Abstract
Background:
Blended learning has become a key pedagogical approach for advancing Education 5.0 by integrating digital and face-to-face instructional practices. However, limited empirical evidence exists on the institutional, psychological, and innovation-related factors influencing preservice mathematics teachers' intention to adopt blended learning for teaching and assessment, particularly in developing countries.
Purpose:
This study examined the factors associated with preservice mathematics teachers’ intention to adopt blended learning for teaching and assessment in Education 5.0 by investigating the roles of institutional training, self-efficacy, attitude, and innovation attributes.
Method:
This cross-sectional survey involved 240 preservice mathematics teachers purposively and conveniently selected from public universities in Southwest Nigeria. A structured questionnaire with items adapted from validated instruments was used for data collection. The measurement and structural models were assessed using Partial Least Squares Structural Equation Modelling (PLS-SEM).
Results:
The measurement model demonstrated satisfactory reliability and validity. Structural model analysis revealed that institutional training was significantly and positively associated with blended learning for teaching (β = .496, p < .001) and for assessment (β = .322, p = .003), but showed no significant relationship with attitude (β = .054, p = .349) or self-efficacy (β = −.100, p = .260). Innovation attributes were significantly and positively associated with both attitude (β = .865, p < .001) and self-efficacy (β = .615, p < .001). Unexpectedly, self-efficacy showed a significant negative relationship with attitude (β = −.270, p = .001) but was significantly and positively associated with blended learning for teaching (β = .279, p = .001). Furthermore, attitude showed a significant positive relationship with blended learning for assessment (β = .270, p = .007) but not with blended learning for teaching (β = .085, p = .148). The model demonstrated moderate explanatory power and satisfactory predictive relevance. It explained 58.4% of the variance in attitude, 40.5% in blended learning for teaching, 34.3% in self-efficacy, and 22.4% in blended learning for assessment.
Conclusion:
The findings underscore the importance of strengthening institutional training and fostering favourable perceptions of innovation to promote preservice mathematics teachers' intention to adopt blended learning for teaching and assessment. Teacher education institutions should provide sustained, practice-oriented blended learning experiences that strengthen both instructional and assessment competencies required for Education 5.0.
1 Introduction
An educational paradigm change has occurred on a global scale as a result of the transition from the Fourth Industrial Revolution (Industry 4.0) to the Fifth Industrial Revolution (Industry 5.0), marking the beginning of a new age in education known as Education 5.0 (Chinchorkar and Jadhav, 2024; Permadi and Risnawati, 2025). Education 5.0 centres on a more humanistic, sustainable, and personalized educational approach, as opposed to Education 4.0, which is primarily digital transformation, automation, and efficiency-led, leveraging technologies such as IoT and Big Data (Ajitha and Vakkil, 2025; Falebita et al., 2025; Parhusip and Sitanggang, 2025; Singh and Kaur, 2024). This new era aims to balance technological innovation with human elements, highlighting the potential for collaboration between humans and autonomous machines to enhance the human condition and create a more sustainable and efficient society (Moola et al., 2024). Within this paradigm, educational technologies are expected not only to improve instructional efficiency but also to support learner-centred, collaborative, and authentic learning experiences that develop higher-order competencies. Consequently, the relevance of blended learning extends beyond technology integration; it represents a pedagogical approach that aligns with the human-centred philosophy of Education 5.0 by combining digital technologies with meaningful face-to-face interactions to promote personalised teaching, continuous assessment, and collaborative knowledge construction. In this regard, blended learning has become a key element of contemporary higher education, especially as a result of the sudden digitalization during the COVID-19 pandemic (Hamad et al., 2024; Tong et al., 2023).
The blended learning model, which combines face-to-face learning with technology-based elements, has become the “new normal” in the educational system (Mohammadi et al., 2025; Turnbull et al., 2021). Teacher education programmes increasingly encourage preservice teachers to develop competencies in blended learning because contemporary classrooms require graduates who can effectively integrate digital and face-to-face instructional approaches (Saputra et al., 2025; Tong et al., 2023). Research suggests that the blended learning approach is highly beneficial for flexibility, collaboration, and digital professional competence (Koloti et al., 2026; Tong et al., 2023). The implementation of this model is undoubtedly highly dependent on teachers' professional competence, confidence, and readiness to integrate technology into pedagogical practice in the era of Education 5.0 (Geng et al., 2019; Ma and Hao, 2024). The teaching and learning of mathematics present unique pedagogical challenges because many mathematical concepts are abstract, cumulative, and require multiple representations, dynamic visualisation, and continuous formative assessment. Consequently, technologies such as dynamic mathematics software, artificial intelligence (AI), and interactive digital platforms can support conceptual understanding, problem-solving, and immediate feedback when integrated appropriately (Falebita et al., 2026; Saputra et al., 2025; Syafitri et al., 2025). Although these technologies have great potential, there remains a gap between the dynamism of these technologies and the capability of mathematics teachers to apply them in their teaching (Syafitri et al., 2025). The use of traditional teaching methods, including lecture-based instruction and note-taking, by many teachers still does not enable the realization of the potential of Society 5.0 (Syafitri et al., 2025). Unlike many subject areas, mathematics instruction requires teachers to orchestrate conceptual explanations, procedural demonstrations, continuous practice, and formative assessment simultaneously. These distinctive pedagogical demands suggest that preservice mathematics teachers may experience blended learning differently from preservice teachers in general, thereby warranting discipline-specific investigation. Such a gap is especially worrisome for preservice teachers, as the gatekeepers of future STEM education, and they must be prepared to lead the digital transformation (Hidayat et al., 2023; Rind et al., 2024).
Understanding preservice mathematics teachers' intention to adopt blended learning requires an integrated theoretical perspective that considers both individual beliefs and perceived characteristics of the innovation. Social Cognitive Theory explains that self-efficacy influences individuals' confidence in performing a behaviour, whereas attitude reflects their overall evaluation of engaging in that behaviour. Attitude has been widely recognised in technology acceptance research as an important proximal determinant of behavioural intention and often develops from users’ beliefs regarding the usefulness and desirability of an innovation (Getenet et al., 2024; Hong et al., 2026; Pozas and Letzel, 2021; Scherer et al., 2017). An essential component of successfully overcoming obstacles in a blended learning setting is teachers' self-efficacy, defined as their confidence in their ability to perform a specific task (Birbal et al., 2018). Research has indicated that people with high self-efficacy tend to perceive challenges as tasks to be accomplished rather than obstacles to be overcome, thereby increasing their task engagement and persistence (Alkaabi et al., 2023; Yeop et al., 2018). Recent evidence also indicates that teachers whose self-efficacy beliefs align with positive attitudes towards blended learning are more likely to implement blended learning effectively in classroom practice (Hong et al., 2025, 2026). Complementing these psychological perspectives, the Diffusion of Innovation (DOI) theory proposes that the perceived characteristics of an innovation, particularly relative advantage, compatibility, trialability, and observability, influence individuals' evaluations of the innovation and subsequently their intention to adopt it. Accordingly, the present study integrates Social Cognitive Theory and DOI to explain how psychological factors and innovation attributes jointly shape preservice mathematics teachers' intention to adopt blended learning within Education 5.0.
DOI therefore provides the external explanatory mechanism, whereas self-efficacy and attitude represent the internal psychological mechanisms that together influence behavioural intention. In the case of preservice mathematics teachers, the innovation is not merely the technology itself, but the new model of Education 5.0 that places the human as the centre of the educational process (Weigand et al., 2022). Although some studies have examined blended learning through the DOI framework, very little has focused on preservice mathematics teachers’ intention to adopt blended learning in the socio-technological context of Education 5.0 (Koloti et al., 2026; Tong et al., 2023). Furthermore, previous studies have frequently conceptualised blended learning as a single behavioural outcome. However, teaching and assessment involve different pedagogical objectives, technological demands, and professional competencies. While blended learning for teaching primarily concerns instructional delivery and learner engagement, blended learning for assessment requires competence in digital assessment design, feedback provision, monitoring learner progress, and maintaining assessment integrity. Treating these behaviours as a single construct may therefore mask important differences in the factors associated with each dimension of blended learning practice. Firstly, although research on the technological aspects of Education 4.0 is extensive, there is a lack of empirical evidence that models the human-centric transition to Education 5.0 in teacher education programmes. Secondly, recent studies tend to focus on general education, overlooking the specific educational needs and supporting technologies in mathematics education. Thirdly, the interaction between self-efficacy, attitudes and innovation attributes has not been comprehensively modelled among preservice mathematics teachers. Finally, little empirical evidence distinguishes behavioural intention to use blended learning for teaching from behavioural intention to use blended learning for assessment, despite their differing pedagogical demands.
The study is important as it could provide a theoretical and practical framework for education stakeholders. This study is designed to model the relationships among self-efficacy, attitude, and innovation attributes in explaining preservice mathematics teachers' intention to adopt blended learning for teaching and assessment and to provide insights for redesigning teacher education programs in line with the demands of Society 5.0. Theoretically, the study extends the Diffusion of Innovation and Social Cognitive Theory perspectives by explaining blended learning intention within the human-centred philosophy of Education 5.0 while distinguishing between teaching and assessment practices. In practice, the results can guide curriculum designers and faculty managers to pay greater attention to technological support and professional development, thereby producing a new generation of mathematics teachers who are technologically competent yet humanistic in their teaching. Finally, the purpose of this study is to establish a connection between technological potential and classroom practice in mathematics learning, so that the learning process becomes more adaptive, interactive, and better oriented to the needs of society in the future. Drawing on the theoretical perspectives and empirical evidence reviewed above, the following hypotheses are formulated and empirically tested to examine the proposed relationships among the study constructs.
H1: Innovation Attributes positively influence Self-Efficacy toward blended learning.
H2: Innovation Attributes positively influence Attitude toward blended learning.
H3: Institutional Training positively influences Self-Efficacy.
H4: Institutional Training positively influences Attitude toward blended learning.
H5: Self-Efficacy positively influences Attitude toward blended learning.
H6: Attitude positively influences Blended Learning for Assessment.
H7: Self-Efficacy positively influences Blended Learning for Assessment.
H8: Institutional Training positively influences Blended Learning for Assessment.
H9: Attitude positively influences Blended Learning for Teaching.
H10: Self-Efficacy positively influences Blended Learning for Teaching.
H11: Institutional Training positively influences Blended Learning for Teaching.
2 Literature review and hypotheses development
2.1 Blended learning in mathematics teacher education and education 5.0
The transition from Education 4.0–5.0 represents a paradigm shift from technology-driven digitalisation towards a human-centric, sustainable, and personalized education system (Chinchorkar and Jadhav, 2024; Permadi and Risnawati, 2025). Education 5.0 emphasizes the synergy between human intelligence and cognitive computing for the well-being of society, whereas the Fourth Industrial Revolution emphasized automation and big data (Moola et al., 2024). Accordingly, educational technologies are no longer viewed simply as tools for improving instructional efficiency but as pedagogical enablers that foster collaboration, creativity, critical thinking, and personalised learning experiences. This shift in teacher education requires moving from being merely technologically literate to developing teachers who can use technology in ways that maintain the human elements of mentorship and social cohesion (Chinchorkar and Jadhav, 2024; Moola et al., 2024). In this context, blended learning provides an appropriate pedagogical model because it integrates digital technologies with face-to-face interaction while preserving meaningful teacher–student engagement. Rather than representing technology adoption alone, blended learning operationalises the human-centred philosophy of Education 5.0 by enabling flexible instruction, continuous feedback, collaborative learning, and personalised educational experiences. In mathematics education, blended learning serves dual pedagogical functions: firstly, it can help students understand complex problems through visualization with tools to explore mathematics concepts and problems; secondly, it can help teachers use assessment strategies that are more flexible, adaptive, and data-driven (Syafitri et al., 2025; Tong et al., 2023).
These dual functions distinguish mathematics education from many other disciplines. Mathematics teachers are expected not only to explain abstract concepts but also to diagnose misconceptions, monitor procedural development, provide continuous formative feedback, and promote mathematical reasoning. Consequently, blended learning for teaching and blended learning for assessment involve related but pedagogically distinct competencies. Whereas blended learning for teaching focuses primarily on instructional delivery and learner engagement, blended learning for assessment requires teachers to design digital assessments, interpret learner analytics, provide timely feedback, and maintain assessment quality. Despite these complementary roles, preservice teachers frequently perceive teaching and assessment as independent instructional activities rather than interconnected components of an integrated pedagogical cycle (Tong et al., 2023). Most previous studies have conceptualised blended learning adoption as a single behavioural construct, thereby overlooking these pedagogical distinctions. Examining blended learning for teaching and blended learning for assessment separately therefore provides a more comprehensive understanding of preservice mathematics teachers' behavioural intentions and better reflects the pedagogical demands of Education 5.0. Although it provides greater engagement and flexibility in learning, evidence remains limited regarding the factors associated with mathematics preservice teachers' intention to use blended learning for both teaching and assessment within the human-centred context of Education 5.0.
2.2 Theoretical foundations
This study adopts an integrated theoretical framework drawing primarily on Diffusion of Innovation (DOI) theory and Social Cognitive Theory (SCT), while recognising the role of attitude as an evaluative mechanism widely established in technology acceptance research. Together, these perspectives explain how external characteristics of blended learning interact with individual psychological beliefs and institutional support to shape preservice mathematics teachers' behavioural intention to use blended learning for teaching and assessment. According to DOI theory, five innovation characteristics influence an individual's decision to adopt an innovation: Relative Advantage, Compatibility, Complexity, Trialability, and Observability (Koloti et al., 2026; Rogers, 2003; Weigand et al., 2022). Although Rogers (2003) conceptualised these characteristics as distinct constructs, contemporary PLS-SEM studies increasingly model them as a higher-order construct when the objective is to capture users' overall perception of an innovation rather than examine the isolated contribution of each attribute. Such a specification provides greater theoretical parsimony, reduces model complexity, and reflects the reality that prospective users typically evaluate innovations holistically instead of independently considering each characteristic. Consistent with this perspective, the present study conceptualises Innovation Attributes as a reflective-formative higher-order construct comprising Relative Advantage, Compatibility, Trialability, and Observability. Complexity, although recognised as a core DOI attribute, was excluded from the higher-order construct. Previous studies have shown that perceived complexity frequently overlaps conceptually with technology self-efficacy because technologies perceived as difficult to use often reduce users' confidence in their ability to implement them. Moreover, contemporary blended learning platforms increasingly emphasise user-friendly interfaces, making Relative Advantage, Compatibility, Trialability, and Observability more salient determinants of behavioural intention within teacher education programmes.
The SCT highlights the importance of self-efficacy, defined as individuals' beliefs in their capability to organise and execute the actions required to achieve desired outcomes (Birbal et al., 2018). For preservice teachers, confidence in technology is an important mediator because their willingness to experiment with, integrate, and sustain blended learning depends largely on their perceived capability to use these technologies effectively (Birbal et al., 2018; Geng et al., 2019). Beyond self-efficacy, attitude represents teachers' overall positive or negative evaluation of blended learning and has consistently been identified as a proximal determinant of behavioural intention within technology acceptance research. Teachers who perceive blended learning as useful, beneficial, and compatible with their instructional goals are more likely to develop favourable attitudes that translate into stronger behavioural intentions. Institutional training complements these individual-level beliefs by providing organisational support, technical guidance, and opportunities for professional learning, thereby facilitating both competence development and positive perceptions of blended learning. Accordingly, DOI explains how innovation characteristics shape teachers' perceptions of blended learning; SCT explains how these perceptions influence self-efficacy; attitude represents teachers' evaluative response to the innovation; and institutional training provides the organisational support that facilitates technology integration. Together, these perspectives offer a comprehensive explanation of preservice mathematics teachers' intention to use blended learning for teaching and assessment within Education 5.0.
2.3 Innovation attributes, self-efficacy, and attitude toward blended learning
Perceived relative advantage and compatibility are important catalysts that can enhance teacher self-efficacy and create a positive professional attitude (Hamad et al., 2024; Jailani et al., 2025; Nuhoğlu et al., 2024). According to DOI theory, innovations perceived as advantageous, compatible, observable, and easy to experiment with reduce uncertainty and encourage users to develop favourable perceptions of the innovation. Such positive perceptions create opportunities for mastery experiences that strengthen self-efficacy while simultaneously fostering favourable attitudes toward technology integration (Henriksen et al., 2019; Quintana-Ordorika et al., 2023). Additionally, as preservice teachers have opportunities to experiment with blended learning tools and observe successful instructional outcomes, their confidence is reinforced through successful performance experiences and vicarious learning, two important sources of self-efficacy identified by SCT (Birbal et al., 2018; Koloti et al., 2026). Likewise, observing the educational benefits of blended learning encourages teachers to evaluate the innovation positively, thereby strengthening favourable attitudes towards its future use. This continuous interaction between innovation perceptions and psychological beliefs aligns with the human-centred philosophy of Education 5.0, where technology is valued not merely for efficiency but for its ability to enhance meaningful teaching and learning experiences (Chinchorkar and Jadhav, 2024).
2.4 Institutional training, self-efficacy, and attitude
Pedagogical transformation requires sustained institutional support and professional development. Higher education institutions that provide structured training programmes, technical assistance, mentoring opportunities, and access to blended learning technologies create environments that facilitate technology integration (Hamad et al., 2024). Such institutional support not only equips preservice teachers with technical competencies but also enhances their confidence to integrate blended learning into classroom practice. Empirical evidence suggests that systematic training strengthens teachers' self-efficacy by providing opportunities for guided practice, reducing technology-related anxiety, and increasing familiarity with digital instructional tools (Birbal et al., 2018; Koloti et al., 2026). Furthermore, institutional commitment to digital innovation signals organisational readiness for blended learning, encouraging teachers to perceive technology positively and develop favourable attitudes toward its use (Erwin and Kuswandi, 2024). The connection among support, self-efficacy, and attitude is a cycle: initial competence leads to increased self-efficacy, resulting in a positive and proactive attitude toward blended learning, which stabilizes competence (Hamad et al., 2024). Consequently, preservice teachers who perceive strong institutional support are more likely to feel competent and develop positive dispositions towards blended learning implementation.
2.5 Self-efficacy and attitude toward blended learning
Social Cognitive Theory posits that individuals' beliefs about their capabilities influence not only their behaviour but also their emotional responses and evaluations of tasks (Bandura, 1986, 2023). Individuals with stronger self-efficacy generally expect successful performance, experience lower levels of anxiety, and perceive challenging tasks as opportunities for growth rather than obstacles. Consequently, they are more likely to develop favourable attitudes toward engaging in those activities. Within technology-enhanced learning environments, teachers who are confident in their ability to use digital technologies tend to perceive technology integration more positively because they believe they can successfully overcome implementation challenges. Recent studies have similarly reported positive associations between teachers' technology self-efficacy and favourable attitudes towards blended learning practice (Hong et al., 2025, 2026; Scherer et al., 2017). Accordingly, higher levels of self-efficacy are expected to promote more positive attitudes toward blended learning among preservice mathematics teachers.
2.6 Determinants of blended learning for teaching and assessment
The implementation of blended learning in mathematics education can be conceptualised through two complementary behavioural intentions: blended learning for teaching (BLT) and blended learning for assessment (BLA). Although both dimensions reflect preservice teachers' intention to integrate blended learning into practice, they require different pedagogical competencies and technology applications. Blended learning for teaching primarily concerns instructional planning, content delivery, learner engagement, and classroom interaction, whereas blended learning for assessment involves designing digital assessments, providing timely feedback, monitoring learner progress, and using assessment information to support instructional decision-making. Teachers with favourable attitudes towards blended learning are more likely to appreciate the pedagogical benefits of integrating digital technologies into both instructional delivery and assessment practices. Similarly, teachers possessing higher self-efficacy are more willing to experiment with innovative instructional approaches and digital assessment tools because they believe they can effectively manage the associated technological and pedagogical demands (Tong et al., 2023). Institutional training further facilitates these behavioural intentions by providing opportunities for skill development, guided practice, and exposure to effective blended learning strategies (Hamad et al., 2024). Consistent with the integrated DOI–SCT framework, attitude represents teachers' evaluative disposition, self-efficacy reflects their perceived capability, and institutional training provides the organisational support necessary for translating positive perceptions into behavioural intention. Collectively, these factors are expected to explain preservice mathematics teachers' intention to use blended learning for both instructional delivery and assessment within the Education 5.0 environment. Therefore, this study proposes that:
2.7 Proposed research model
Figure 1 presents the proposed conceptual framework underpinning this study. The framework integrates the Diffusion of Innovation (DOI) theory and Social Cognitive Theory (SCT) to explain preservice mathematics teachers' behavioural intention to use blended learning for teaching and assessment within the context of Education 5.0. Innovation Attributes are conceptualised as a reflective-formative higher-order construct comprising Relative Advantage (RA), Compatibility (Comp), Trialability (TRI), and Observability (OBS). Institutional Training (IT) represents the organisational support provided by teacher education institutions, while Self-Efficacy (SE) and Attitude (ATT) represent the internal psychological mechanisms through which teachers evaluate and respond to blended learning. The model proposes that innovation attributes positively influence self-efficacy and attitude, while institutional training positively influences both self-efficacy and attitude. Self-efficacy is further expected to strengthen attitude toward blended learning. Subsequently, attitude, self-efficacy, and institutional training are hypothesised to influence two complementary behavioural intentions: Blended Learning for Teaching (BLT) and Blended Learning for Assessment (BLA). These two constructs capture distinct yet related pedagogical intentions associated with instructional delivery and assessment practices. Although Blended Learning for Teaching and Assessment (BLTA) is included in the measurement model as a reflective-formative higher-order construct comprising BLT and BLA, it is not treated as an endogenous construct within the structural model. Rather, BLTA provides an overall representation of preservice teachers' behavioural intention to integrate blended learning across teaching and assessment practices. Consequently, no structural hypotheses are specified between BLT, BLA, and BLTA, nor are direct structural paths estimated from the antecedent constructs to BLTA.
Figure 1
3 Methodology
3.1 Research design
This study adopted a quantitative cross-sectional survey research design within the post-positivist paradigm. The study, grounded in a post-positivist research paradigm, was underpinned by the assumption that relationships among latent constructs could be objectively examined using statistical modelling techniques (Abu-Alhaija, 2019). The cross-sectional survey design enabled data collection from respondents at a single point in time, facilitating evaluation of the conceptual model and the hypothesized relationships among the study variables (Creswell, 2014). However, because the study used a cross-sectional design, the structural relationships are interpreted as statistical associations rather than evidence of temporal prediction or causal effects.
3.2 Sample and sampling technique
This study was conducted on a sample of 240 preservice mathematics teachers selected from selected public universities in Southwest Nigeria. The sample size was considered adequate for the PLS-SEM model based on the complexity of the proposed structural model and the minimum sample requirement for estimating the specified relationships. The most complex endogenous construct in the structural model had three direct predictors, namely Attitude, Self-Efficacy, and Institutional Training. The achieved sample of 240 therefore provided substantially more observations than the minimum requirement for estimating the model (Hair et al., 2017a). Preservice mathematics teachers were selected because the study specifically examined their intention to use blended learning for future mathematics teaching and assessment. They represent an important population for examining technology-related beliefs and intentions before these teachers enter sustained professional practice. Two non-probability sampling techniques were employed. First, public universities offering Mathematics Education programmes and having exposure to blended learning programmes, digital learning platforms, or technology-supported teaching were purposively identified as eligible institutions. The institutional eligibility criteria were therefore based on the availability of a Mathematics Education programme and evidence of access to technology-supported or blended learning environments. Second, convenience sampling was used to recruit preservice mathematics teachers who were accessible and willing to participate during the data-collection period. A total of 13 public universities met the eligibility criteria, of which seven participated in the study. Approximately 279 eligible preservice mathematics teachers were approached, and 240 usable responses were obtained, representing a response rate of 86.02%. The use of non-probability sampling limits the generalisability of the findings beyond the participating institutions and respondents; therefore, the findings should be interpreted within the context of preservice mathematics teacher education in Southwest Nigeria.
Table 1 presents the demographic information of the mathematics preservice teachers who participated in the study. It indicates that the respondents were 101 (42.1%) male and 139 (57.9%) female preservice mathematics teachers. Regarding age, 151 (62.9%) were between 20 and 24 years, 41 (17.1%) were under 20 years, and 48 (20.0%) were over 24 years. Regarding teaching practice experience, 130 (54.2%) of the respondents were undertaking their first teaching practice, and 110 (45.8%) were undertaking their second. The varied characteristics of the respondents (gender, age, teaching practice experience) also ensured a well-balanced and representative sample for analyzing preservice mathematics teachers' intention to use blended learning in mathematics teaching and assessment.
Table 1
| Variable | Level | N | % |
|---|---|---|---|
| Gender | Male | 101 | 42.1 |
| Female | 139 | 57.9 | |
| Age | Less than 20 years | 41 | 17.1 |
| 20–24 years | 151 | 62.9 | |
| Above 24 years | 48 | 20.0 | |
| Teaching practice experience | First | 130 | 54.2 |
| Second | 110 | 45.8 | |
| Total | 240 | 100 | |
Demographic information of respondents.
N= number of respondents; %= Percentage.
3.3 Instrument for data collection
A structured questionnaire, adapted from validated instruments in technology acceptance, blended learning, educational innovation, and digital competence, was used to collect data. Rather than adopting the original items without modification, the adaptation followed a contextualisation procedure involving item identification, wording modification, expert review, and pilot testing. First, relevant items were identified from established instruments that measured constructs conceptually consistent with the proposed model. Second, the wording of the items was modified to reflect the specific context of preservice mathematics teachers and their intended use of blended learning for teaching and assessment. Third, the adapted items were reviewed by experts in mathematics education, educational technology, educational psychology, and measurement and evaluation to assess their relevance, clarity, representativeness, and contextual appropriateness. Based on their recommendations, ambiguous, repetitive, or contextually unsuitable wording was revised before pilot administration. The questionnaire has two sections. The first section (Section A) collected respondents' demographic information such as gender, age, and experience in teaching practice. In the second section (Section B), there were 45 items representing the nine latent constructs in the proposed conceptual model. These latent constructs are Relative Advantage (RA), Compatibility (COMP), Trialability (TRI), Observability (OBS). Institutional Training (IT), Self-Efficacy (SE), Attitude (ATT), Blended Learning for Teaching (BLT) and Blended Learning for Assessment (BLA). Each item was rated on a 5-point Likert scale, with extremes at 1 (Strongly Disagree) and 5 (Strongly Agree). The rationale for using a 5-point scale was to ensure measurement sensitivity, response variability and parameter estimation in structural equation modelling. Blended Learning for Teaching (BLT) and Blended Learning for Assessment (BLA) are the dependent constructs with five items each. The BLT and BLA items were adapted from Venkatesh et al. (2003) and Ye et al. (2022). The items were contextualized to assess preservice mathematics teachers' intentions to use blended learning in their teaching planning, teaching process, and students' engagement, and to measure respondents' intentions to use blended learning technologies for student assessment, student feedback, and student evaluation. Self-Efficacy (SE) and Attitude (ATT) were the psychological constructs involved. Five items from the ICT Self-Efficacy Scale (Ye et al., 2022) were adapted to measure self-efficacy. The items were adapted to assess respondents' confidence in using blended learning technologies for teaching and assessment. The items used to measure attitude were based on the Technology Acceptance Model (TAM) (Davis, 1989). The items were reworded to test participants' attitudes towards the use of blended learning in mathematics. Five items adapted from Butt et al. (2022) were used as measures of institutional training (IT). The items assessed perceptions of institutional support, the availability of training programmes and workshops, and encouragement to implement BL. Innovation attributes were adapted from the perceived characteristics framework developed by Rogers (2003) and Alhammadi et al. (2023). Specifically, the Compatibility (COMP) was measured using five items assessing how closely blended learning aligned with respondents' teaching styles, teaching practices, and professional needs. Relative Advantage (RA) included five items that assessed how much teachers perceived blended learning to be a better alternative than traditional approaches to teaching in terms of enhancing instructional quality, engagement, learning outcomes, and instructional effectiveness. Trialability (TRI) items focused on respondents' opportunities to experiment with the blended learning model before its implementation was complete, while Observability (OBS) items examined the visibility of successful BL practices and their results for peers and teachers. The four innovation characteristics were subsequently modelled as dimensions of the higher-order Innovation Attributes construct. Complexity, although recognised as a core DOI characteristic, was not included in the empirical measurement model because the study focused on the innovation characteristics most directly representing perceived benefits, compatibility, opportunities for experimentation, and visible outcomes of blended learning. The questionnaire was subjected to content and face validation by experts in mathematics education, educational technology, measurement and evaluation, and educational psychology. The experts independently examined the items for construct relevance, clarity, wording, redundancy, and alignment with the intended constructs. Their comments were incorporated before the pilot administration. Prior to the main data collection, the adapted questionnaire was pilot-tested with 40 preservice mathematics teachers who were comparable to the target population but were not included in the final sample. The pilot test was conducted to determine whether the items were understandable, contextually appropriate, and sufficiently reliable for the main study. Based on the pilot feedback and reliability results, necessary modifications were made to the wording and presentation of the instrument before the final administration.
3.4 Method of data collection
The study used a structured questionnaire to collect data from preservice mathematics teachers from selected public universities in Southwest Nigeria. Ethical clearance and permission were sought from the appropriate authorities of the respective institutions before the data collection. The study was explained to the respondents, and informed consent and voluntary participation were obtained. To ensure easy access and high response rates, the questionnaire was distributed both electronically and in paper form. The electronic questionnaire was disseminated via online platforms associated with the respondents' classes, and printed copies were distributed during classes and teaching practice meetings. Respondents had enough time to complete the questionnaire. Participants were told that their information would only be used for the study, and they were guaranteed anonymity and confidentiality. The completed responses were subsequently screened, coded and prepared for statistical analysis. After collection, all returned questionnaires were subjected to systematic data screening before statistical analysis. Responses were checked for completeness, and questionnaires with substantial missing responses were excluded from the analysis. Duplicate electronic submissions were identified and removed where applicable. Missing values in retained responses were examined to determine their extent and pattern. Response patterns were also inspected for evidence of straight-lining or invariant responding, particularly where respondents selected the same response category across a substantial proportion of items. The dataset was further examined for unusual response patterns and extreme observations before the final dataset was prepared for PLS-SEM analysis. Following this screening process, 240 usable responses were retained for the analysis.
3.5 Data analysis
The data analysis was conducted in three sequential stages. Demographic characteristics of the respondents were summarized using descriptive statistics, including frequencies and percentages. Second, in accordance with the Partial Least Squares Structural Equation Modelling (PLS-SEM) analysis, the measurement model was validated for establishing reliability and validity of the lower order constructs (LOCs): Institutional Training (IT), Self-Efficacy (SE), Attitude (ATT), Relative Advantage (RA), Compatibility (COMP), Trialability (TRI), Observability (OBS), Blended Learning for Teaching (BLT), and Blended Learning for Assessment (BLA). Outer loadings, Cronbach's alpha (CA), and composite reliability (CR) were used to assess indicator reliability and internal consistency (CA and CR), respectively. Convergent validity was checked by Average Variance Extracted (AVE), and discriminant validity was checked by Heterotrait-Monotrait ratio (HTMT) and Fornell-Larcker criterion. The variance inflation factor (VIF) was used to assess multicollinearity among indicators. After establishing the adequacy of the lower-order measurement models, the two higher-order constructs, Innovation Attributes and Blended Learning for Teaching and Assessment (BLTA), were evaluated. Innovation Attributes was specified as a formative higher-order construct comprising Relative Advantage, Compatibility, Trialability, and Observability. BLTA represented the broader measurement construct comprising Blended Learning for Teaching and Blended Learning for Assessment. The higher-order constructs were assessed separately from the substantive structural relationships to avoid treating BLT and BLA as structural predictors of their own higher-order construct. The structural model was evaluated using collinearity assessment, path coefficients (β), coefficients of determination (R2), and statistical significance testing. Collinearity was assessed using VIF values, while the strength and direction of hypothesised relationships were examined using standardised path coefficients. The explanatory power of the endogenous constructs was assessed using R2 values. Effect size (f2) was not included in the reported structural model assessment because the corresponding results were not reported separately for the hypothesised paths. Predictive assessment was conducted using the PLSpredict procedure. The reported Q2predict values, together with the Root Mean Square Error (RMSE) and Mean Absolute Error (MAE), were used to evaluate the model's out-of-sample predictive performance. These statistics were interpreted as predictive assessment indicators rather than as Stone-Geisser Q2 values obtained through the traditional blindfolding procedure. Thus, the term Q2predict is used throughout the manuscript to distinguish the PLSpredict procedure from blindfolding-based predictive relevance. Finally, bootstrapping with 5,000 resamples was employed to determine the significance of the hypothesised structural relationships at the 0.05 significance level. The path coefficients, standard errors, t-values, and p-values generated through bootstrapping were used to determine whether each hypothesised relationship was statistically significant. The descriptive analyses were conducted using IBM SPSS, while the measurement model, higher-order construct assessment, structural model, and PLSpredict analyses were performed using SmartPLS 4.
4 Results
4.1 Measurement model assessment
4.1.1 Assessment of lower-order constructs
Table 2 presents the results of the reliability and convergent validity assessment for the lower-order constructs. Indicator loadings were examined alongside internal consistency and convergent validity rather than relying solely on loading thresholds. Indicators with very low loadings were considered for removal only where their deletion improved measurement quality without compromising the conceptual coverage of the construct. The following indicators were removed during measurement model refinement: ATT2 (0.530), BLA1 (0.345), BLT1 (0.247), COMP1 (0.475), RA1 (0.544), RA5 (0.585), SE2 (0.431), SE3 (0.517), TRI2 (0.507), and OBS5 (0.307). These items recorded comparatively weak outer loadings and were reviewed in relation to construct reliability, convergent validity, and conceptual coverage. Their removal was retained only where the remaining indicators continued to adequately represent the underlying construct. In particular, the removal of SE2 and SE3 was considered carefully because self-efficacy is represented by only three retained indicators (SE1, SE4, and SE5). Before the removal, the CA and AVE were 0.677 and 0.442, respectively. The retained items continue to capture respondents' confidence in using blended-learning technologies for instructional and assessment purposes. The retained indicators had outer loadings ranging from 0.562 to 0.925 and were statistically significant (p < .001).
Table 2
| Construct | Item | Outer loadings | P-values |
|---|---|---|---|
| Attitude (ATT) | ATT1 | 0.657 | 0.000 |
| ATT3 | 0.794 | 0.000 | |
| ATT4 | 0.748 | 0.000 | |
| ATT5 | 0.631 | 0.000 | |
| Blended Learning for Assessment (BLA) | BLA2 | 0.721 | 0.000 |
| BLA3 | 0.801 | 0.000 | |
| BLA4 | 0.814 | 0.000 | |
| BLA5 | 0.858 | 0.000 | |
| Blended Learning for Teaching (BLT) | BLT2 | 0.685 | 0.000 |
| BLT3 | 0.844 | 0.000 | |
| BLT4 | 0.758 | 0.000 | |
| BLT5 | 0.799 | 0.000 | |
| Compatibility (COMP) | COMP2 | 0.617 | 0.000 |
| COMP3 | 0.877 | 0.000 | |
| COMP4 | 0.780 | 0.000 | |
| COMP5 | 0.925 | 0.000 | |
| Institutional Training (IT) | IT1 | 0.562 | 0.000 |
| IT2 | 0.838 | 0.000 | |
| IT3 | 0.833 | 0.000 | |
| IT4 | 0.742 | 0.000 | |
| IT5 | 0.773 | 0.000 | |
| Observability (OBS) | OBS1 | 0.862 | 0.000 |
| OBS2 | 0.891 | 0.000 | |
| OBS3 | 0.911 | 0.000 | |
| OBS4 | 0.813 | 0.000 | |
| Relative Advantage (RA) | RA2 | 0.798 | 0.000 |
| RA3 | 0.807 | 0.000 | |
| RA4 | 0.868 | 0.000 | |
| Self-Efficacy (SE) | SE1 | 0.799 | 0.000 |
| SE4 | 0.832 | 0.000 | |
| SE5 | 0.799 | 0.000 | |
| Trialability (TRI) | TRI1 | 0.771 | 0.000 |
| TRI3 | 0.714 | 0.000 | |
| TRI4 | 0.834 | 0.000 | |
| TRI5 | 0.827 | 0.000 |
Outer loading of lower-order constructs.
Table 3 presents the reliability and convergent validity of the lower-order constructs. Cronbach's alpha values ranged from 0.671 to 0.893, with the lowest value occurring for Attitude (α = 0.671), which is marginally below the conventional 0.70 criterion but was retained because the corresponding composite reliability (rho_c) and AVE indicated acceptable measurement quality. Composite reliabilities, rho_a, ranged from 0.703 to 0.912, and rho_c from 0.801 to 0.926, while AVE values ranged from 0.504 to 0.757. Thus, the constructs generally demonstrated satisfactory internal consistency and convergent validity.
Table 3
| Cronbach's Alpha | CR (rho_a) | CR (rho_c) | AVE | |
|---|---|---|---|---|
| Attitude (ATT) | 0.671 | 0.703 | 0.801 | 0.504 |
| Blended Learning for Assessment (BLA) | 0.812 | 0.826 | 0.877 | 0.641 |
| Blended Learning for Teaching (BLT) | 0.776 | 0.785 | 0.856 | 0.600 |
| Compatibility (COMP) | 0.814 | 0.832 | 0.881 | 0.654 |
| Institutional Training (IT) | 0.814 | 0.847 | 0.868 | 0.572 |
| Observability (OBS) | 0.893 | 0.912 | 0.926 | 0.757 |
| Relative Advantage (RA) | 0.767 | 0.792 | 0.864 | 0.680 |
| Self-Efficacy (SE) | 0.739 | 0.741 | 0.852 | 0.657 |
| Trialability (TRI) | 0.805 | 0.821 | 0.871 | 0.628 |
Reliability and convergent validity of lower-order constructs.
Table 4 presents the results of the discriminant validity assessment using the Heterotrait–Monotrait ratio (HTMT) and the Fornell–Larcker criterion. The HTMT values ranged from 0.162 to 0.935. With the exception of the relationship between Attitude (ATT) and Relative Advantage (RA) (HTMT = 0.935), all HTMT values were below the conservative threshold of 0.90, indicating adequate discriminant validity among the constructs (Hair et al., 2021; Henseler et al., 2015). Although the HTMT value between ATT and RA slightly exceeded the recommended threshold, this may be attributed to the conceptual relatedness of the two constructs, as respondents who perceive greater advantages in blended learning are also more likely to exhibit favourable attitudes toward its adoption. The Fornell–Larcker criterion further confirmed discriminant validity, as the square root of the Average Variance Extracted (AVE) for each construct (diagonal elements) exceeded its correlations with all other constructs. This indicates that each construct shared more variance with its own indicators than with other constructs in the model (Fornell and Larcker, 1981). Overall, the results provide satisfactory evidence that the lower-order constructs are empirically distinct and suitable for subsequent structural model analysis.
Table 4
| Heterotrait-monotrait ratio (HTMT) - Matrix | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| ATT | BLA | BLT | COMP | IT | OBS | RA | SE | TRI | |
| ATT | |||||||||
| BLA | 0.573 | ||||||||
| BLT | 0.447 | 0.703 | |||||||
| COMP | 0.809 | 0.197 | 0.483 | ||||||
| IT | 0.555 | 0.469 | 0.643 | 0.567 | |||||
| OBS | 0.878 | 0.162 | 0.308 | 0.669 | 0.362 | ||||
| RA | 0.935 | 0.704 | 0.317 | 0.599 | 0.261 | 0.563 | |||
| SE | 0.466 | 0.166 | 0.465 | 0.587 | 0.303 | 0.684 | 0.318 | ||
| TRI | 0.578 | 0.246 | 0.360 | 0.782 | 0.544 | 0.723 | 0.392 | 0.654 | |
| Fornell-Larcker criterion | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| ATT | BLA | BLT | COMP | IT | OBS | RA | SE | TRI | |
| ATT | 0.710 | ||||||||
| BLA | 0.377 | 0.800 | |||||||
| BLT | 0.304 | 0.567 | 0.774 | ||||||
| COMP | 0.586 | 0.152 | 0.403 | 0.808 | |||||
| IT | 0.315 | 0.400 | 0.559 | 0.455 | 0.756 | ||||
| OBS | 0.652 | −0.050 | 0.182 | 0.596 | 0.180 | 0.870 | |||
| RA | 0.685 | 0.571 | 0.221 | 0.482 | 0.167 | 0.483 | 0.825 | ||
| SE | 0.220 | 0.001 | 0.360 | 0.477 | 0.126 | 0.575 | 0.238 | 0.810 | |
| TRI | 0.398 | 0.056 | 0.240 | 0.662 | 0.365 | 0.647 | 0.337 | 0.555 | 0.788 |
Discriminant validity.
4.1.2 Assessment of higher-order construct
Table 5 presents the assessment of the reflective-formative higher-order constructs, namely Innovation Attributes and Blended Learning for Teaching and Assessment (BLTA). The evaluation focused on the significance and relevance of the formative weights, as well as collinearity among the lower-order constructs. Formative higher-order constructs are considered satisfactory when the indicator weights are statistically significant, and multicollinearity is within acceptable limits (VIF < 3.3 or, more conservatively, < 5.0). For the Innovation Attributes higher-order construct, all four lower-order constructs contributed significantly to its formation, with weights ranging from 0.272 to 0.341 (t = 18.720–29.872, p < .001). Among the dimensions, Observability (β = 0.341) made the largest contribution, followed by Compatibility (β = 0.323), Trialability (β = 0.296), and Relative Advantage (β = 0.272). The VIF values ranged from 1.431 to 2.104, indicating no multicollinearity among the formative dimensions. Similarly, the higher-order construct Blended Learning for Teaching and Assessment (BLTA) was significantly formed by its two lower-order dimensions: Blended Learning for Assessment (β = 0.565, t = 37.654, p < .001) and Blended Learning for Teaching (β = 0.565, t = 38.724, p < .001). Both dimensions contributed equally to the higher-order construct, suggesting that preservice teachers' adoption of blended learning encompasses both instructional delivery and assessment practices. The VIF values of 1.740 and 2.268, respectively, were well below the recommended threshold, confirming that multicollinearity was not a concern. Generally, the results indicate that all lower-order constructs significantly and uniquely contributed to their respective higher-order constructs. The significant formative weights and acceptable VIF values provide evidence of the validity of the higher-order measurement model, supporting its use in the subsequent structural model analysis.
Table 5
| Higher-order construct | Lower-order construct | Weight | t-value | p-value | VIF |
|---|---|---|---|---|---|
| Innovation attributes | Relative advantage | 0.272 | 18.720 | 0.000 | 1.431 |
| Compatibility | 0.323 | 29.872 | 0.000 | 2.077 | |
| Trialability | 0.296 | 25.706 | 0.000 | 2.104 | |
| Observability | 0.341 | 20.904 | 0.000 | 2.060 | |
| Blended Learning for Teaching and Assessment (BLTA) | Blended learning for assessment | 0.565 | 37.654 | 0.000 | 1.740 |
| Blended learning for teaching | 0.565 | 38.724 | 0.000 | 2.268 |
Assessment of higher-order construct.
4.2 Structural model results
Table 6 presents the results of the structural model analysis, including the standardized path coefficients (β), variance inflation factor (VIF), t-statistics, p-values, and effect sizes (f2) for the hypothesized relationships. The VIF values ranged from 1.062 to 1.727, which are below the recommended threshold of 3.3, indicating that multicollinearity was not a concern in the structural model (Hair et al., 2017b). The results showed that Institutional Training (IT) was significantly and positively associated with Blended Learning for Assessment (BLA) (β = .322, t = 2.970, p = .003, f2 = .119) and Blended Learning for Teaching (BLT) (β = .496, t = 5.797, p < .001, f2 = .366). However, IT was not significantly associated with Attitude (ATT) (β = .054, t = 0.937, p = .349, f2 = .006) or Self-Efficacy (SE) (β = −.100, t = 1.127, p = .260, f2 = .013). Innovation Attributes showed strong positive associations with both ATT (β = .865, t = 17.142, p < .001, f2 = 1.043) and SE (β = .615, t = 9.988, p < .001, f2 = .500). Self-Efficacy was significantly and negatively associated with ATT (β = −.270, t = 3.416, p = .001, f2 = .115). However, SE was not significantly associated with BLA (β = −.104, t = 1.450, p = .147, f2 = .013), but was significantly and positively associated with BLT (β = .279, t = 3.457, p = .001, f2 = .123). Similarly, ATT was significantly and positively associated with BLA (β = .270, t = 2.685, p = .007, f2 = .079), but its association with BLT was not statistically significant (β = .085, t = 1.447, p = .148, f2 = .010). Overall, the findings indicate that Innovation Attributes showed the strongest associations with the psychological constructs of Attitude and Self-Efficacy, whereas Institutional Training showed significant positive associations with both blended teaching and assessment. Self-Efficacy was positively associated with blended teaching but not blended assessment, while Attitude was positively associated with blended assessment but not blended teaching.
Table 6
| Path | Path coefficients | VIF | STDEV | T stat. | P-values | f2 |
|---|---|---|---|---|---|---|
| ATT → BLA | 0.270 | 1.178 | 0.100 | 2.685 | 0.007 | 0.079 |
| ATT → BLT | 0.085 | 1.178 | 0.059 | 1.447 | 0.148 | 0.010 |
| IT → ATT | 0.054 | 1.167 | 0.057 | 0.937 | 0.349 | 0.006 |
| IT → BLA | 0.322 | 1.128 | 0.109 | 2.970 | 0.003 | 0.119 |
| IT → BLT | 0.496 | 1.128 | 0.086 | 5.797 | 0.000 | 0.366 |
| IT → SE | −0.100 | 1.151 | 0.089 | 1.127 | 0.260 | 0.013 |
| Innovation attributes → ATT | 0.865 | 1.727 | 0.050 | 17.142 | 0.000 | 1.043 |
| Innovation attributes → SE | 0.615 | 1.151 | 0.062 | 9.988 | 0.000 | 0.500 |
| SE → ATT | −0.270 | 1.523 | 0.079 | 3.416 | 0.001 | 0.115 |
| SE → BLA | −0.104 | 1.062 | 0.071 | 1.450 | 0.147 | 0.013 |
| SE → BLT | 0.279 | 1.062 | 0.081 | 3.457 | 0.001 | 0.123 |
Structural path coefficients.
4.2.1 Coefficient of determination and predictive relevance
Table 7 presents the coefficient of determination (R2) values for the endogenous constructs. The R2 value indicates the proportion of variance in an endogenous construct explained by its predictor variables and is commonly used to assess the explanatory power of the structural model. R2 values of 0.75, 0.50, and 0.25 are generally considered substantial, moderate, and weak, respectively (Hair et al., 2017a). As presented in Table 7 and Figure 2, the model explained 58.4% of the variance in Attitude (ATT) (R2 = .584), indicating moderate explanatory power. The model also accounted for 40.5% of the variance in Blended Learning for Teaching (BLT) (R2 = .405), suggesting a moderate level of explanatory power. Similarly, 34.3% of the variance in Self-Efficacy (SE) (R2 = .343) was explained by the predictor constructs, indicating moderate explanatory power. In contrast, Blended Learning for Assessment (BLA) recorded an R2 value of 0.224, indicating that the model explained 22.4% of the variance, representing weak explanatory power. However, the higher-order constructs BLTA and Innovation Attributes each recorded an R2 of 0.999, indicating that 99.9% of their variance was explained by the respective predictor constructs. In summary, the R2 values indicate that the proposed structural model has varying levels of explanatory power across the endogenous constructs.
Table 7
| Endogenous construct | R-square | Explained variance | Interpretation |
|---|---|---|---|
| ATT | 0.584 | 58.4% | Moderate |
| BLA | 0.224 | 22.4% | Weak |
| BLT | 0.405 | 40.5% | Moderate |
| BLTA | 0.999 | 99.9% | Extremely high |
| Innovation attributes | 0.999 | 99.9% | Extremely high |
| SE | 0.343 | 34.3% | Moderate |
R2 values.
Figure 2
Table 8 presents the model's out-of-sample predictive assessment using the PLSpredict procedure. All constructs recorded positive Q2predict values, indicating predictive capability. The highest Q2predict values were observed for Innovation Attributes (0.999) and Attitude (0.533), followed by Self-Efficacy (0.308), Blended Learning for Teaching (0.293), Blended Learning for Teaching and Assessment (0.278), and Blended Learning for Assessment (0.130). These results indicate variation in predictive performance across the endogenous constructs, with the strongest prediction obtained for Innovation Attributes and Attitude. Prediction errors were further assessed using RMSE and MAE. Innovation Attributes recorded the lowest prediction errors (RMSE = 0.025; MAE = 0.019), whereas BLA recorded the highest (RMSE = 0.948; MAE = 0.621). The RMSE and MAE values are therefore interpreted comparatively across the constructs rather than against arbitrary absolute cut-off values. Overall, the results indicate that the model possesses meaningful predictive capability across the endogenous constructs, with particularly strong predictive performance for Innovation Attributes and ATT. The relatively lower Q2predict values and higher prediction errors observed for BLA suggest that additional factors may be required to improve the prediction of preservice mathematics teachers' adoption of blended learning for assessment. The predictive assessment therefore provides additional evidence that the proposed model has useful out-of-sample predictive capability, although its predictive performance varies across the individual constructs.
Table 8
| Construct | Q2predict | RMSE | MAE |
|---|---|---|---|
| ATT | 0.533 | 0.686 | 0.538 |
| BLA | 0.130 | 0.948 | 0.621 |
| BLT | 0.293 | 0.850 | 0.614 |
| BLTA | 0.278 | 0.863 | 0.618 |
| Innovation attributes | 0.999 | 0.025 | 0.019 |
| SE | 0.308 | 0.845 | 0.599 |
Predictive relevance (Q2).
5 Discussion
5.1 Innovation attributes, attitude and self-efficacy
A central finding was the strong positive association between innovation attributes and both attitude and self-efficacy. This finding indicates that preservice mathematics teachers who perceived blended learning as advantageous, compatible with their practices and needs, observable in its outcomes, and sufficiently trialable tended to report more favourable attitudes and stronger confidence regarding its use. The relatively large effect sizes associated with these relationships further indicate that Innovation Attributes constituted an important explanatory component of the psychological readiness of the respondents. The finding provides strong support for the Diffusion of Innovation (DOI) perspective, particularly its proposition that potential adopters evaluate an innovation according to characteristics that reduce uncertainty and increase its perceived value (Rogers, 2003). In the context of mathematics teacher education, the importance of these attributes is particularly understandable because blended learning requires preservice teachers to reconsider established approaches to lesson planning, interaction, resource use, and assessment. When an innovation is perceived as compatible with existing pedagogical practices and capable of providing observable benefits, the psychological uncertainty associated with adoption is likely to be reduced. Trialability may further allow preservice teachers to experiment with blended approaches before assuming the demands associated with sustained implementation. This finding is consistent with Alhammadi et al. (2023) and Hamad et al. (2024), who reported that perceptions of innovation characteristics, particularly usefulness and compatibility, were associated with technology acceptance and teachers' intentions to use innovative technologies. However, the present study extends this evidence by showing that innovation attributes are not only related to favourable attitudes but are also strongly associated with self-efficacy. Thus, the relevance of DOI in this context appears to extend beyond the conventional explanation of adoption intention. Perceptions of an innovation may also correspond with teachers' beliefs about their capacity to work with that innovation.
The relationship with self-efficacy also provides a useful connection between DOI and Social Cognitive Theory (SCT). From an SCT perspective, efficacy beliefs develop through experiences that influence individuals' judgements about their capability to perform particular behaviours. An innovation that is perceived as understandable, compatible, observable, and possible to experiment with may provide a more manageable psychological environment in which such efficacy beliefs can develop. The finding therefore suggests that the external characteristics of blended learning and the internal confidence of preservice teachers should not be considered entirely separate dimensions of adoption. Rather, perceptions of the innovation may form part of the psychological context within which technology-related efficacy develops. Importantly, the present study does not establish that favourable perceptions of innovation cause higher self-efficacy or more positive attitudes because the data were collected cross-sectionally. Instead, the findings indicate strong empirical associations that support the usefulness of integrating DOI and SCT when examining preservice teachers' readiness for blended learning. This integration represents an important contribution because it positions innovation characteristics as an external adoption-related condition and self-efficacy as an internal psychological correlate within the same explanatory framework.
5.2 Institutional training and blended learning adoption
Institutional training produced a different pattern. It was not significantly associated with attitude or self-efficacy, but it was significantly associated with both blended learning for assessment and, more strongly, blended learning for teaching. This pattern is important because it suggests that institutional support may be more closely connected with the practical enactment of blended learning than with the psychological dispositions that precede adoption. The absence of a significant association between institutional training and attitude indicates that training opportunities alone may not necessarily correspond with more favourable evaluations of blended learning. Similarly, the non-significant relationship between institutional training and self-efficacy suggests that participation in institutional training does not automatically translate into stronger beliefs about technological capability. This contrasts with evidence from studies such as Getenet et al. (2024) and Nuhoğlu et al. (2024), which associated digital learning experiences and blended learning experiences with stronger efficacy-related outcomes. One possible explanation is that the present construct captured institutional training and support rather than the quality, intensity, duration, or mastery experiences generated by that training. Training that provides exposure to platforms and procedures may therefore be insufficient to develop efficacy unless it includes repeated practice, feedback, authentic classroom application, and opportunities to experience successful performance.
The stronger association between institutional training and blended learning for teaching provides a complementary interpretation. Blended teaching requires access to platforms, instructional resources, digital infrastructure, technical guidance, and institutional procedures. Consequently, institutional support may reduce practical barriers that preservice teachers cannot independently address. This is consistent with Ibrahim and Nat (2019) and Turnbull et al. (2021), who emphasised institutional readiness and support as important conditions for technology adoption. The finding also accords with Tong et al. (2023), who highlighted institutional and pedagogical support as important components of effective blended learning implementation. The association with blended learning for assessment was also significant, although weaker than the relationship with teaching. This suggests that institutional training may support assessment-related adoption by increasing familiarity with digital assessment procedures, feedback mechanisms, and technology-supported evaluation. However, the comparatively stronger relationship with teaching may reflect the broader technological and pedagogical demands associated with integrating digital tools into instructional delivery. In other words, institutional training appears to be more closely related to what preservice teachers report being prepared to do than to how positively they evaluate blended learning or how confident they feel in using it. Taken together, these findings refine the role of institutional support in the adoption process. Institutional training should not be viewed simply as an antecedent of positive attitudes or technological confidence. Its stronger relationship with behavioural dimensions suggests that institutional conditions may operate partly by making implementation more feasible. This distinction is particularly relevant in the Nigerian context, where access to infrastructure, institutional digital platforms, technical support, and opportunities for authentic technology use may constrain the translation of individual readiness into classroom practice.
5.3 Self-efficacy and attitude toward blended learning
The relationship between self-efficacy and attitude requires particularly cautious interpretation. The structural model showed a significant negative association between the two constructs, whereas the bivariate relationship reported in the discriminant validity analysis was positive. This difference indicates that the negative structural coefficient should not immediately be interpreted as evidence that greater technological confidence is associated with less favourable attitudes toward blended learning. Rather, the reversal in direction after the inclusion of innovation attributes and other predictors suggests a possible suppression or sign-reversal effect arising from shared variance among the predictors. Innovation Attributes, for example, were strongly associated with both self-efficacy and attitude. Once this common explanatory variance is statistically controlled, the remaining variance in self-efficacy may relate differently to attitude. The significant coefficient should therefore be understood as a conditional structural association rather than a straightforward negative relationship between the two psychological constructs.
This interpretation is important because the finding differs from the broader evidence suggesting a positive relationship between technological self-efficacy and attitudes toward technology. Ye et al. (2022), Getenet et al. (2024), and Scherer et al. (2017), for example, reported positive relationships between technology-related confidence and attitudes or engagement. The discrepancy should therefore not be explained solely by proposing that highly efficacious preservice teachers are more aware of infrastructural difficulties, because infrastructure awareness was not directly measured in the present study. Such an explanation remains a possibility requiring future empirical investigation rather than a conclusion supported by the current data. The more defensible contribution of this finding is methodological and theoretical. It demonstrates that the relationship between efficacy and attitude may appear positive at the bivariate level but change when theoretically related predictors are incorporated into a multivariate structural model. This highlights the importance of examining predictor overlap and conditional relationships rather than interpreting individual path coefficients in isolation. It also suggests that future studies should explicitly investigate whether institutional readiness, technological constraints, perceived workload, or pedagogical concerns explain the residual component of self-efficacy that remains after innovation-related perceptions are controlled.
5.4 Attitude and the differentiated intention to adopt blended teaching and assessment
Attitude was significantly associated with blended learning for assessment but not with blended learning for teaching. This distinction provides an important insight into the multidimensional nature of blended learning adoption. A favourable attitude toward blended learning appears to correspond more closely with willingness to use it for assessment than for instructional delivery. One explanation is that technology-supported assessment may represent a comparatively bounded form of pedagogical change. Online quizzes, assignments, feedback, grading, and progress monitoring can often be incorporated into existing instructional routines without requiring a complete redesign of classroom pedagogy. Blended teaching, in contrast, requires more extensive changes to lesson planning, instructional interaction, classroom management, resource development, and student engagement. Thus, positive attitudes may be sufficient to correspond with assessment-related adoption but may not translate into teaching-related adoption when implementation demands are substantially greater.
This finding partly supports the technology acceptance perspective associated with Davis (1989), in which favourable attitudes are expected to correspond with technology-related behavioural intentions. It also aligns with Birbal et al. (2018) and An et al. (2022), who reported the importance of positive attitudes in technology adoption. However, the non-significant association with blended learning for teaching qualifies this relationship. The present findings suggest that attitude is not uniformly related to all forms of blended learning adoption. The relationship depends on the particular pedagogical activity for which technology is being considered. The finding therefore contributes to the literature by demonstrating the importance of distinguishing between teaching and assessment rather than treating blended learning adoption as a single undifferentiated behavioural outcome. In mathematics education, where teaching involves complex representations, problem-solving activities, classroom interaction, and pedagogical decision-making, the threshold for translating a positive attitude into instructional adoption may be higher than for adopting technology-supported assessment practices.
5.5 Self-efficacy and the differentiated intention to adopt blended teaching and assessment
Self-Efficacy showed the opposite pattern to Attitude: it was significantly associated with blended learning for teaching but not with blended learning for assessment. This finding indicates that confidence in using blended learning technologies corresponds more strongly with instructional application than with assessment-related adoption. The relationship is theoretically consistent with SCT because teaching with blended technologies requires individuals to judge whether they can successfully perform multiple technology-mediated instructional tasks. Preservice teachers who believe they can operate digital platforms, develop resources, manage online and face-to-face activities, and facilitate student interaction may consequently be more willing to envisage blended teaching as part of their future professional practice. The finding is consistent with Henriksen et al. (2019), Pozas and Letzel (2021), and Nuhoğlu et al. (2024), who identified technology-related self-efficacy as an important correlate of teachers’ or preservice teachers' technology integration.
The absence of a significant association between self-efficacy and blended learning for assessment is nevertheless informative. Assessment decisions may depend less on individual technological confidence and more on institutional policies, curriculum requirements, assessment procedures, available platforms, and expectations concerning evidence of learning. A preservice teacher may therefore feel capable of using digital assessment tools but still not intend to use them when institutional or curricular conditions do not support such practices. This distinction reinforces the argument that individual capability and contextual opportunity represent complementary rather than interchangeable conditions for technology adoption. The contrasting patterns for attitude and self-efficacy further strengthen the case for separating blended teaching from blended assessment. Attitude was associated with assessment but not teaching, whereas Self-Efficacy was associated with teaching but not assessment. These findings suggest that different psychological resources may become salient at different stages or forms of technology integration. Such differentiation is particularly relevant to teacher education because preparing future teachers for blended learning requires more than developing general positive attitudes or technical confidence; it requires targeted preparation for the distinct demands of instructional and assessment practices.
6 Implications for theory and education 5.0
The combined findings provide support for a complementary application of DOI and SCT while also clarifying the limits of a two-theory explanation. DOI is particularly useful for explaining why perceptions of blended learning characteristics are strongly associated with attitude and self-efficacy, whereas SCT provides a stronger basis for understanding the role of efficacy beliefs in technology-related behaviour. Institutional training, meanwhile, represents an important contextual condition through which individual readiness may be translated into practical adoption. The findings therefore suggest that blended learning adoption is best understood as an interaction between perceptions of the innovation, individual psychological readiness, and institutional conditions rather than as the outcome of a single theoretical mechanism. The findings provide important implications for understanding blended learning adoption within the Education 5.0 context. Unlike technology-centred models of educational transformation, Education 5.0 emphasises the purposeful integration of digital technologies with human capabilities to create personalised, adaptive, sustainable, and learner-centred educational experiences. From this perspective, the value of blended learning is not determined simply by whether technology is available, but by whether preservice teachers perceive it as pedagogically meaningful and are sufficiently prepared to use it to respond to learners’ diverse needs. The strong associations between innovation attributes, attitude, and self-efficacy suggest that preservice teachers are more psychologically prepared for Education 5.0 when blended learning is perceived as advantageous, compatible, observable, and suitable for experimentation. This indicates that technology adoption should focus not merely on digital competence but on understanding how technology can enhance human-centred mathematics teaching. The differentiated relationships involving attitude and self-efficacy further reinforce this perspective. Attitude was associated with blended assessment, whereas self-efficacy was associated with blended teaching, suggesting that Education 5.0 requires different forms of readiness for different pedagogical functions. Institutional training, in turn, was more strongly associated with actual blended teaching and assessment practices than with psychological dispositions. This highlights the importance of institutional environments that provide sustained opportunities for authentic technology integration rather than one-off training. Overall, the findings suggest that advancing Education 5.0 in mathematics teacher education requires more than increasing technology use. It requires the integration of innovation awareness, technological self-efficacy, pedagogical capability, and institutional support so that digital technologies function as partners in personalised, adaptive, and human-centred teaching and assessment. The study therefore extends technology-adoption research by demonstrating that the transition toward Education 5.0 involves not only whether preservice teachers adopt blended learning, but also how and for which pedagogical purposes they use it.
7 Conclusion
This study aimed to analyze the factors that influence preservice mathematics teachers' intention to implement blended learning in teaching and assessment processes in the era of Education 5.0. The results indicate that the process of adopting blended learning is multifaceted, with institutional, psychological, and innovation-related factors playing a role. The findings showed that the institutional training results significantly predicted preservice teachers' intentions to use blended learning in teaching and assessment. Thus, it was determined that structured institutional training can help prepare future educators for technology-based education. Moreover, the results showed that the innovation attributes were significant predictors of preservice teachers' attitudes and self-efficacy, suggesting that perceived usefulness, compatibility, trialability, and visibility of blended learning practices were key factors in technology acceptance. Additionally, the results of this study revealed that self-efficacy and attitude have significant and complex effects on intentions to adopt blended learning. Self-efficacy had a positive effect on teaching-related adoption, while attitude showed a stronger effect on assessment-related adoption. The results indicate that equipping preservice teachers for Education 5.0 is not just a matter of technical skills but also involves cultivating positive perceptions, confidence, and hands-on experience with digital pedagogies and digital assessment methods. In general, the study provides insight into the adoption of blended learning among future mathematics teachers by developing a model, supported by empirical data from these teachers, that combines institutional support, psychological readiness, and innovation characteristics. The results highlight the importance of the sustainable implementation of blended learning by the teacher education institution in creating an environment that supports preservice teachers in developing the knowledge, confidence, and motivation needed for effective technology integration.
8 Limitations and suggestions for future research
Despite the benefits of this study, some limitations should be noted. First, the study used a cross-sectional survey design, collecting data at a single point in time. The results are thus based on respondents' perceptions and intentions at this stage of their training and may not reflect changes in adoption behaviour over time. Longitudinal designs are recommended for future studies to investigate the growth in preservice teachers' intentions as they gain more classroom experience. Second, the research was based on self-reported questionnaire data, which can be driven by individuals' perceptions, social desirability, and/or the individual's understanding of blended learning. Qualitative interviews, classroom observations, or actual technology usage data could be added to future research to achieve more insightful information on the intentions and implementation of blended learning in the classroom. Third, the study was limited to preservice mathematics teachers from public universities in Southwest Nigeria, and the findings may not generalize to other regions, disciplines, or educational contexts. Future research should aim to broaden the sample to more universities in other geographical locations and preservice teachers from different subject areas. Lastly, while the study included relevant institutional, psychological, and innovation-related factors, other factors such as digital competence, technological infrastructure, perceived barriers, and digital literacy could be relevant in the context of blended learning adoption. The model could be expanded and further developed to capture additional contextual and individual factors, thereby improving understanding of blended learning adoption in the evolving landscape of Education 5.0.
Statements
Data availability statement
The raw data supporting the conclusions of this article will be made available by the author, without undue reservation.
Ethics statement
The studies involving humans were approved by Faculty of Education Research and Ethics Committee of Federal University Oye – Ekiti, Nigeria. The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study.
Author contributions
OF: Investigation, Visualization, Resources, Validation, Conceptualization, Formal analysis, Data curation, Project administration, Writing – review & editing, Methodology, Writing – original draft, Software, Supervision, Funding acquisition.
Funding
The author declares that financial support was received for this work and/or its publication. The author declares that no funds were received for conducting the study. Its publication was funded by the Directorate of Research and Innovation, University of Zululand, South Africa.
Acknowledgments
The author sincerely thanks the preservice mathematics teachers from the participating public universities in Southwest Nigeria for their voluntary participation and valuable contributions to this study.
Conflict of interest
The author declares that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Generative AI statement
The author declares that generative AI was not used in the creation of this manuscript.
Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.
Publisher’s note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
Supplementary material
The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/feduc.2026.1926606/full#supplementary-material
References
1
Abu-AlhaijaA. S. (2019). From epistemology to structural equation modeling: an essential guide in understanding the principles of research philosophy in selecting the appropriate methodology. Aust. J. Basic Appl. Sci.13 (9), 122–128. 10.22587/ajbas.2019.13.9.12
2
AjithaM.VakkilM. (2025). Revolutionizing education: embracing the future with education 5.0. Int. J. Emerg. Knowl. Stud.4 (2), 49–55. 10.70333/ijeks-04-02-s-011
3
AlhammadiK.MarashdehH.HussainM. (2023). Assessing the effect of innovation diffusion and technology readiness theories on attitude, behavioral intention and implementation of smart learning. Cross Cult. Strat. Manag.30 (4), 657–675. 10.1108/CCSM-12-2022-0213
4
AlkaabiA. M.QablanA.AlkatheeriF.AlnaqbiA. R. M.AlawlakiM.AlameriL.et al (2023). Experiences of university teachers with rotational blended learning during the COVID-19 pandemic: a qualitative case study. PLoS One18 (10) e0292796. 10.1371/journal.pone.0292796
5
AnX.ChaiC. S.LiY.ZhouY.ShenX.ZhengC.et al (2022). Modeling English teachers’ behavioral intention to use artificial intelligence in middle schools. Educ. Inf. Technol.28 (5), 5187–5208. 10.1007/s10639-022-11286-z
6
BanduraA. (1986). Social Foundations of Thought and Action: a Social Cognitive Theory. Englewood Cliffs, NJ: Prentice-Hall.
7
BanduraA. (2023). Social Cognitive Theory: An Agentic Perspective on Human Nature. Hoboken, NJ: John Wiley & Sons.
8
BirbalR.RamdassM.HarripaulC. (2018). Student teachers’ attitudes towards blended learning. J. Educ. Hum. Dev.7 (2), 9–26. 10.15640/jehd.v7n2a2
9
ButtS.MahmoodA.SaleemS. (2022). The role of institutional factors and cognitive absorption on students’ satisfaction and performance in online learning during COVID 19. PLoS One17 (6), e0269609. 10.1371/journal.pone.0269609
10
ChinchorkarS.JadhavJ. (2024). Commentary: transforming education 4.0 to education 5.0: sustainable education. J. Res. Innov. Teach. Learn.17 (2), 408–413. 10.1108/jrit-09-2024-194
11
CreswellJ. W. (2014). Research Design: Qualitative, Quantitative and Mixed Methods Approaches, 4th Edn. Thousand Oaks, CA: SAGE Publications, Inc.
12
DavisF. D. (1989). Perceived usefulness, perceived ease of use, and user acceptance of information technology. MIS Q.13 (3), 319–340. 10.2307/249008
13
ErwinE.KuswandiD. (2024). Tinjauan pustaka: model pembelajaran blended learning di era society 5.0. Inopendas Jurnal Ilmiah Kependidikan7 (1), 39–47. 10.24176/jino.v7i1.11553
14
FalebitaO. S.KokP. J.AyanwoyeO. K.OgunjobiA. O. (2025). Virtual reality in education 4.0: pre-service teachers’ technology readiness and behavioral intention. Int. J. Technol. Knowl. Soc.21 (2), 151–177. 10.18848/1832-3669/cgp/v21i02/151-177
15
FalebitaO. S.PopoolaA. A.TalasiT.AkinwamideC. O. (2026). Enhancing mathematics interest and retention through student teams achievement division (STAD): a quasi-experimental study in Nigerian secondary schools. Eur. J. STEM Educ.11 (1), 23. 10.20897/ejsteme/18459
16
FornellC.LarckerD. F. (1981). Evaluating structural equation models with unobservable variables and measurement error. J. Mark. Res.18 (1), 39–50. 10.1177/002224378101800104
17
GengS.LawK. M. Y.NiuB. (2019). Investigating self-directed learning and technology readiness in blending learning environment. Int. J. Educ. Technol. High. Educ.16 (1). 10.1186/s41239-019-0147-0
18
GetenetS.HaeuslerC.RedmondP.CantleR.CrouchV. (2024). First-year preservice teachers’ understanding of digital technologies and their digital literacy, efficacy, attitude, and online learning engagement: implication for course design. Technol. Knowl. Learn.29 (3), 1359–1383. 10.1007/s10758-023-09724-z
19
HairJ. F.HultG. T. M.RingleC. M.SarstedtM. A. (2017a). Primer on Partial Least Squares Structural Equation Modeling (PLS-SEM), 2nd Edn. Thousand Oaks, CA: Sage.
20
HairJ. F.HultG. T. M.RingleC. M.SarstedtM.DanksN. P.RayS. (2021). “Evaluation of reflective measurement models,” in Partial Least Squares Structural Equation Modeling (PLS-SEM) Using R: A Workbook, eds. HairJ. F.Jr.HultG. T. M.RingleC. M.SarstedtM.DanksN. PRayS (Cham: Springer International Publishing), 75–90. 10.1007/978-3-030-80519-7_4
21
HairJ. F.MatthewsL. M.MatthewsR. L.SarstedtM. (2017b). PLS-SEM or CB-SEM: updated guidelines on which method to use. Int. J. Multivar. Data Anal.1 (2), 107–123. 10.1504/IJMDA.2017.087624
22
HamadF.ShehataA.HosniN. A. (2024). Predictors of blended learning adoption in higher education institutions in Oman: theory of planned behavior. Int. J. Educ. Technol. High. Educ.21 (1). 10.1186/s41239-024-00443-8
23
HenriksenD.MehtaR.RosenbergJ. M. (2019). Supporting a creatively focused technology fluent mindset among educators: survey results from a five-year inquiry into teachers’ confidence in using technology. J. Technol. Teach. Educ.27 (1), 63–95. 10.70725/084183mmdjft
24
HenselerJ.RingleC. M.SarstedtM. (2015). A new criterion for assessing discriminant validity in variance-based structural equation modeling. J. Acad. Mark. Sci.43 (1), 115–135. 10.1007/s11747-014-0403-8
25
HidayatM. L.PrayitnoH. J.AnifS.MeccawyM.KhanzadaT. J. S. (2023). Science learning in answering digital competency needs of preservice math and science teachers in the STEM context. J. Pendidik. IPA Indones.12 (3), 410–422. 10.15294/jpii.v12i3.43960
26
HongA. L. T.StapaM.TianK. X. (2025). Teachers’ self-efficacy beliefs and blended learning practice in the Malaysian primary ESL classroom: exploring belief–practice alignment. Malay. J. ELT Res.22 (2), 40–57. 10.52696/BGDV8743
27
HongA. L. T.StapaM.TianK. X. (2026). Aligning teachers’ attitudes with blended learning practices in Malaysian primary ESL classrooms. Inter. J. Lang. Educ. Appl. Linguist., 16 (1), 76–87. 10.15282/ijleal.v16i1.13188
28
IbrahimM. M.NatM. (2019). Blended learning motivation model for instructors in higher education institutions. Int J Educ Technol High Educ.16 (1), 12. 10.1186/s41239-019-0145-2
29
JailaniN.RosliR.MahmudM. S. (2025). Factors influencing mathematics teachers’ blended learning: a systematic review. Int. J. Learn. Teach. Educ. Res.24 (1), 397–419. 10.26803/ijlter.24.1.20
30
KolotiA. C.JitaT.JitaL. C. (2026). Pre-service Teachers’ experiences of blended learning: a diffusion of innovation theory perspective using mixed methods. Int. J. Learn. Teach. Educ. Res.25 (1), 234–250. 10.26803/ijlter.25.1.12
31
MaK.HaoT. (2024). Predictors of pre-service teachers’ learning engagement in blended teaching. Dist. Educ.46 (3), 477–495. 10.1080/01587919.2024.2387035
32
MohammadiM.PaasivaaraM.KasurinenJ. (2025). Blended learning in higher education: good practices in platforms and teachers support, enhancing students motivation. Educ. Inf. Technol.30 (18), 26001–26024. 10.1007/s10639-025-13770-8
33
MoolaZ.RamailaS.DhurumrajT. (2024). Teachers’ views on the interdependence of humanity and technology in life sciences teaching and learning within the context of the 5IR. Int. J. Learn. Teach. Educ. Res.23 (7), 461–475. 10.26803/ijlter.23.7.24
34
NuhoğluH.SarkinDBŞAşiroğluS. (2024). Digital transformation in science education: teachers’ self-efficacy of distance learning and blended learning experiences. Particip. Educ. Res.11 (3), 126–145. 10.17275/per.24.38.11.3
35
ParhusipA.SitanggangM. P. (2025). “Educational technology innovation in transformation education era 5.0,” in Advances in Social Science, Education and Humanities Research/Advances in Social Science, Education and Humanities Research, 919, 65–72. 10.2991/978-2-38476-398-6_10
36
PermadiM.RisnawatiA. (2025). Period front education era revolution industry 4.0 going to society 5.0. Matriks Jurnal Sosial Dan Sains7 (1), 27–34. 10.59784/matriks.v7i1.29793
37
PozasM.LetzelV. (2021). “Do you think you have what it takes?” – exploring predictors of pre-service teachers’ prospective ICT use. Technol. Knowl. Learn.28 (2), 823–841. 10.1007/s10758-021-09551-0
38
Quintana-OrdorikaA.EsturoE. C.PortilloJ.RuízU. G. (2023). Integrating the maker pedagogical approach in teacher training: the acceptance level and motivational attitudes. Educ. Inf. Technol.29 (1), 815–841. 10.1007/s10639-023-12293-4
39
RindA. A.AsadM. M.SherwaniF. (2024). Pre-service teachers' digital self-efficacy towards education 5.0: a narrative literature review. In Digital Transformation in Higher Education, Part B: Cases, Examples and Good Practices, eds. LytrasM. D.SerbanA. C.AlkhaldiA.MalikS.AldosemaniT. (Bingley: Emerald Publishing Limited), 29–56. 10.1108/978-1-83608-424-220241003
40
RogersE. M. (2003). Diffusion of Innovations, 5th Edn. New York Singh and kaur 2024 – New Dehli Simon and Schuster.
41
SaputraS. A.SiswantoD. H.SuryatamaH. (2025). The urgent need for mathematics learning innovation in the era of society 5.0. Acad. J. Teach. Learn.4 (2), 71–81. 10.59373/academicus.v4i2.82
42
SchererR.TondeurJ.SiddiqF.BaranE. (2017). The importance of attitudes toward technology for pre-service teachers’ technological, pedagogical, and content knowledge: comparing structural equation modeling approaches. Comput. Human. Behav.80, 67–80. 10.1016/j.chb.2017.11.003
43
SinghD.KaurJ. (2024). “The transformation of education: from education 1.0 to education 5.0,” in Education 5.0: Shaping the Future of Learning in a Digital Age, eds. SinghR.GehlotA.MisraN.Mohan JoshiL. (New Delhi: Wisdom Leaf Press (A division of ICAPSR)), 01–07. 10.55938/wlp.v1i5.174
44
SyafitriK. S. H.Yani T.A.SiregarN. (2025). Inovasi teknologi pembelajaran matematika diantara desakan era society 5.0 di kota pontianak. JP2M11 (1), 635–644. 10.29100/jp2m.v11i1.6927
45
TongD. H.UyenB. P.NganL. K. (2023). Blended learning in mathematics teacher education: a systematic review. Int. J. Emerg. Technol. Learn.18 (17), 196–222. 10.3991/ijet.v18i17.36985
46
TurnbullD.ChughR.LuckJ. (2021). Transitioning to E-learning during the COVID-19 pandemic: how have higher education institutions responded to the challenge?Educ. Inf. Technol.26 (5), 6401–6419. 10.1007/s10639-021-10633-w
47
VenkateshV.MorrisM. G.DavisG. B.DavisF. D. (2003). User acceptance of information technology: toward a unified view. MIS Q.27 (3), 425–478. 10.2307/30036540
48
WeigandH.-G.Donevska-TodorovaA.FaggianoE.IannoneP.MedováJ.TabachM.et al (2022). “MEDA3 Mathematics education in digital age 3,” in Proceedings of the 13th ERME Topic Conference (ETC13) Held on 7–9 September 2022 in Nitra, Slovakia (HAL (Le Centre Pour La Communication Scientifique Directe). Available online at:https://hal.science/hal-03925304(Accessed June 15, 2026).
49
YeL.KuangM.LiuS. (2022). ICT self-efficacy, organizational support, attitudes, and the use of blended learning: an exploratory study based on English teachers in basic education. Front. Psychol.13, 941535–941535. 10.3389/fpsyg.2022.941535
50
YeopM. A.YaakobM. F. M.WongK. T.DonY.ZainF. M. (2018). Implementation of ICT policy (blended learning approach): investigating factors of behavioural intention and use behaviour. Int. J. Instr.12 (1), 767–782. 10.29333/iji.2019.12149a
Summary
Keywords
attitude, blended learning, education 5.0, innovation attributes, institutional training, preservice mathematics teachers, self-efficacy
Citation
Falebita OS (2026) Modeling mathematics preservice teachers' intention to adopt blended learning for teaching and assessment in education 5.0: integrating self-efficacy, attitude, and innovation attributes. Front. Educ. 11:1926606. doi: 10.3389/feduc.2026.1926606
Received
02 July 2026
Revised
09 August 2026
Accepted
31 August 2026
Published
17 September 2026
Volume
11 - 2026
Edited by
Jumoke Iyabode Oladele, University of Ilorin, Nigeria
Reviewed by
Andy Lim Teik Hong, University of Technology Malaysia, Malaysia
Philip Dorsah, C K Tedam University of Technology and Applied Sciences, Ghana
Updates
Copyright
© 2026 Falebita.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
*Correspondence: Oluwanife Segun Falebita oluwanifefalebita@gmail.com; falebitaO@unizulu.ac.za
Disclaimer
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.