pedagogy experts who will ensure that the training emphasizes on ensuring that the instructors use the LMS to motivate the students to achieve the learning outcomes of the course. Furthermore, university administrators should encourage instructors to attend workshops and conferences centered on the use of eLearning systems. Finally, the quality assurance department should evaluate the instructors based on the use of the LMS and the quality of their courses hosted on the LMS.

Thirdly, we observed that facilitating conditions has a causal effect on the students’ behavioral intention to use the learning management system. This means that the universities have the organizational and technological infrastructure required to support the use of the LMS. University administrators, therefore, should ensure that devices such as laptops and desktops are made available for the use of the LMS by students who do not have access to internet-ready devices. Also, there should be a help desk support team in place, for example, an instructional technology unit, where training and support can be rendered to users of the LMS. The LMS developers can incorporate intuitive online help services and tutorials to support the students’ use of the LMS when the help desk team is not available. The fourth implication is based on the significant relationship between social influence and students’ behavioral intention, and the insignificant relationship of learning value on behavioral intention to use the LMS. The students attribute their use of the LMS to the influence of their peers, instructors, and the university management system. Universities need to adopt policies that enforce and monitor the use of LMS in a way that will encourage both students and instructors to see the value that can be gained by using the LMS.

6.2 Limitations and further research

The cross-sectional and self-reported nature of survey data restricts our ability to draw causal inferences. Future studies may utilize qualitative or mixed-method approaches for a deeper understanding. Although multiple approaches, e.g., assurance of confidentiality and anonymity of the respondents, and the use of artificial intelligence approaches were used to overcome the potential threats of common method bias, future studies are encouraged to use longitudinal design or multiple data sources to eradicate this tendency. The respondents are students, which limits our ability to generalize the findings. Future research may test and validate the model in different cultural settings or countries. This study did not consider perceived enjoyment, user satisfaction, and attitude towards technology; this may be a fruitful research avenue.

7 Conclusion

This study has examined the factors that have contributed to the acceptance of LMS in 4 universities in Nigeria. The study proposed a model to identify student’s acceptance of learning management systems. Survey data collected from 1116 students were analyzed using artificial neural network (ANN) and structural equation modeling (SEM) techniques. The findings of the study show that social influence, facilitating conditions, system quality, perceived ease of use, and perceived usefulness are important predictors for students’ behavioral intention to use LMS. Students’ behavioral intention to use LMS also functions as a predictor for actual usage of LMS. In sum, this study adds to our understanding of the factors that influence students to use the LMS, specifically in Nigeria. Also, the results will also help university administrators, governing bodies such as the National University Commission (NUC), Tertiary Education Trust Fund (TETFund) and software developers to plan, develop and implement LMS for higher education institutions in Nigeria. This will be a giant step towards improving the quality of instruction and learning in institutions of higher learning in Nigeria.