The Fraunhofer IVI Accident Prevention School (FAPS) was developed to close the existing knowledge gap in road safety education for young people aged 13 to 15 (Erbsmehl et al., 2019). The aim of the training is to sensitise pupils to critical traffic situations, raise their awareness of accident risks and promote their ability to act in a proactive, safe and anticipatory manner in road traffic. The training also teaches pupils that even correct behaviour on their part cannot always prevent accidents. It is important to note that accidents can still occur as a result of the misconduct of other road users. Moreover, the FAPS programme encourages young people to engage with real traffic situations, thereby enabling them to transfer the skills they have acquired to everyday life.

The training is based on a modular structure that combines sequential learning steps: The introductory modules 0 and 1 aim to raise awareness by encouraging pupils to report on their own experiences, identify critical points on their own way to school and assess the frequency of accidents in their school surroundings. In modules 2 and 3, real accident statistics are evaluated and accident hotspots in the school area are analysed with the help of anonymised police recorded accident reports. In module 4, participants use software to reconstruct traffic accidents statically on tablets. In this process, they analyse the causes of accidents, consider the perspectives of all involved accident parties and examine visibility conditions such as viewing angles, view obstacles and blind spots. The change of perspective enables them to understand critical situations realistically and derive alternative strategies for safe behaviour in road traffic.

Before the start of the SOTERIA Horizon Europe project first virtual reality scenarios were available. The aim within the SOTERIA project was to develop an educational concept based on these scenarios that would enable the integration of a VR module (module 5) into the existing training programme. Module 5 expands on the existing training programme with the use of VR technology and deepens the knowledge gained in module 4 through a fully immersive learning experience. Based on official accident statistics and detailed accident analyses, particularly serious and frequently occurring accident scenarios involving pedestrians and cyclists were identified and reconstructed as real accidents in a dynamic 3D environment. Figure 2 shows sample screenshots of the VR module.

In contrast to the tablet-based reconstruction in module 4, pupils experience the accident situations from a first-person perspective of all involved parties in real time. Obstructed views, angles of vision and blind spots become physically and emotionally tangible, as learners actively move around in the virtual space and can take on the perspectives. This leads to a particularly intense understanding of the causes of accidents, risky behaviour and the potentially fatal consequences of critical traffic situations.

The study was carried out using a control group design. In Munich, the training was conducted in its previous form up to module 4, while in Saxony, the new educational concept, including the VR module (module 5), was implemented. This structure not only enabled age-appropriate teaching of road safety skills but also allows a direct comparability between the conventional tablet-based training and the extended VR-based method. Both variants were evaluated as part of the SOTERIA project to investigate the effect of the immersive learning environment on the safety awareness and anticipatory behaviour of young people.

The VR module was developed using the Unreal Engine, where all accident scenarios were modelled, animated, and rendered as 360° immersive video sequences. These scenes are delivered through a WebXR-based 360° web application (“JARVIS 360 WebXR Player”), which runs directly in the integrated browser of standalone VR headsets. As the system is based on standard WebXR technology, it is hardware-independent and can be used across different VR devices without architectural changes. Originally tested on Oculus Quest headsets, the application is now operated on Pico VR devices.

Immediately after completing the training at their respective schools, the participating pupils received standardised questionnaires. The questionnaires used in the city of Munich and in Saxony contained an identical core set of questions, which allowed a direct comparison of perception, acceptance and learning effectiveness between the tablet-based and VR-based training. Additionally, the questionnaire administered in Saxony incorporated specific questions concerning the VR module, with the objective of capturing the participants' experiences with immersive technology and its impact on their safety awareness. As the evaluation followed a post-test-only design and no pre-test measurements were conducted, changes in knowledge, attitudes, or safety awareness could not be assessed directly.

The common questions in both questionnaires consisted of different sections focussing on the perception of the learning method, for example whether it was perceived as varied and appealing, as well as on the participants' willingness to recommend the prevention programme to others. In addition, the questionnaire included a section on the contents and aspects of the training that were particularly memorable, and which allowed for the drawing of conclusions regarding the respondents' own safety behaviour and their attitude towards road safety. Moreover, the questionnaire was designed to undertake a comprehensive evaluation of the programme, addressing both perceived strengths and weaknesses, and proposals for enhancement.

The specific questionnaire section on the VR component focused on the independent use of VR glasses and software, the comprehensibility of the virtual representations, the support in understanding the risks of motorised traffic for pedestrians and cyclists and the appropriateness of the duration of VR use.

Contextual information was also collected to identify possible influencing factors, including the location of the school (urban or rural) and the type of secondary school (grammar or high school). This enabled an investigation of the extent to which school conditions may influence the perception, acceptance and impact of the training variants.

The questionnaire was developed based on established evaluation criteria commonly used in educational media assessment. To ensure content validity, methodological approaches from schoolbook research were considered, including the Bielefelder Raster (Laubig et al., 1986), the Wiener Raster (Bamberger, 1998), and more recent tools like the Levanto Tool (Wirthensohn, 2012), which include criteria such as relevance, comprehensibility, didactic quality, effectiveness, transferability, and user appropriateness. These frameworks provided a meta-level rationale for selecting question types and evaluating dimensions such as usability, perceived learning outcomes, acceptance and behavioural relevance.

Furthermore, concepts from project evaluation – particularly relevance, effectiveness, efficiency, and process evaluation – were incorporated to justify the inclusion of items related to learning impact, perceived behavioural change, and situational awareness. This methodological foundation ensured that our questionnaire captures both descriptive and evaluative aspects relevant to road safety education.

All questionnaire items used a 4-point Likert scale without a neutral midpoint (“agree”, “rather agree”, “rather disagree”, “disagree”). For some items, an additional “not applicable” option was provided. The format was chosen to encourage clear directional responses and follows common practice in educational media evaluation.

The questionnaire was provided in printed form and filled out by the pupils directly after the accident prevention session during regular class time. As the questionnaire was administered on paper, no digital platform or software tool was involved.

A total of 247 participants were invited to complete the questionnaire as part of the study. Of the 93 participants from the city of Munich, 27 (29.03%) were from grammar schools (“Gymnasium”) in municipal areas, and 66 (70.97%) were from high schools (“Oberschule”) in rural areas. In comparison, in the Saxon accident prevention school, 110 of the 154 participants who utilised the VR module were enrolled from urban grammar schools (71.43%), while 44 participants (28.57%) were recruited from rural ones. To evaluate the effectiveness of digital media in accident prevention, a comparison was carried out between the tablet-based results in Munich and the extended VR approach within Saxony.

The data analysis consisted of a simple Chi-Square test as method to determine if there was a significant variation in responses between the two test groups. Should a Chi-Square value be higher than the specified significance level, it can be deduced that the observed data varies significantly from the expected data. The significance level was determined to be α=0.05, which corresponds to a probability of 5%. If the calculated p-value exceeds the significance level (p>α), it can be deduced that the result is not significant, and the null hypothesis is upheld. This finding indicates that VR users in Saxony attain equivalent learning outcomes to pupils in Munich who do not participate in the VR module. The software utilised for this purpose was MS Excel. The following analysis will examine the specific results of the VR module.

The ensuing figures offer a comprehensive evaluation of the specific components of the accident prevention school, accompanied by a meticulous comparison of the outcomes from the city of Munich and the federal state Saxony. The findings demonstrated a high level of consistency between both regions, with pupils expressing a generally positive approval of the course, irrespective of the additional VR module. The initial aspect requested and analysed pertained to the overall satisfaction levels of the course, both between the two locations and two types of schools. The difference is evident when analysing the participants' satisfaction with the learning methodology employed in the accident prevention programme (see Figure 3). In the city of Munich, as illustrated on the left side of the diagram, with 77.78% of respondents enrolled in urban grammar schools and 81.82% in rural high schools, the proportion of respondents who rate the course as 'good' or 'rather good' is almost equal. However, a discrepancy was observed in the ratings given by respondents attending school in rural and urban areas. In Munich, where the course was found to be more favourably evaluated in rural schools, 30.30% of individuals attending school in rural areas evaluated the subject as "good", while this figure was halved, at 14.81%, among their urban counterparts (p=0.654). This outcome was expected due to the absence of statistical significance (p > α). This phenomenon stands in contrast to the situation in Saxony, where the proportion of respondents who expressed a 'good' rating was notably higher in urban areas (70.91%) compared to rural areas (40.91%). In this case, the difference was found to be significant (p≤α with p=0.035).

In the following the data from Munich, excluding the VR module, is compared with the data from Saxony, including the VR module, regarding the impact of the accident prevention schooling. The questionnaire's key findings indicate that the VR module significantly increased the pupils’ satisfaction with the programme, as illustrated on the left side in Figure 4. In Munich, 36.56% of pupils provided a positive evaluation of the accident prevention school, with an additional 35.48% rating it as "rather good". In contrast, the general rating in Saxony is higher, with 93% of Saxon pupils expressing satisfaction following the completion of the VR module. The implementation of VR has been developed to enhance the programme's appeal among pupils. This is evidenced by the outcomes of the evaluation in terms of the pupils' recommendations of the course, as presented in Figure 4 on the right. In the city of Munich, 53.76% of respondents stated that they would or rather would recommend the accident prevention school to a friend. In contrast, participants in Saxony reported a significantly higher level of recommendation in the study, with more than 80% of respondents (p≤0.05).

These are key positive aspects, as is the memorability of the course, which is also beneficial for the VR module, as illustrated on the left in Figure 5. The questionnaire administered in Munich further revealed that 16.13% of respondents expressed the opinion that the employed pedagogical approach would leave a lasting impression. The outcomes of the questionnaire in Saxony revealed that this figure was almost double, with 30.52% of participants holding the same view (p≤0.05). The project work had a significantly varying degree of influence on the attitudes towards road safety behaviour of participants in Munich and Saxony. The study revealed that in the absence of the VR module in Munich, only 12.90% of participants exhibited a change in attitude, as demonstrated in Figure 5 on the right. In Saxony, where the accident prevention school was employed with VR, this figure increased to 31.82% (p≤0.05).

The following question is presented in summary form, without the inclusion of explicit numerical data or graphical representations. The feedback received identifies the VR module as a contributing factor to the establishment of a more sustainable awareness of road safety. Next to a higher rate of recommendation, a greater proportion of respondents (>75%) stated agreement and rather agreement that more attention will be paid to safe road behaviour from Saxon participants than from Munich ones, both in terms of the behaviour of one's own and other road users. While approximately 33% of respondents in both regions expressed a certain degree of agreement, only 18% of Munich pupils indicated full willingness to dedicate more attention to the issue of road safety. This figure is more than half lower than the percentage of respondents in Saxony who expressed a similar degree of agreement.

As previously mentioned, the initial questionnaire themes of this segment are also evidenced solely by quantitative data. It is important to note that the participants from Saxony highlighted the significant impact of the VR module on the aforementioned changes in attitudes towards road safety. This is in accordance with the contribution of the VR module to attitude change in road safety behaviour. The results of the questionnaire section, which was exclusively concerned with the VR module, indicated that approximately half of the pupils (45.45%) participating in Saxony attested to the VR module having a role in modifying their attitudes, with an additional 28% expressing a rather positive contribution. The VR application's design was intuitively structured, enabling pupils to engage with the VR module autonomously. This finding was further substantiated by the responses to the administered questionnaire, with over 70% of respondents expressing complete approval and only a small percentage rejecting or partially rejecting the application. The consensus among most test subjects was that the VR application contributed to an enhanced comprehension of traffic and hazardous situations.

The question of whether new insights into the visibility of pedestrians and cyclists were gained, and whether the VR application provided significant support in this regard, can be answered by examining Figure 6. As illustrated on the left-hand side, 53.25% of Saxon participants obtained novel insights into the visibility, with a further 25.97% stating they rather did. Additionally, 38.96% of pupils reported that VR use had been beneficial, with an additional 38.96% stating that it had rather enhanced their perception of pedestrians and cyclists (see right-hand side).

The fact that most participants indicated a desire for additional time for the VR module during the accident prevention schooling also serves to emphasise the favourable opinion of the implementation of the VR module in accident prevention training, irrespective of region or school level.

It is crucial to acknowledge the limitations of the study. Firstly, the comparison between Munich and Saxony is not based on a controlled experimental design. Rather, it is based on two distinct regional implementations. Consequently, any observed differences in outcomes cannot be attributed solely to the presence of the VR module. Other contextual variables, such as teaching quality, school type and location, and socio-demographic factors, may also have influenced the results. These are just some of the possible influencing factors. Even the educational system differs from state to state in Germany.

Moreover, the study predominantly relies on self-reported data, which is subject to potential biases, such as social desirability or recall bias. In addition, the evaluation followed a post-test-only design, as no pre-test measurements were conducted. Therefore, changes in knowledge, attitudes, or safety awareness cannot be assessed directly. While subjective perceptions are important indicators of programme acceptance, they do not necessarily reflect actual behavioural changes in real traffic situations.

Attitudinal change and memorability were only measured in the short term, immediately after programme participation. The long-term effects, which are relevant to accident prevention, were not examined. Furthermore, participants expressed a desire for more time, suggesting that the duration of exposure to the VR module was limited. This may have limited the intervention's potential impact, which partly explains the moderate levels of reported attitude change.

These limitations highlight several avenues for future research. Future studies should employ more rigorous experimental designs, including pre- and post-test measurements and, where feasible, control groups, to better isolate the specific effects of VR-based interventions and reduce potential confounding factors. Further research would contribute to more robust conclusions and help to assess the durability of learning outcomes and behavioural changes over time. It would be particularly valuable to investigate whether the observed improvements in awareness and attitudes towards safer behaviour translate into safer behaviour in real-world traffic environments.

Given the disparities observed between regions and school types, future research should also examine the influence of contextual factors more systematically. These include local traffic conditions, differences between urban and rural environments, school characteristics, and the role of environmental and socio-cultural variables in shaping the effectiveness of VR-based education. Such research would contribute to the development of more tailored and equitable interventions.

The findings of this study have several implications for educational policy and practice. The evident advantages of VR suggest that the incorporation of immersive technologies into accident prevention programmes has the potential to significantly enhance their effectiveness and acceptance among pupils. Consequently, policymakers should consider the implementation of VR modules on a broader scale within school courses.

In view of the disparities observed between regions and school types, it is imperative to ensure that such innovations are implemented in a manner adapted to the specific characteristics of the school environment. This may require the adaptation of content to suit local traffic conditions, including the differentiation between urban and rural areas. Additional support should be provided for schools with limited technological infrastructure.

The favourable reception of the VR module, coupled with the expressed desire for increased usage, indicates a strong willingness among pupils to engage with digital learning tools. It is imperative to invest in teacher training and technical resources if the full potential of these technologies is to be realised. Further investigation into the optimal exposure duration and instructional design within VR environments is recommended to maximise learning outcomes. Despite its potential, it is important to note that virtual reality should not be considered a replacement for conventional teaching methods, but rather a valuable supplement that can improve experiential learning and risk awareness (Makransky et al., 2019). It has been identified that a combination of theoretical instruction and immersive simulation is a potentially beneficial approach to promote sustainable behavioural change in the context of road safety (Merchant et al., 2014). Studies within educational psychology and human computer interaction have repeatedly demonstrated that VR environments have the capacity to enhance motivation and emotional involvement, which are critical predictors of success in this domain (Bart et al., 2008; Đorđević, 2024; Merchant et al., 2014; Meyer et al., 2019; Radianti et al., 2020).