Charting the Future of Maritime Education and Training: A Technology-Acceptance-Model-Based Pilot Study on Students’ Behavioural Intention to Use a Fully Immersive VR Engine Room Simulator

Abstract

Fully immersive engine room simulators are increasingly recognised as prominent tools in advancing maritime education and training. However, end-users’ acceptance of these innovative technologies remains insufficiently explored. To address this research gap, this case-specific pilot study applied the Technology Acceptance Model (TAM) to explore maritime engineering students’ intentions to adopt the newly introduced head-mounted display (HMD) virtual reality (VR) engine room simulator as a training tool. Sampling (N = 84) was conducted at the Faculty of Maritime Studies, University of Rijeka, during the initial simulator trials. Structural Equation Modelling (SEM) revealed that perceived usefulness was the primary determinant of students’ behavioural intention to accept the simulator as a tool for training purposes, acting both as a direct predictor and as a mediating variable, transmitting the positive effect of perceived ease of use onto the intention. By providing preliminary empirical evidence on the key factors influencing maritime engineering students’ intentions to adopt HMD-VR simulation technologies within existing training programmes, this study’s findings might offer valuable insights to software developers and educators in shaping future simulator design and enhancing pedagogical practices in alignment with maritime education and training (MET) standards.

1. Introduction

A proficient and well-trained maritime workforce is essential for a thriving and effective shipping industry [1]. The availability of highly skilled human capital to perform maritime labour is enhanced by the globally acknowledged maritime education and training (MET) model, which aims to standardise maritime education curricula, learning outcomes, and certifications in alignment with the Standards of Training, Certification and Watchkeeping (STCW) Convention [2,3,4]. Today, rapid changes in global business and operational dynamics within the maritime industry and the recent COVID-19 pandemic have prompted significant shifts in MET approaches, including the advocation and integration of innovative technologies, such as fully immersive VR training simulators, into MET curricula [5,6,7].

Virtual reality is a computer-generated environment that allows users to engage with and alter their perceptions by obtaining 3D sensory information through isolated screens or wearable displays, such as head-mounted displays. Generally, VR technology can be categorised into three main types: non-immersive, semi-immersive, and fully immersive. Non-immersive VR utilises standard computer monitors, allowing users to indirectly interact with a digital 3D environment through input devices such as a mouse, keyboard, or controller. Semi-immersive VR utilises a concave screen or wall projectors, providing more holistic sensory experiences (e.g., conventional maritime simulators). Fully immersive VR involves tracking users’ movement and adjusting the displayed virtual content accordingly through an HMD. There is a widespread consensus that fully immersive VR offers significant advantages over non-immersive and semi-immersive VR due to its elevated perceptual fidelity and an enhanced sense of perceived immersion, allowing users to interact with virtual environments in a manner similar to real-world experiences [8,9,10,11].

Driven by constant advancements in computing power and declining equipment costs, the VR industry is experiencing rapid expansion, with its global market projected to grow from less than USD 18 billion in 2024 to over USD 60 billion by 2029 [12]. The increasing adoption of VR technology is evident across various educational and professional training sectors [13,14,15,16,17]. VR technology is also increasingly being integrated into the maritime education and training industry [6,18,19,20,21,22], reflecting a broader global shift in the maritime industry toward the adoption of innovative technologies [10]. However, integrating high-tech digital technologies (e.g., HMD-VR engine room simulators) in MET curricula presents a particularly complex challenge. This complexity stems from the necessity to effectively bridge theoretical knowledge with practical skills while ensuring compliance with the STCW training and certification standards [6,23]. To address these challenges, a structured approach to understanding students’ adoption and acceptance of innovative technologies in MET is essential [24]. One of the most widely used theoretical frameworks for understanding end-users’ acceptance of technology or systems is the Technology Acceptance Model (TAM) [25]. This framework posits two key variables, perceived usefulness (PU) and perceived ease of use (PEU), as fundamental determinants affecting individual behavioural intention (BI) to accept a technology or information system [26,27]. To date, the original TAM framework and its extended modifications have been employed within various educational and technological domains [28,29,30,31,32,33], including Educational Virtual Environments (EVEs) [34,35,36,37,38]. While TAM is well documented in the maritime industry [39,40,41], its application in maritime higher education remains very limited [23,42], particularly in the context of understanding students’ acceptance of cutting-edge technology such as HMD-VR maritime simulators. Therefore, to address the gap in the existing literature, the main aim of this case-specific pilot study, conducted at the Faculty of Maritime Studies, University of Rijeka, is to explore the extent to which the core TAM variables, perceived ease of use and perceived usefulness, interact and influence maritime engineering students’ behavioural intention to use the HMD-VR engine room simulator for learning and training purposes. The decision to conduct a case-specific study was motivated by the fact that the integration of HMD-VR technology still represents a novel training modality within MET institutions [6]. Therefore, the newly introduced HMD-VR engine room simulator laboratory at the Faculty of Maritime Studies, University of Rijeka, provided an appropriate setting for conducting a pilot investigation into students’ acceptance of this emerging technology.

In line with the research context and the study’s aim, the following research question is stipulated:

1. To what extent can the core constructs of the Technology Acceptance Model explain maritime engineering students’ future behavioural intention to use the HMD-VR engine room simulator for education and training purposes?

HMD-VR Technology in Maritime Engineering Simulator Training

Cambridge University Press defines a simulation as “a model of a set of problems or events that can be used to teach someone how to do something or the process of making such a model” [43]. Simulation-based education, rooted in Kolb’s Experiential Learning Theory [44], has been widely adopted in different educational domains [45,46,47], including MET [48,49]. Aligned with constructivist learning principles, this pedagogical approach places learners at the centre of the learning process, allowing them to actively develop an understanding of the topic through exploration, reflection, and experience [47,50].

Over the past three decades, simulator-based training has become a cornerstone of the MET curricula, providing a structured and standardised platform for students and professional seafarers to develop an array of operational competencies in alignment with the STCW Convention [3,4,51]. Among the diverse range of maritime simulators used in MET [48,52], Marine Engine Room Simulators (MERSs) have gained particular significance in maritime engineering training. To date, MERSs have been proven to have a prominent role in knowledge acquisition, reinforcing procedural compliance, and developing critical technical and decision-making skills under simulated stress conditions [49,53,54,55]. However, limitations inherent to all conventional simulators, including MERSs, regarding constrained levels of representational fidelity, dependency on instructor-led delivery, substantial maintenance expenses, the necessity for fixed physical infrastructure, and increasing global trends in maritime digitalisation, have catalysed a paradigm shift toward seeking innovative technologies for advancing traditional simulator-based training methods [6,10]. Recent evidence increasingly supports the efficacy of immersive VR-based simulators in MET, often indicating their pre-eminence over conventional simulator training modalities, such as through fostering learners’ engagement, active responsiveness, learning control and autonomy, conceptual comprehension, and knowledge retention [6,18,51,56]. Additionally, immersive VR simulators have proven instrumental in ensuring the continuity of learning during external disruptions, as exemplified during the global COVID-19 pandemic [7,57].

In contrast to their relatively established use in navigation simulation training [21,58,59], the integration of fully immersive VR simulators in maritime engineering education remains nascent. One of the few examples is a fully immersive engine room simulator developed by the company Adricom and piloted by the Faculty of Maritime Studies, University of Rijeka. This simulation platform demonstrates the pedagogical potential of HMD-VR by offering a cost-effective, portable, and educationally robust tool tailored specifically to undergraduate and graduate engineering students with no prior onboard experience [22]. In particular, the simulator is purposefully designed to facilitate student engagement in engine diagnostics, the interactive manipulation of control panels, and comprehensive procedural walkthroughs supplemented by real-time feedback mechanisms and performance monitoring [60,61,62].

Despite its enormous potential, the integration of HMD-VR simulation technology into EVEs still faces various logistical challenges. Evidence from different educational domains highlights several similar key barriers, including substantial initial investments for the development of specific immersive content, the need for constant technical support, and the lack of a standardised instructional framework tailored to the context-specific educational curricula [9,63,64]. The last-mentioned issue is particularly important for MET institutions as VR-based training must be aligned with the competencies and assessment benchmarks outlined by the MET standards to be efficient. This endeavour requires sustained collaboration among maritime educators, MET curriculum developers, software engineers, and end-users (e.g., students).

2. Theoretical Framework and Hypothesis Development

2.1. Technology Acceptance Model

The Technology Acceptance Model, developed and validated by F.D. Davis [26,27], is one of the most influential and widely applied theoretical frameworks for understanding the factors influencing users’ behaviours toward accepting technology or information systems [25]. Grounded in the theories of Reasoned Action [65] and Planned Behaviour [66], the core TAM framework posits that users’ behavioural intention to use a novel technological system is primarily determined by two cognitive beliefs: perceived ease of use and perceived usefulness [27]. BI refers to the user’s readiness to exert a planned effort to use the technology in the future [67,68], reflecting a complex interaction of both cognitive and/or affective factors that may lead to its adoption or rejection [37,69]. Across educational and learning environments, BI typically refers to students’ willingness to use innovative technology for learning and education in the future and/or their likelihood of recommending it to others [70,71,72]. PEU is defined as the degree to which a person believes that using a particular system would be effortless [27]. In the context of VR technology, it is typically assessed through the user’s subjective perceptions of the understandability, simplicity, intuitiveness, and user-friendliness of the VR technology platform [35,56,73]. PU, on the other hand, is defined as the degree to which a person believes that using a particular system would enhance their job performance [27]. Within the context of a VR learning environment, PU is conceptualised as the extent to which students perceive VR technology as beneficial for enhancing their learning performance [34,74]. In the present study, PU is assessed as the degree to which students believe training on the HMD-VR engine room simulator enhances their active learning. Active learning, as a pedagogical approach, places the learner at the centre of the training or educational process, highlighting elements such as interaction, customisation, engagement, and the application of previously acquired knowledge [75,76]. In the context of EVEs, active learning refers to the level of autonomy afforded by a VR platform, which, in turn, enables students to actively apply theoretical knowledge in VR-simulated training scenarios, increase responsiveness and engagement, and benefit from controlled, self-paced experiential training in a risk-free environment [70,73,77].

2.2. Cross-Domain Review of Interrelationships Among TAM Constructs

Over the past 25 years, a substantial amount of meta-analytic, systematic, and context-specific research has consistently confirmed the TAM’s theoretical and empirical robustness across various technological, organisational, and educational settings.

For instance, the authors of [78] conducted one of the first comprehensive conceptual TAM reviews and studied its empirical application, analysing over 100 studies published since the model’s introduction in the mid-1980s. Although their review did not include a statistical meta-analysis, their synthesis of empirical findings provided valuable insights into the consistency of the TAM’s core variables across various domains. Specifically, they reported that the relationship between PU and BI was statistically significant in 74% of the studies analysed. While the PEU significantly influenced PU in 69% of cases, its direct effect on BI was notably weaker, with significance observed in less than 60% of the studies.

The authors of [31] conducted a meta-analysis synthesising cross-domain TAM data from 26 empirical studies. Their results revealed that PU was the strongest predictor of users’ BI, with path coefficients ranging from 0.09 to 0.91 and a mean of β = 0.49. The findings also indicate that compared to the modest direct influence of PEU on BI (range β = 0.07–0.59; mean β = 0.27), its effect on PU was more substantial, with coefficients ranging from 0.42 to 0.54 and a mean β of 0.48.

Similarly, the authors of [79] expanded upon this work by analysing 88 studies involving over 12,000 participants. Their result supported previous findings confirming that the PU-to-BI relationship remained the strongest, with path coefficients ranging from 0.46 to 0.55 and a mean of β = 0.51. The direct effect of PEU on PU was also consistently significant, with a coefficient ranging from 0.43 to 0.54 and a mean of β = 0.48. Considerably weaker PEU-to-BI relationships, ranging from 0.15 to 0.27 with a mean of β = 0.19, also confirmed the uncertain and often limited direct role of perceived ease of use in predicting behavioural intention to use innovative technology.

Focusing on TAM application in e-learning, the authors of [80] conducted a meta-analysis of 42 studies, encompassing a total of 494 effect sizes. Their findings confirmed consistently stronger PU-BI (mean β = 0.40) and PEU-PU (mean β = 0.40) relationships compared to that between PEU and BI (mean β = 0.29). Complementing these findings, the authors of [30] conducted a systematic literature review of 71 studies focused on TAM application in the broader field of education. Their findings reaffirmed the continued dominance of PU in shaping BI, highlighting the indirect role of PEU primarily through its impact on PU.

With increasing interest in integrating VR technology into educational environments, the application of the TAM and its extensions has garnered growing attention within the academic community [34,35,38,72,74,81,82,83,84,85]. Across many reviews, the direct and significant effects of PU on BI and the effects of PEU on PU were consistent with evidence in the broader TAM literature. However, the findings present slight variations concerning the direct effect of PEU on BI. Specifically, in earlier studies conducted on less immersive VR platforms [74,83], PEU was found to directly influence BI; however, as VR technology became more immersive and technologically complex, the direct impact of PEU on BI diminished [34,35,72], with other affective variables (e.g., perceived immersion) gaining prominence.

Even though the application of TAM has received increasing scholarly attention in the maritime field, including education and training, its implementation remained largely outside fully immersive VR technologies. For instance, ref. [42] investigated the differential acceptance of Electronic Chart Display and Information (ECDIS) by comparing the attitudes of experienced shipmates to those of junior maritime students with no practical experience at Southern Taiwan University. Within the student group, PEU positively influenced PU (β = 0.33), while PU emerged as a significant direct predictor of BI (β = 0.25). Similarly, when exploring the determinants of adopting Global Positioning System (GPS) technology among artisanal fishers in Ghana [39], it was found that PU significantly predicts BI. In a broader maritime vocational context, the authors of [40] conducted a TAM-based study among 539 maritime logistic professionals to assess the adoption of digital transformation technologies. Their findings reaffirmed the central role of PU as the strongest predictor of BI (β = 0.70), while PEU was confirmed as a reliable antecedent of PU (β = 0.20), consistent with previous studies on the TAM.

2.3. Hypotheses

The literature review on empirical evidence across different technological, organisational, and educational domains, including application within the maritime sector, has consistently confirmed the TAM’s robustness. In particular, the model core constructs, PEU and PU, have been shown to form stable direct and indirect predictive capacities in explaining users’ behavioural intention to use innovative technology or systems. Reviewed cross-sectional meta-analyses, systematic reviews, and VR-context-specific studies consistently identified PU as the dominant predictor of BI. The direct relationship between PEU and BI is generally modest, primarily influencing behavioural intention indirectly through its strong and positive effect on PU. Based on this empirical evidence, this study proposes a conceptual model comprising four hypotheses to investigate the relationship between perceived core TAM variables and their influence on maritime engineering students’ intention to continue utilising the HMD-VR engine room simulator as a training tool (Figure 1).

H1. 

PU will have a positive and significant effect on BI.

H2. 

PEU will have a weak and non-significant effect on BI.

H3. 

PEU will have a positive and significant effect on PU.

H4. 

The effect of PEU on BI will be fully mediated by PU.

3. Materials and Methods

3.1. Hardware

The hardware specifications of all desktop computers employed to operate the VR engine room simulator, used in conjunction with the Meta Quest 2 (256 GB) VR headset, are comprehensively outlined in

Table 1. Hardware specifications of desktop computers.

Component	Specifications	PC 1	PC 2	PC 3	PC 4
CPU	Manufacturer	Intel	Intel	Intel	Intel
	Generation	13th	13th	13th	12th
	Model	Core i7-13700F	Core i7-13700F	Core i7-13700F	Core i7-12700F
	Frequency	2.10 GHz	2.10 GHz	2.10 GHz	2.10 GHz
RAM	Capacity	32 GB	32 GB	32 GB	8 GB
	Type	DDR5-4800	DDR5-4800	DDR5-4800	DDR4-2400
	Speed	2400 MHz	2400 MHz	2400 MHz	1200 MHz
GPU	Manufacturer	NVIDIA	NVIDIA	NVIDIA	NVIDIA
	Model	GeForce RTX 4070 Ti	GeForce RTX 4070 Ti	GeForce RTX 3080	GeForce RTX 4070 Ti
	VRAM	12 GB	12 GB	10 GB	12 GB

Intel and NVIDIA are located in Santa Clara, California, U.S.A.

. Each system was configured with a 64-bit Windows 10 Pro Version 22H2 operating system and integrated an Intel Wi-Fi 6E AX210 160 MHz wireless network interface card, which was explicitly configured to function as a dedicated hotspot on the 5 GHz frequency band for each respective HMD-VR. This point-to-point wireless configuration enhanced streaming stability, video rendering quality, and latency performance. Additionally, Meta Quest 2’s Air Link feature was configured to operate at its maximum supported bitrate of 200 Mbps to leverage data transmission capacity.

Software

The HMD-VR engine room simulator was developed using Blender 4.2 software for 3D modelling and using Unreal Engine 5.4 as the primary game development platform. The virtual engine room environment was designed using an authentic technical blueprint sourced from a roll-on/roll-off (RO-RO) vessel and designed as a double-deck spatial configuration with dimensions of 16.1 metres by 19.6 metres.

The lower deck houses two main four-stroke diesel engines, each connected to a shaft generator and propeller shaft, with three auxiliary diesel generators, a range of cooling systems, valve networks, pipelines, pumps, and other essential engine room apparatus. The upper deck includes two starting air compressors, a service air compressor, two Heavy Fuel Oil (HFO) purifiers, a diesel oil purifier, and associated systems such as lubricating and fuel oil pipelines, tanks, and control valves (Figure 2). The simulator was developed for compatibility with the Windows-based operating system and designed to support both wired and wireless VR headset connectivity, allowing interoperability with various HMD platforms. Key functional features, including actions and scenarios, are depicted in Figure 3.

3.2. Sample

The study population comprised 84 male students enrolled in undergraduate, graduate, and special education maritime engineering programmes at the Faculty of Maritime Studies at the University of Rijeka. Most participants (65.6%) pursued undergraduate or graduate degrees, while the remaining students (34.5%) were enrolled in specialised engineering programmes. The age range spanned from 18 to 46 years, with a median age of 22. Furthermore, 61.9% of participants reported having prior experience with VR simulators for entertainment purposes (i.e., gaming).

3.3. Experimental Protocol and Sampling

Data collection was conducted in the newly established VR laboratory within the Faculty of Maritime Studies, University of Rijeka, during two discrete sampling intervals: from 15 May to 15 July and from 1 September to 14 October 2024. During regularly scheduled coursework, students were voluntarily introduced to the VR laboratory and participated in the HMD-VR engine room simulator trials. Following class sessions, those who expressed interest assembled in the VR laboratory, where they received an introduction to the laboratory followed by a comprehensive oral briefing by lab staff outlining the underlying rationale and objectives of the VR lab initiative, contextualised with the MET framework. The briefing further included a detailed outline of the HMD-VR engine room simulator, encompassing its technical features and interactive functionalities.

After the introductory session, lab staff members conducted a live demonstration of the simulator using both a VR headset and a wall-mounted projector, enabling participants to observe real-time system operation and user interaction. Following the demonstration, each participant was invited to complete a brief, structured, guided tutorial specifically designed to facilitate user familiarisation with the simulator interface and functionality, all while under the supervision of lab personnel. Thereafter, participants were allowed to independently engage in a series of predefined trial exercises, with each session averaging approximately 20 min in duration.

Upon completion of the trial exercises, students were requested to complete a post-use questionnaire design to evaluate their experiences with the HMD-VR engine room simulator. The data were collected through a computer-based, self-administered instrument implemented via the Lime Survey (6.3.5) software. On average, participants required approximately 10 min to complete the questionnaire. Notably, there was full compliance among the sample, with all 84 completed questionnaires deemed valid for subsequent analysis. The full procedural workflow is presented in Figure 4.

3.4. Measurement Instrument

A measurement instrument was developed to measure engineering students’ acceptance of the HMD-VR engine room simulator for education and training purposes. The full questionnaire consisted of eight sections; however, for this study, three sections were used for analysis. Perceived ease of use was measured by four items adopted from [86]. Four items that measured perceived usefulness were adopted from [70]. Behavioural intention was assessed using three items drawn from [70,72]. All measurement items were systematically adapted to reflect the specific context of the current study. Items were presented as declarative statements and rated on a five-point Likert scale, where a score of one indicated “strongly disagree” and five indicated “strongly agree”.

3.5. Data Analysis

The data analysis process consisted of two primary steps. First, Confirmatory Factor Analysis (CFA) and Structural Equation Modelling (SEM) were conducted using Jeffrey’s Amazing Statistics Program 19.3.0 (JASP) to evaluate the measurement model’s validity and reliability, and then structural relationships were examined, including mediation effects. Second, the IBM Statistical Package for the Social Sciences 23.0 (SPSS) was utilised to conduct descriptive statistics, including age, study level, and previous experience with HMD-VR technology.

Measurement and structural model fit were assessed using multiple indices, following established guidelines [87,88,89]. The chi-square (χ²) test was used to evaluate the absolute model fit, with a non-significant result indicating an adequate fit. The relative chi-square (χ²/df) was deemed acceptable if below 3. Comparative Fit Index (CFI) and Tucker–Lewis Index (TLI) values above 0.90 indicate acceptable fit, with values above 0.95 indicating excellent fit. A Root Mean Square Error of Approximation (RMSEA) value below 0.08 suggests an acceptable fit, while values below 0.05 indicate a good fit. A Standardised Root Mean Square Residual (SRMR) value below 0.08 generally indicates a good model fit. With 11 items and a sample size of 84, the item-to-sample ratio was approximately 7.6:1, which is deemed adequate [90].

Internal consistency was evaluated using Cronbach’s alpha (α) [91] and McDonald’s omega (ω) [92]. Cronbach’s alpha values above 0.7 are considered acceptable, and values above 0.80 indicate good reliability. McDonald’s omega values above 0.70 also indicate good reliability. All standardised item loadings should be statistically significant and exceed the minimum acceptable threshold of 0.5 [90].

Convergent validity was assessed through Average Variance Extracted (AVE) and composite reliability (CR). AVE values above 0.5 indicate that the construct explains more than half of the variance in its indicators, while CR values above 0.70 suggest acceptable composite reliability [93,94]. Discriminant validity was assessed using the Heterotrait–Monotrait Ratio (HTMT) criterion, where HTMT values below 0.85 suggest adequate discriminant validity [95]. Following the validation of the measurement model, SEM was employed to assess the relationship among latent constructs and examine the potential mediation effect.

4. Results

4.1. Confirmatory Factor Analysis (CFA)

A Confirmatory Factor Analysis (CFA) was performed to evaluate the measurement model’s fit and construct validity. The model hypothesised three latent factors: PEU (four items), PU (five items), and BI (three items). With 11 items and a sample size of 84, the item-to-sample ratio was approximately 7.6:1, which is deemed adequate [90]. Model fit was evaluated using multiple fit indices. The chi-square test of model fit was not significant, indicating an overall good fit to the data (χ²(51) = 54.326, p = 0.349). The relative chi-square (χ²/df) was below the recommended threshold of 3, indicating a good fit (χ²/df = 1.065). All additional fit indices indicated were within recommended margins (CFI = 0.992, TLI = 0.99, RMSEA = 0.028 (90%; 0.00–0.077; p = 0.717), and SRMR = 0.052), indicating a good fit. The results thus suggest that the hypothesised measurement model adequately represents the data.

Internal consistency was examined using Cronbach’s alpha (α) and omega (ω) coefficients. Both indicated acceptable reliability (≥0.7) for each latent construct PEU (α = 0.79, ω = 0.797), PLB (α = 0.861, ω = 0.856), and BI (α = 0.764, ω = 0.768). In addition, all items loaded onto their respective constructs were statistically significant (p < 0.001), with loadings ranging between 0.531 and 0.879. Convergent validity was assessed through AVE and CR, with all constructs meeting the recommended rule of thumb (AVE > 0.50 and CR > 0.70), including PEU (AVE = 0.504, CR = 0.793), PLB (AVE = 0.552, CR = 0.872), and BI (AVE = 0.521, CR = 0.762), indicating satisfactory convergent validity for all latent constructs. All HTMT values were below 0.85, indicating acceptable discriminant validity. The CFA results suggest that the measurement model demonstrated a good model fit and satisfactory convergent and discriminant validity, providing a solid basis for SEM analysis (

Table 2. Confirmatory Factor Analysis and model fit.

Latent Construct	Items	SFL	α	AVE	CR	Model Fit Indices and HTMT
Perceived Ease of Use (PEU)	PEU 1	0.619	0.790	0.504	0.793	χ2(51) = 54.326, p = 0.349 χ2/df = 1.065 CFI = 0.992 TLI = 0.990 RMSEA = 0.028 SRMR = 0.052 HTMT PEU-PU = 0.474 PEU-BI = 0.435 PLB-BI = 0.742
	PEU 2	0.649
	PEU 3	0.838
	PEU 4	0.683
Perceived Usefulness (PU)	PU 1	0.531	0.855	0.552	0.872
	PU 2	0.830
	PU 3	0.879
	PU 4	0.709
	PU 5	0.821
Behavioural Intention (BI)	BI 1	0.705	0.759	0.521	0.762
	BI 2	0.757
	BI 3	0.693

).

4.2. SEM Analysis

The evaluation of the structural model exhibited a good fit, indicating that the hypothesised relationships among the PEU, PU, and BI aligned with the data (χ²/df = 1.067; CFI = 0.992; TLI = 0.989; RMSEA = 0.028; SRMR = 0.056). The model explained a large proportion of variance, as measured by the squared multiple correlations (R²), in both PU (71.8%) and BI (55.4%), suggesting moderate-to-high predictive accuracy for the constructs under investigation [96].

PU emerged as a key determinant of BI, exerting a positive and statistically significant impact (β = 0.443, p < 0.001), thus confirming H1. The direct effect of PEU on BI was weak and insignificant (β = 0.064, p = 0.610), consistent with the expectation stated in H2. However, PEU demonstrated a statistically significant positive direct effect on PU (β = 0.535, p < 0.001), thus supporting H3. The mediation pathway from PEU to BI via PU and the total effect of PEU on BI were positive and statistically significant (β = 0.375, p < 0.001 and β = 0.443, p < 0.001, respectively). The results support H4, revealing that the influence of PEU on BI was entirely channelled through PU (

Table 3. Standardised structural estimates and hypothesis testing results.

Hypothesis	Path	β	t-Value	p-Value	Supported?
H1	PU → BI	0.694	6.447	<0.001	Yes
H2	PEU → BI	0.064	0.504	0.610	Yes
H3	PEU → PU	0.535	5.479	<0.001	Yes
H4	PEU → PU → BI (indirect effect)	0.375	3.996	<0.001	Yes
	PEU → BI (total effect)	0.443	3.770	<0.001

).

5. Discussion

The present TAM-based pilot study offers valuable empirical insights into maritime engineering students’ acceptance of fully immersive VR technology for education and training. The findings reinforce the growing body of empirical evidence attesting to the robustness of the TAM and offer novel, context-specific insights into its applicability within education and training in higher maritime institutions. As such, the results provide a compelling foundation for theoretical refinement.

In alignment with the broader TAM literature [30,35,80], PU emerged in this study as a dominant predictor of students’ BI to adopt HMD-VE engine room simulators in the future, thereby empirically supporting H1. The value of path coefficient of β = 0.694 (p < 0.001) notably exceeds the average PU-BI path coefficients reported in other TAM meta-analysis studies [31,79,80] in which the average coefficients ranged between β = 0.40 and β = 0.51. Even in recent VR-related educational research (e.g., [72,81,85]), PU-BI, though significant, mostly falls below the threshold identified in the present study. While TAM studies conducted in the maritime domain, such as [39,42], eagerly remained outside of the fully immersive VR context (e.g., digital chart plotting and GPS use), they also consistently validated PU as a reliable predictor of BI.

The elevated PU-BI path coefficient observed in this pilot study may be attributed to several context-specific factors. One plausible explanation might lie in the study-specific conceptualisation of PU. Namely, the present study explicitly framed PU as students’ perception of how the HMD-VR engine room simulator contributed to active learning rather than being conceptualised in more general or abstract terms related to the VR platform’s utility (e.g., increasing learning/academic performance). By anchoring the PU in a specific, experiential, active learning process, such as engagement, self-paced learning, or knowledge comprehension, the construct was closely aligned with students’ immediate educational and training goals. Such targeted construct operationalisation likely enhanced the exemplarity power of PU in predicting BI, particularly within the practice-oriented context of maritime engineering education and training, where competency development through hands-on, active learning is paramount [49,77]. Another possible contributing factor refers to the novelty of HMD-VR technology within the context of maritime education and training at the Faculty of Maritime Studies. As a newly implemented instructional modernisation, the simulator likely evoked a sense of technological advancement and pedagogical modernisation, which may have positively influenced students’ perception of its usefulness and, in turn, strengthened their behavioural intention to adopt it in the future. However, this interpretation should be approached with caution. Namely, it is plausible that the strong effect of PU on BI may partly reflect the initial “novel effect” or “wow phenomenon” often associated with early exposure to emerging technologies [97]. Thus, to determine the sustained effect of PU on HMD-VR engine room acceptance, this assumption should be further examined in future longitudinal studies involving repeated exposure and follow-up assessments.

The findings of this study also revealed a statistically significant and positive relationship between PEU and PU, confirming H2. The observed path coefficient (β = 0.535, p < 0.001) falls within the established range reported in prior MET studies [31,79], further reinforcing the robustness of this relationship across different technological and educational contexts. As such, the findings fully align with the core assumption of the TAM, which further posits that users are more likely to perceive innovative technology or systems as applicable if they find them easy to use [27,79,84,98]. This dynamic is particularly relevant in the MET context, where hands-on learning and technical precision are fundamental; thus, the intuitiveness and fluidity of system interaction are likely to significantly contribute to users’ perceived usefulness of innovative technology [24,42]. Accordingly, these findings carry practical implications for both software developers and educators involved in designing and implementing fully immersive VR technologies in maritime engineering education and training. For developers, the results underscore the significance of prioritising usability and intuitive system interaction. For educators, emphasis should be placed on the continuous development of step-by-step tutorials and familiarisation sessions, especially during the early stages of simulator adoption.

The weak and non-significant effect of PEU on BI observed in this study (β = 0.064, p = 0.610) empirically supports H3. The notably low path coefficients observed in this study reinforced findings from prior meta-analyses [31,79], which have consistently shown that PEU often plays a limited direct role in predicting BI. Recent studies conducted in fully immersive [35,72,81] and other technologically complex VR environments [84] have also demonstrated that ease of use alone is often insufficient to drive users’ behavioural intention toward future use of VR technology [35,72]. Building on the foundational work of [27], which posited that the direct effect of PEU on BI tends to diminish with increased user experience, the present finding offers a possible context-specific explanation for the absence of a significant PEU-BI relationship. Namely, from the outset of their academic formation, sampled maritime engineering students are routinely exposed to interactions with complex technological systems (e.g., MERS and engine diagnostic systems) as mandated by MET standards. Furthermore, over 60% of participants reported prior experience using VR simulators for entertainment (i.e., gaming) purposes. Taken together, albeit with caution, these prior experiences are likely to have shaped students’ expectations regarding the complexity of human–technology interaction, fostering their anticipation of interaction intricacy when engaging with the new system. Consequently, during the HMD-VR engine room trials, students may not have regarded usability elements such as interaction simplicity or user-friendliness as critical and sufficient drivers of acceptance of the simulator for education and training per se.

However, the study findings revealed that the influence of PEU on BI remains significant, albeit indirectly via PU, thus supporting the mediated pathway proposed in H4 and reinforcing the well-documented PEU-PU-BI path reported across the TAM literature [72,78,82]. This observation is of special importance in the context of maritime education and training, where the integration of advanced VR simulation technologies, such as the HMD-VR engine room simulator, is increasingly being advocated as a future tool in enhancing student professional skills and competencies [5,6,7]. Given the high baseline familiarity of the sampled population with complex engineering technical systems and their prior experience with VR technology for entertainment purposes, simply ensuring that VR simulators are easy to use may not be sufficient to motivate their continued use. Instead, for students to accept new technology, they must perceive that such innovations offer clear educational training benefits, such as promoting active learning that contributes directly to developing their competencies and readiness for real-world maritime operations.

6. Conclusions

This context-specific pilot study offered several important contributions to advancing knowledge within maritime education, particularly through the lens of the TAM applied to fully immersive VR-based learning and training environments. By investigating the relationship between the core TAM constructs, PEU and PU, and their direct and indirect effects on BI in the context of the HMD-VR engine room simulator, the study provided a novel, theory-driven perspective of the adoption of VR-based simulation technology in modern higher education, a sector where TAM-related empirical research remains relatively scarce. Furthermore, by focusing on maritime engineering students engaged in the early stages of HMD-VR development, the study delivered first-hand, context-specific insights into distinctive pedagogical and operational demands needed for the continuous integration of advanced simulation tools to not only support competency-based maritime engineering training but also ensure alignment with international MET standards.

Despite its contribution to the current body of knowledge, this study has several limitations that should be considered. First, the research was conducted during the pilot phase of the VR laboratory, which is still under development. Consequently, though representative of early adopters, the sample may not fully reflect the broader population of maritime engineering students who will engage with the simulator once it is fully implemented in the MET programme. Future research should, therefore, be conducted after the full integration of the simulator to capture a more representative and longitudinal perspective of its acceptance. Second, due to its case study design, the research was conducted at the Faculty of Maritime Studies in Rijeka, leading to a sample that may not fully capture the diversity of the broader maritime engineering population, thus limiting the findings’ generalisability. Although the findings provide valuable and preliminary insights into how the TAM’s core constructs interact within this context-specific educational setting, future research should involve more extensive and diverse samples, ideally across multiple maritime institutions equipped with similar HMD-VR technologies, to validate and extend the present results. The third limitation concerns gender imbalance within the sample. This sampling outcome was not intentional but reflects the existing gender structure of the maritime engineering programme at the Faculty of Maritime Studies, University of Rijeka, where female student representation is exceptionally low. According to the official faculty records, only 0.8%, or two female students, are currently enrolled in the maritime engineering programme. Gender imbalance is a well-documented issue not only in maritime and associated engineering education and training studies [16,60,86,99] but also across distinct domains of the maritime industry sectors, where seafaring roles have historically been male-oriented [100]. Future research should thus examine to what extent gender moderates the acceptance of VR-based simulation technology and, most importantly, explore structural and cultural factors that contribute to the persistent gender imbalance in the maritime education and training sector. Given that the study focused specifically on core TAM constructs, future research should employ the full analytical potential of SEM to explore further additional affective and attitudinal constructs, such as perceived immersion, cognitive load, and attitudes toward usage, socio-demographic factors, and prior experience, to deepen insights into the acceptance of HMD-VR simulation-based technology. Finally, future research should also consider broadening the theoretical scope by integrating complementary models such as the Unified Theory of Acceptance and Use of Technology (UTAUT) [101,102] and the Diffusion of Innovations theory [103], as they could offer additional explanatory power by integrating social influence, facilitation conditions, and innovation characteristics into the analysis. Such a multidimensional approach would help capture the complex factors influencing VR technology acceptance in maritime engineering education and training contexts.

In conclusion, the study findings offer valuable guidance to educators and software designers aiming to develop and implement fully immersive HMD-VR-based simulation platforms in maritime engineering higher education. As MET institutions increasingly advocate for the application of immersive technologies within existing curricula, a user-centred, evidence-based approach grounded in the TAM framework appears essential, as applied in the current study.