Embodied virtual reality ethnography in intangible cultural heritage preservation: community co-creation of the Qiang New Year festival and its impact on cultural knowledge and attitudes

1 Introduction

Under the dual pressures of cultural homogenization and intergenerational detachment, the preservation of intangible cultural heritage (ICH) has long transcended mere archival documentation. UNESCO’s inscription of the Qiang New Year festival on the List of Intangible Cultural Heritage in Need of Urgent Safeguarding was motivated not by a lack of documentary records, but by the rupture of emotional bonds between younger generations and these traditions (UNESCO, 2009). Urbanization, the displacement of local knowledge systems by formal education, and the pervasive reach of mainstream popular culture have collectively estranged Qiang youth from their ancestral heritage. The 2008 Wenchuan earthquake, which devastated numerous Qiang villages, further inflicted catastrophic damage upon the material substratum of cultural continuity. This predicament raises a specific and open question: When a cultural tradition faces the threat of rupture, can technological means leave an authentic imprint upon individuals who have never experienced the ritual firsthand? This is not merely knowledge about the tradition, but a genuine concern for its fate. The present study is undertaken precisely within this problematic.

While contemporary heritage preservation has developed internal protocols, digital archives, high-definition documentaries, and online databases have effectively addressed the archival imperative, they have struggled to reconstruct emotional connection. They are, in essence, information transmission systems. Watching a video of a Qiang sacrificial rite is fundamentally distinct from beating a drum beside a bonfire and feeling the rhythm of the ritual course through one’s body. Immersive virtual reality (VR) offers a technological means to bridge this chasm. Unlike conventional digital media, VR can engender a sense of embodied presence, transforming users from passive spectators into situated participants (Slater, 2009). Recent scholarship has begun to explore the application of VR in cultural heritage communication (Liu et al., 2025; Zhao et al., 2025; Zhang et al., 2025; Makransky and Lilleholt, 2018; Wang and Ametefe, 2026), with preliminary evidence indicating that VR can enhance both cultural knowledge acquisition and the quality of user experience. Nevertheless, two notable lacunae persist in the extant literature. First, the preponderance of studies has focused on knowledge acquisition and technological usability, with comparatively little systematic investigation into how VR experiences may activate cultural identification at an affective level: that is, enabling cultural outsiders not merely to learn about a culture, but to develop a genuine emotional affiliation with it. Second, even fewer studies have examined whether such emotional affiliation can translate into tangible heritage preservation intentions in the real world; in particular, there is a conspicuous absence of efforts to integrate embodied experience, cultural identification, and preservation willingness within a unified structural framework for empirical validation. Furthermore, extant VR heritage research has largely prioritized technical implementation, with relatively few studies deeply embedding source community co-design within the development pipeline (Liu et al., 2025; Zhang et al., 2025). This omission constitutes a methodological gap with respect to both cultural authenticity and community trust.

Taking the Qiang New Year celebration in Mao County, Sichuan Province, as a case study, the present research adopts an explanatory sequential mixed-methods design. Building upon 8 months of ethnographic fieldwork and community-based participatory co-creation, we developed an embodied VR ethnographic experience system. Through a quasi-experimental evaluation (N = 200; VR condition versus content-equivalent slideshow control) and follow-up interviews, the study addresses two central research questions: (1) can embodied VR experiences yield measurable knowledge and attitudinal gains among cultural outsiders? (2) Do immersive presence and perceived cultural authenticity influence heritage support intentions via the mediating mechanism of cultural identification, independently of technology acceptance?

The principal contributions of this study are twofold. At the methodological level, the adoption of a content-equivalent control design allows for a relatively clean separation of the medium’s immersive effect from the content-exposure effect, complemented by the development of a knowledge assessment instrument validated by Qiang cultural experts. At the theoretical level, by integrating presence theory and the cultural identification framework with the technology acceptance model (TAM), we propose the embodied cultural presence model (ECPM) as a provisional analytical lens and empirically test the mediating pathway of cultural identification via Partial least squares structural equation modeling (PLS-SEM). Additionally, the comprehensive documentation of the community co-design process furnishes a replicable methodological pathway for the digital transmission of other endangered intangible cultural heritage, although this aspect is situated closer to practice documentation than to an independent scholarly contribution.

2 Literature review and theoretical framework

Conventional preservation modalities, such as archival storage, exhibitions, and textual records, are inherently limited in conveying the lived, experiential dimensions of cultural traditions, prompting the field to turn to interactive media. A comprehensive survey by Bekele et al. (2018) documents the widespread application of VR and AR in cultural heritage education. Liu et al. (2025) evaluated user and learning experiences across three heritage VR applications using a content-equivalent multi-condition design, substantiating the advantages of immersive technologies in knowledge acquisition and emotional engagement and providing a direct methodological referent for the present study. Zhang et al. (2025) developed the mixed-reality system KiteMR to preserve traditional kite craftsmanship, underscoring the critical role of community participation in cultural authenticity. Zhao et al. (2025) systematically reviewed VR heritage applications and advanced design recommendations for immersive heritage experiences. However, many existing projects prioritize visual fidelity over systematic evaluation of learning outcomes or attitudinal impacts (Makransky and Lilleholt, 2018).

2.1 Limitations of existing frameworks: the conceptual boundaries of TAM

TAM (Davis, 1989) explains behavioral intention via perceived ease of use (PEOU) and perceived usefulness (PU), with perceived enjoyment (PE) subsequently introduced for hedonic systems (Davis et al., 1992). However, directly applying TAM to cultural heritage VR encounters a structural conceptual gap. TAM’s core dependent variable, behavioral intention to use the technology, aligns with productivity tools but misaligns with heritage VR’s central concern: whether an experience enables users to forge emotional bonds with an endangered culture and consequently act to perpetuate it. TAM measures the person–technology relationship; heritage VR research must simultaneously measure the person–culture relationship. Applying TAM in isolation systematically occludes cultural identification mechanisms. This is not a limitation inherent to TAM, but a context misalignment. Gong et al. (2025) similarly extended TAM for museum VR, identifying immersion and content quality as antecedents of perceived usefulness and enjoyment, further underscoring the necessity of incorporating experiential dimensions. The present study therefore retains TAM’s technology acceptance pathway while introducing three complementary theoretical perspectives that render the cultural identification pathway explicit as an independent mechanism.

2.2 Theoretical integration: the embodied cultural presence model (ECPM)

ECPM integrates three theoretical foundations selected for their explanatory relevance to heritage VR rather than comprehensive theoretical coverage.

Presence theory (Slater, 2009) provides a mechanistic account of how immersive experiences engender psychological consequences. Slater distinguishes immersion (objective sensory stimulation intensity) from presence (subjective “being there”): in high-immersion environments, the brain processes virtual events as if they were real, activating emotional responses analogous to those of experience. ECPM positions immersive presence (IP) and perceived cultural authenticity (PCA), which is the users' subjective judgment that VR content faithfully represents authentic traditions, at the front end of the cultural identification pathway. Empirical studies confirm presence as a key mediator linking immersive experience quality to visitor engagement and behavioral intentions (Lee et al., 2020; Tussyadiah et al., 2018). Nam et al. (2023) further distinguish object-based from existential authenticity, aligning with the PCA construct in ECPM.

Experiential learning theory (Kolb, 1984) provides a mechanistic rationale for how embodied participation yields knowledge gains. Kolb’s experiential learning cycle (concrete experience → reflective observation → abstract conceptualization → active experimentation) posits that meaning-making from personally participating in an event is more enduring than reading or viewing descriptions, simultaneously activating cognitive, affective, and somatic channels. Manipulating ritual artifacts, drumming, and performing sacrificial gestures before a virtual altar all lie closer to the “concrete experience” pole than does viewing a slideshow. ECPM thus anticipates enhanced cognitive (knowledge acquisition) and affective (cultural identification activation) outcomes from embodied participation.

Social identity theory (Tajfel and Turner, 1979) holds that self-concept is partially constituted by group membership, which can activate affective investments and behavioral predispositions. The present study extends this logic cross-culturally, operationalizing cultural identification (CI) as perceived closeness, belonging, and affective resonance toward a cultural group following the experience, even among cultural outsiders. ECPM’s core proposition resides here: a carefully designed embodied VR experience, by enabling active ritual participation, interaction with virtual community members, and multisensory immersion in the culture’s cadence, may activate affective structures akin to cultural belonging in outsiders. Once activated, this affective bond is hypothesized to mediate the translation of immersive experience into behavioral intentions supporting heritage preservation, which constitutes ECPM’s fundamental theoretical divergence from standard TAM.

ECPM is a provisional conceptual integration from a single cultural case rather than a fully elaborated theoretical system; its applicability across cultures and heritage typologies awaits further empirical scrutiny.

2.3 Structure of ECPM: two parallel pathways

ECPM comprises two analytically distinguishable yet experientially co-occurring influence pathways (see Figure 1).

The technology acceptance pathway adheres to classical TAM logic: PEOU → PU → BI-VR, with PE influencing both perceived usefulness and behavioral intention. This pathway determines users' acceptance of the VR technology itself, which is the foundational condition for heritage VR to exert cultural effects.

The cultural identification pathway represents ECPM’s core extension: IP + PCA → CI → Heritage support intention (HSI). Cultural identification is conceptualized not as a byproduct of technology acceptance but as an independent outcome jointly determined by embodied experience quality and perceived cultural authenticity, ultimately linked to heritage preservation behavioral intentions. These two pathways may dissociate: a user may exhibit high technology acceptance with little cultural resonance, or vice versa. Comprehensive heritage VR evaluation therefore necessitates simultaneous consideration of both pathways.

Model evaluation adheres to the PLS-SEM criteria established by Hair et al. (2019), with standardized root mean square residual (SRMR) as the primary fit index, supplemented by composite reliability (CR), AVE, and HTMT discriminant validity reporting. Path coefficients reflect statistical associations rather than causal relations; conclusions are framed with appropriate circumspection.

2.4 Research hypotheses

Grounded in the foregoing framework, the present study advances the following four hypotheses.

H1: Perceived ease of use (PEOU) and perceived enjoyment (PE) positively predict perceived usefulness (PU) and behavioral intention to use VR (BI-VR), consistent with the TAM pathway.

H2: Immersive presence (IP) and perceived cultural authenticity (PCA) positively predict cultural identification (CI) and knowledge acquisition, reflecting the front-end effects of the cultural identification pathway.

H3: Perceived enjoyment (PE) positively predicts cultural identification (CI). The theoretical warrant for this hypothesis derives from the proposition that states of affective engagement lower cognitive defenses and increase openness to cultural content (Fredrickson, 2001; broaden-and-build theory). When an experience is perceived as pleasurable rather than task-oriented, cultural symbols are more likely to elicit identificatory affective responses rather than critical, spectatorial distance.

H4: Cultural identification (CI) positively predicts heritage support intentions (HSI). The cultural affinity activated through the VR experience is expected to be associated with behavioral intentions to support heritage preservation activities.

H2, H4 jointly constitute the core test of ECPM’s cultural identification pathway. Should both receive empirical support, this would provide preliminary evidence for an independent associative mechanism linking embodied immersive experience to heritage preservation behavioral intentions and being mediated by cultural identification. Methodologically, this association is examined via the indirect effect test in PLS-SEM and is not construed as implying strict causal inference.

3 Materials and methods

This study employed an explanatory sequential mixed-methods design (Leech and Onwuegbuzie, 2009): a quantitative quasi-experiment was conducted first, followed by follow-up interviews intended to explain and deepen the interpretation of the quantitative findings. The research proceeded in three phases: ethnographic engagement and community co-design, VR system development, and experimental evaluation. Iterative collaboration with the target community was maintained throughout all phases to ensure cultural authenticity.

3.1 Ethics approval

This study involved human subjects research (N = 200 participants) and was conducted in accordance with the Declaration of Helsinki. Ethical approval was granted by the Human Research Ethics Committee (HREC/JEPeM) of Universiti Sains Malaysia (USM), approval number USM/JEPeM/PP/25050423, issued on August 15, 2025. Written informed consent was obtained from all participants prior to their involvement.

3.2 Phase 1: ethnographic engagement and community co-design

The research team undertook an intensive eight-month fieldwork immersion within the Qiang community of Mao County, Sichuan Province, observing and documenting the preparations for and ceremonial activities of the Qiang New Year festival. With the informed consent of community leaders and participants, we integrated ourselves into the daily rhythms of village life to establish rapport and conducted focused observations during the Qiang New Year celebrations in November 2022. Extensive field notes were recorded concerning key rituals, including the Mountain Sacrifice (Jishan), an early morning offering of wine and meat to the mountain deities; the Sheepskin Drum Dance, a collective dance performed beside a bonfire at dusk, accompanied by traditional drumming and chanting (Figure 2); and the Torch Procession, a nocturnal circumambulation of dwellings by residents bearing torches to ward off malevolent spirits and invoke blessings.

In addition, in-depth semi-structured interviews were conducted with 15 cultural knowledge bearers using the local language and subsequently translating their responses into Mandarin. Interviewees included three Shibi priests, four elders serving as traditional custodians, three village officials engaged in heritage preservation, and five youth representatives. Interviews centered on three core dimensions: the symbolic significance of each ritual, the interviewee’s personal affective connection to the festival, and perspectives on the challenges of intergenerational transmission. Representative questions included: “What is the single most important thing you would wish young people to understand about the Qiang New Year?” (directed at elders) and open-ended inquiries regarding attitudes toward the festival and ideas for its revitalization (directed at youth). Concurrently, the research team collected audio recordings of traditional songs, ritual prayers, and chants, as well as photographs and sketches of traditional attire, ceremonial paraphernalia, and sacred sites, constituting primary source material for VR content development.

Following preliminary analysis of the ethnographic data, two community co-design workshops were convened to collaboratively develop the virtual reality content. During the first workshop, documentary materials were shared with villagers to verify their accuracy and to stimulate reflective narration. Community members participated actively: they corrected minor inaccuracies, supplied contextual clarifications, and articulated the cultural values underpinning each custom. Elders emphasized that any representation must convey the significance of each ritual, not merely “how it is performed,” underscoring the necessity of integrating cultural meaning and oral history rather than simply depicting actions. In the second workshop, the concept of a virtual reality recreation of the festival was introduced. Utilizing accessible visual aids, including 360-degree panoramic photographs of the village and a rudimentary VR demonstration via Google Earth VR, a participatory brainstorming session was facilitated to explore what a virtual Qiang New Year experience ought to encompass. Participants exhibited considerable enthusiasm. One young participant remarked, “Include our New Year songs and dances; make it feel celebratory, not like an exhibit in a museum.” Elders reiterated their earlier guidance: “Make sure young people understand why we perform each ritual, not just the movements themselves.” These contributions directly informed three design principles that subsequently shaped development: embedding cultural narration throughout the experience, fostering a festive atmosphere, and prioritizing the transmission of meaning over visual spectacle. The workshops simultaneously cultivated a sense of shared ownership over the digital representation; participants felt that the manner of digitizing their culture was something with which they were personally involved, rather than an external imposition. A five-member local advisory group (including one Shibi priest and two schoolteachers) was subsequently constituted to review early VR prototypes and provide ongoing feedback.

3.3 Phase 2: virtual reality system development

Guided by ethnographic insights, the VR application was developed using the Unity 3D game engine and the XR Interaction Toolkit. Development and testing were conducted using the PICO Neo 3 VR headset, equipped with hand controllers and room-scale tracking. A tethered PC connection was employed to achieve high-fidelity graphics and maintain stable laboratory performance. The final VR experience, lasting approximately 20 min, simulates the full-day progression of the Qiang New Year from dawn to night. To maximize user engagement, the experience was structured as an interactive narrative in which the user assumes an active role as an assistant to the Shibi priest. This narrative device both provides the user with a mentor figure within the virtual world and legitimizes the user’s active participation. The experience unfolds across four chronologically sequenced scenes: dawn (mountain sacrifice), morning (village preparations), evening (Sheepskin Drum Dance), and night (torch procession and blessing) (See Table 1).

TABLE 1. Core elements of the Qiang Year ritual and their implementation in the VR experience.

3.4 Scene 1: mountain sacrifice (dawn)

The experience commences at dawn, situated at a hillside altar overlooking a virtual Qiang village. A Shibi priest non-player character (NPC) guides the user through the ritual of offering sacrifices to the mountain deities. Using the hand controllers, the user picks up virtual wine and meat and places them into a fire pit. Physics-based object interaction enhances realism. The ritual is accompanied by authentic chanting recorded from a Shibi priest during fieldwork, with haptic feedback signaling the moment the offering is accepted: the virtual flames surge upward. The scene culminates in a virtual sunrise, rewarding the user for completing the ritual (Figure 3).

3.5 Scene 2: village preparations (morning)

Later in the morning, the user arrives at the village courtyard, where villagers are engaged in festival preparations. The user may freely explore via teleportation movement. Interactive tasks, such as hanging prayer flags or lighting incense before a household shrine, serve both to familiarize the user with VR controls and to convey cultural knowledge through NPC dialogue. For instance, one NPC explains, “We hang these flags so that our prayers may travel with the wind.” This scene functions as a semi-tutorial, employing environmental storytelling in lieu of direct instruction.

3.6 Scene 3: Sheepskin Drum Dance (evening)

As night falls, a bonfire is lit in the village square, and NPC villagers gather for the central Sheepskin Drum Dance. The user plays a sheepskin drum alongside the NPCs, guided by a rhythm mini-game mechanism wherein visual cues are synchronized with authentic field-recorded Qiang folk music. This ensures rhythmic accuracy and affords the user a palpable sense of communal participation. Midway through the dance sequence, the user’s perspective briefly shifts to a third-person view, revealing their own avatar dancing within the circle, which is a deliberate design choice intended to enhance the user’s sense of community belonging.

3.7 Scene 4: torch procession and blessing (night)

In the final scene, the user holds a virtual torch and joins a torch-lit procession circumambulating the village. At key junctures, the Shibi narrator shares legends and cultural exegesis, supplemented by visual apparitions of spirits. Toward the conclusion, a voice recognition system detects the Qiang blessing phrase, “Suoluo.” Upon successful recognition, the ritual concludes with virtual fireworks and cheering NPCs, one of whom addresses the user: “You are now a member of our extended family,” thereby conveying a sense of inclusion and cultural belonging.

3.8 Scene 5: Cave Quest (underground exploration)

Following the torch procession, the experience concludes with an optional exploratory segment in which the user descends into an underground chamber beneath the village. Guided by narrative cues from the Shibi priest, the user is tasked with locating a chest containing a symbolic artifact linked to the community’s heritage. The scene is structured as a branching task sequence (Figure 4): the user locates the chest, engages in NPC dialogue, opens the chest, ignites a torch to illuminate the surroundings, and retrieves the artifact. Interactive elements include a grab-interactable chest, a teleport arc for navigation, and a torch-based light source that reveals details of the cavern environment (Figures 3d–f, 5). This segment was designed to extend the sense of embodied exploration and discovery established in earlier scenes, reinforcing the user’s role as an active participant in the narrative.

3.9 Community iterative review

The development process adhered to an iterative prototyping methodology. Once functional prototypes were completed, two community members were invited to physically experience the application and provide feedback. Major revisions based on this feedback included adjusting the timbre of the drum sound to more closely approximate the sonorous quality of an authentic sheepskin drum; enriching the diversity of NPCs to reflect the equitable participation of both men and women in actual rituals; and increasing the variety of traditional attire depicted. The final version was confirmed by community reviewers as both authentic and emotionally resonant.

4 Phase 3: experimental evaluation

4.1 Experimental design and conditions

VR condition (n = 100): Participants wore the PICO Neo 3 headset and completed the 20-min immersive VR experience of the Qiang New Year.

Control condition (n = 100): Participants viewed a 20-min slideshow presentation (static images + voice-over narration + on-screen text) in the same laboratory environment. The slideshow covered content entirely equivalent to the VR experience across five knowledge dimensions: ritual sequence and structure; symbolic significance of specific rituals; material culture and sacred objects; the role of the Shibi priest; and the communal functions of the festival.

The content-equivalent design, drawing on the methodological approach of Liu et al. (2025), aimed to render the medium of presentation the primary variable differentiating the two conditions, thereby enabling an assessment of the relative contribution of immersive, embodied experience. It must be noted that this design does not imply that all experiential dimensions between conditions are comparable; the operational agency, autonomous decision-making, and temporal pacing inherent to the VR condition are themselves components of the medium difference, rather than extraneous “noise” to be controlled away. Consequently, the inferences drawn in this study are situated at the level of the “composite effect of the VR presentation mode,” rather than a granular decomposition of the specific sub-components of the medium effect. Future research incorporating a 2D video condition or a 360° VR condition without embodied interaction would facilitate a more refined attribution of the specific sources of the medium effect.

4.2 Participants

Two hundred participants were recruited from a local university and the surrounding community. Inclusion criteria stipulated: no Qiang ethnic background and no prior in-depth familiarity with Qiang New Year customs, which were verified via a pre-screening questionnaire to ensure that all participants were genuine cultural novices. Participant ages ranged from 18 to 30 years (M = 22.4 years, SD = 3.1 years), and 58% identified as female. Individuals with severe motion sickness, epilepsy, or other contraindications for VR use were excluded. All participants received a modest honorarium upon completion of the study.

4.3 Experimental procedure

Upon arrival at the laboratory, participants completed a pre-test questionnaire in a quiet, standardized room, comprising: (a) a 15-item multiple-choice cultural knowledge test; (b) attitude and cultural identification scales; and (c) basic demographic information. Participants then proceeded to their respective intervention. The VR group received a 3-min standardized equipment orientation before commencing the immersive experience; the control group viewed the slideshow presentation in the same room. The duration of the intervention was 20 min for both conditions. Within 5 minutes of completing the intervention, all participants immediately completed a post-test questionnaire, repeating the knowledge test and attitude measures. The VR group additionally completed measures of perceived ease of use, perceived usefulness, perceived enjoyment, immersive presence (adapted from the ITC Sense of Presence Inventory), and perceived cultural authenticity; the control group completed a baseline assessment of slideshow engagement. Following the post-test, 40 participants (30 from the VR condition and 10 volunteers from the control condition) were invited to participate in brief semi-structured interviews, which were audio-recorded and transcribed verbatim.

4.4 Cultural knowledge assessment instrument

The cultural knowledge test consisted of 15 multiple-choice questions (maximum score = 15), each with four response options, covering five thematic dimensions of the Qiang New Year festival: (1) ritual sequence and structure; (2) symbolic significance of specific rituals; (3) material culture and sacred objects; (4) the role of the Shibi priest; and (5) the communal functions of the festival.

The instrument development process proceeded as follows. The research team drafted initial items based on ethnographic field notes and materials collected during the eight-month community engagement. Two Qiang cultural advisors (one Shibi priest and one community elder) reviewed the items for factual accuracy and cultural appropriateness, resulting in the modification or deletion of questions deemed ambiguous or culturally sensitive. An initial pool of 22 items was pilot-tested with a sample of 30 university students. Items with extreme difficulty values (p < 0.15 or p > 0.90) or poor item-total correlations (r < 0.20) were eliminated, yielding a final set of 15 items with a KR-20 reliability coefficient of 0.76. To mitigate content overlap effects, whereby any advantage observed in the VR condition might reflect direct item-to-content correspondence rather than genuine knowledge internalization, items were intentionally designed to assess inferential reasoning and conceptual understanding rather than verbatim recall of narrated content. For example, rather than asking directly, “What does the priest say during the mountain sacrifice?” items probed underlying cultural logic, such as, “Which of the following best describes the purpose of the Mountain Sacrifice during the Qiang New Year?” The full item list, scoring rubric, item-level statistics, and content validity coverage matrix are provided in Supplementary Appendix A. The item-level content validity index (I-CVI) met or exceeded 0.78 for all retained items, ensuring systematic coverage of the five knowledge dimensions (Polit et al., 2007).

4.5 Data analysis

Quantitative data were analyzed using SPSS v26 and SmartPLS v4. For each participant, knowledge test scores (number of correct responses out of 15 items, range 0–15) and attitude scale scores (mean of Likert-type items rated on a 1–5 scale) were computed for both pre-test and post-test. The attitude scale demonstrated satisfactory internal consistency reliability (post-test Cronbach’s α = 0.88). To evaluate intervention effects, paired-samples t-tests were employed to examine within-group change, and independent-samples t-tests and mixed-design analysis of variance (ANOVA) with group as the between-subjects factor and time as the within-subjects factor were used to compare differences between conditions. Effect sizes were calculated using Cohen’s d for between-group comparisons and partial eta-squared (η²) for ANOVA. The significance threshold was set at α = 0.05. To address potential common method bias, Harman’s single-factor test was conducted by entering all indicator items into an exploratory factor analysis. The first unrotated factor accounted for 28.3% of the total variance, well below the 50% threshold, suggesting common method bias is unlikely to constitute a serious threat to validity (Podsakoff et al., 2003).

For the extended TAM structural model, PLS-SEM analysis (SmartPLS v4) was conducted exclusively on data from the VR condition (n = 100). The choice of PLS-SEM over covariance-based structural equation modeling (CB-SEM) was justified by the exploratory nature of the model, the relatively modest sample size, and the inclusion of formative constructs. Model evaluation followed the guidelines of Hair et al. (2019) and proceeded in two stages. The measurement model was assessed via indicator outer loadings (>0.70), composite reliability (CR > 0.70), average variance extracted (AVE > 0.50), and the heterotrait–monotrait (HTMT) ratio of correlations (<0.85). The structural model was evaluated using the standardized root mean square residual (SRMR) as the primary fit index, alongside R² and Q² values obtained via blindfolding. Path coefficient significance was tested using 5,000 bootstrap resamples, with t-values, p-values, and 95% confidence intervals reported. A complete summary of path coefficients is provided in Supplementary Appendix B, and the HTMT discriminant validity matrix appears in Table 4 in the main text.

4.6 Qualitative analysis

Qualitative data analysis adopted a constructivist grounded theory approach (Charmaz, 2006) and entailed systematic coding of 40 in-depth interview transcripts (30 from the VR condition and 10 from the control condition).

To ensure rigor and methodological transparency, the entire coding process was undertaken collaboratively by two researchers following an iterative trajectory of “independent coding → negotiation of consensus → thematic synthesis.”

4.6.1 Phase 1: open coding

Researchers approached the transcripts from an inductive standpoint, scanning line-by-line without imposing a priori categories in order to capture participants' most nascent psychological experiences. For instance, spatial perceptions such as “sense of being there” were distinguished from social-cognitive dimensions such as “affective resonance with virtual priest interactions.” This phase generated 218 initial code labels.

4.6.2 Phase 2: axial coding and inter-rater reliability

Following independent open coding, the two researchers met to compare their respective code sets. Through multiple meetings, initial codes were compared and refined until consensus on code definitions was achieved. To quantitatively assess coding consistency, 12 interview transcripts (30%) were randomly selected for independent rating, yielding a Cohen’s kappa coefficient (κ) of 0.83. According to the benchmarks of Landis and Koch (1977), this value indicates “almost perfect” inter-rater agreement, thereby substantially bolstering the trustworthiness of the analytic conclusions. On this foundation, thematically similar initial codes were aggregated into higher-order categories. For example, codes pertaining to spatial orientation and sensory simulation were integrated under “embodied immersion,” while codes concerning emotional fluctuation and empathy were grouped under “affective cultural engagement.”

4.6.3 Phase 3: selective coding and theme identification

During selective coding, the research team engaged in deep examination of the causal relationships and interactional mechanisms among categories, ultimately distilling four core themes: embodied participation enhances memory retention, cultural-affective connection transcends content exposure, limited engagement in the control condition, and opportunities for experience design improvement. These themes not only provide a coherent explanation of how immersive experience drives shifts in individual cultural perception but also provide qualitative empirical support for the mixed-methods integration presented later in the article.

5 Results

This section reports, in sequence: baseline equivalence tests; three primary quantitative outcomes (cultural knowledge, cultural attitudes, and heritage support intentions); technology acceptance and user experience measures; PLS-SEM structural model results; and qualitative interview analysis. The significance threshold for all statistical tests was set at α = 0.05, two-tailed.

5.1 Baseline group equivalence

Prior to reporting intervention effects, baseline comparability between conditions at pre-test was examined. Results revealed no significant differences between the VR and control groups in pre-test cultural knowledge scores (VR: M = 7.5, SD = 2.7; control: M = 7.5, SD = 2.6; t (198) = 0.04, p = 0.968) or cultural attitude scores (VR: M = 3.1, SD = 0.7; control: M = 3.2, SD = 0.8; t (198) = 0.90, p = 0.371). These findings indicate that random assignment yielded comparable groups at baseline, and that systematic pre-intervention differences can be largely ruled out.

5.2 Cultural knowledge

Cultural knowledge scores among VR-condition participants exhibited a substantial increase from pre-test to post-test, significantly exceeding the magnitude of change observed in the control condition. The VR group achieved a post-test mean score of 12.9 (SD = 2.3), representing an average gain of 5.4 items relative to their pre-test mean of 7.5 (SD = 2.7), which is an increase in accuracy from 50% to 86%. The control group attained a post-test mean of 9.8 (SD = 2.4), an average gain of 2.3 items from their pre-test mean of 7.5 (SD = 2.6). An independent-samples t-test confirmed a significant difference in post-test scores between conditions (t (198) = 9.5, p < 0.001, Cohen’s d = 1.3).

A cautious interpretation of this effect size is warranted. A value of d = 1.3 is conventionally categorized as a large effect and, within the domain of human–social science experimental research, falls toward the upper range of observed magnitudes. Its interpretation must be situated within the specific context of this study: participants in both conditions possessed virtually no prior knowledge of the Qiang New Year festival (pre-test means approximated random guessing levels); the VR condition received a structured, multisensory, embodied experience over a 20-min period; and the control condition received an equivalent quantity of content delivered via a qualitatively distinct presentation format. Within this context of “zero cultural baseline + brief high-intensity intervention,” a large effect size has reasonable contextual justification. Nevertheless, it should equally be understood as a preliminary finding under specific conditions rather than a robust estimate directly generalizable to other cultural heritage VR contexts. Moreover, because the control group received Qiang cultural materials equivalent in breadth and depth to those presented in the VR condition, the observed between-group difference primarily reflects the contribution of the VR medium’s embodied, interactive, and multisensory properties rather than a mere content-exposure effect. Because the study did not include a third condition (e.g., a non-embodied digital presentation of identical content), the relative contributions of specific sub-components of the medium effect, such as interactivity, spatial presence, and narrative structure, cannot be finely disaggregated. This constitutes an inherent limitation of the study design and will be further addressed in Discussion.

5.3 Cultural attitudes and heritage support intentions

The VR experience was associated with significant improvement in cultural attitudes. Participants in the VR condition showed an increase in affective connection toward Qiang culture from pre-test (M = 3.1, SD = 0.7) to post-test (M = 4.2, SD = 0.6), moving from slightly above the scale midpoint to a considerably elevated level. The control group exhibited a modest increase from 3.2 (SD = 0.8) to 3.5 (SD = 0.7). Mixed-design ANOVA revealed a significant time × group interaction effect (F (1, 198) = 46.3, p < 0.001, partial η² = 0.19), indicating a statistically meaningful difference in the magnitude of attitude improvement between conditions, with a large effect size.

Results for heritage support intentions followed a similar pattern. The VR group increased from a pre-test mean of 3.0 (SD = 0.9) to a post-test mean of 4.1 (SD = 0.8), whereas the control group moved only marginally from 3.0 (SD = 0.8) to 3.3 (SD = 0.9). The between-group difference at post-test was significant (t (198) = 6.5, p < 0.001, Cohen’s d = 0.9, 95% CI [0.61, 1.19]). Across all three primary outcome dimensions: cultural knowledge, cultural attitudes, and heritage support intentions, the VR condition consistently outperformed the control condition at post-test (see Table 2).

TABLE 2. Pre-test and post-test outcomes for VR and control conditions (mean ± standard deviation).

Knowledge scores are based on a 15-item assessment instrument. The control condition received a content-equivalent slideshow presentation on the Qiang New Year. All between-group differences are significant at p < 0.001.

5.4 Technology acceptance and user experience

Technology acceptance among VR-condition participants was generally high. Means for each dimension were as follows: perceived ease of use (M = 6.5/7, SD = 0.6), perceived usefulness (M = 6.7/7, SD = 0.5), and perceived enjoyment (M = 6.8/7, SD = 0.5). The relatively high means should be interpreted in light of the participant profile: university students with prior smartphone and gaming experience who voluntarily enrolled in a technology-based study, a profile known to yield higher technology acceptance scores (Teo, 2011). Ceiling effects cannot be fully ruled out and limit generalizability to less technology-familiar populations and immersive presence (M = 6.3/7, SD = 0.7). Item-level means for perceived cultural authenticity all exceeded the scale midpoint (for 4-point items, M > 3.5), indicating that participants generally endorsed the fidelity of the VR experience as a representation of Qiang culture.

Pearson correlation analysis revealed systematic associations between the quality of the immersive experience and learning outcomes. Participants reporting higher immersive presence scores exhibited greater knowledge gains (r = 0.48, p < 0.001) and more pronounced improvements in cultural attitudes (r = 0.51, p < 0.001). Higher perceived cultural authenticity scores were similarly associated with larger knowledge gains (r = 0.42, p < 0.001) and attitude improvements (r = 0.46, p < 0.001). These correlation coefficients indicate moderate-to-strong positive associations between immersive experience quality and learning outcomes, although correlational analysis cannot establish causal direction.

5.5 PLS-SEM: extended TAM framework

To test the structural pathways of the proposed embodied cultural presence model (ECPM), PLS-SEM analysis was performed on data from the VR condition (n = 100). The analysis adhered to the PLS-SEM reporting guidelines of Hair et al. (2019) and proceeded via sequential evaluation of the measurement model and the structural model.

5.5.1 Measurement model

Prior to assessing structural relationships, the reliability and validity of the measurement model were examined. All indicator outer loadings exceeded the recommended threshold of 0.70 (range: 0.71–0.93), indicating satisfactory indicator reliability. Table 3 reports composite reliability (CR) and average variance extracted (AVE) for each construct. All CR values exceeded 0.80, and all AVE values exceeded 0.50, providing support for convergent validity. Discriminant validity was evaluated using the heterotrait–monotrait (HTMT) ratio of correlations. As shown in Table 4, all HTMT values fell below the recommended threshold of 0.85, confirming that the constructs are empirically distinguishable.

TABLE 3. Measurement model: composite reliability and convergent validity.

CR, composite reliability; AVE, average variance extracted. CR > 0.80 and AVE > 0.50 support convergent validity.

TABLE 4. HTMT discriminant validity matrix.

All HTMT values are <0.85, supporting discriminant validity (Hair et al., 2019).

In this study, immersive presence (IP) and perceived cultural authenticity (PCA) were combined into a higher-order construct termed cultural immersion. Complete disaggregated path coefficients (IP-only model and PCA-only model) are available in Supplementary Appendix B. Within ECPM, IP and PCA jointly constitute the front-end drivers of the cultural identification pathway and are theoretically understood to reflect a common underlying construct: namely, users' subjective perception of the quality of an immersive cultural experience. In the PLS-SEM operationalization, IP and PCA were treated as formative dimensions of this higher-order construct in order to simplify the presentation of structural paths and to mitigate multicollinearity concerns. The present analysis therefore treats IP and PCA as constituents of the higher-order construct cultural immersion (see Supplementary Appendix B for full disaggregated paths) to facilitate methodological scrutiny.

5.5.2 Structural model

Overall model fit was assessed using the standardized root mean square residual (SRMR). The obtained SRMR value (0.063) falls below the threshold of 0.08 recommended by Hair et al. (2019), indicating acceptable model fit. Explained variance for the key endogenous constructs was as follows: cultural identification (R² = 0.52), behavioral intention-VR use (R² = 0.61), and heritage support intention (R² = 0.58). These values suggest that the model exhibits moderate-to-strong explanatory power for each endogenous variable. Predictive relevance values obtained via the blindfolding procedure were all positive (cultural identification Q² = 0.34, behavioral intention-VR use Q² = 0.42, heritage support intention Q² = 0.38), indicating that the model possesses predictive capacity beyond mean-level prediction. Table 5 reports standardized path coefficients, t-values, p-values, and 95% bootstrap confidence intervals (based on 5,000 resamples) for all hypothesized paths.

TABLE 5. Structural model path coefficients (bootstrap n = 5000).

β, standardized path coefficient; 95% CI, obtained via 5,000 bootstrap resamples. Cultural immersion is a composite higher-order construct comprising IP and PCA (see in-text explanation). All paths are significant at p < 0.01. Path coefficients reflect statistical association strength and do not imply strict causal inference.

The test results for each hypothesized path are as follows. H1 was supported: PEOU positively predicted PU (β = 0.30, p < 0.001) and BI-VR (β = 0.24, p < 0.001); PE also positively predicted PU (β = 0.45, p < 0.001) and BI-VR (β = 0.38, p < 0.001), consistent with standard TAM pathways (Figure 6). H2 was supported: Cultural Immersion (IP + PCA) positively predicted cultural identification (β = 0.48, p < 0.001), reflecting the front-end effects of the cultural identification pathway. H3 was supported: PE positively predicted cultural identification (β = 0.29, p < 0.001), aligning with the theoretical expectation that affective engagement facilitates openness to cultural content. H4 was supported: Cultural identification positively predicted heritage support intentions (β = 0.50, p < 0.001), providing preliminary empirical support for the core proposition of ECPM.

Mediation analysis revealed a significant indirect effect of Cultural Immersion on heritage support intentions via cultural identification (indirect effect β = 0.24, 95% CI [0.14, 0.35], p < 0.001), indicating that the mediating role of cultural identification between these constructs is statistically robust. The direct effect of Cultural Immersion on heritage support intention after controlling for the mediator was non-significant (β = 0.09, p = 0.21), indicating full mediation. It bears reiterating that PLS-SEM path coefficients reflect conditional associations rather than experimental causal relationships. The interpretation of cultural identification as a mediator is here structural; the underlying mechanistic causality awaits examination in future research employing experimental manipulation designs.

5.6 Qualitative results

Inter-coder reliability reached κ = 0.83, meeting the threshold for strong agreement (Landis and Koch, 1977). The four themes below were generated through inductive analytic procedures, each illustrated with representative quotations.

5.6.1 Theme 1: embodied participation enhances memory retention

This theme emerged from open codes relating to spatial verisimilitude, sensory fidelity, and the sensation of “being there,” and appeared in 26 of 30 VR interviews (87%). Participants consistently attributed knowledge acquisition to their own operational actions rather than passive viewing and spontaneously invoked bodily and sensory details when describing memories. The theme aligns theoretically with presence theory (Slater, 2009) and is congruent with the high mean ITC Sense of Presence Inventory score (M = 6.3/7).

Representative quotations:

“I won't forget it, because I experienced it all myself—offering the wine, beating the drum. I really felt like I was there, inside the village, not standing outside looking in.” VR participant, female, 24

“The mountains at dawn, the bonfire, the chanting… I wasn't just watching a video. My hand trembled a little when I offered the wine. It sounds strange, but I genuinely felt I was doing something meaningful.” VR participant, male, 21

“Even the smells—it was like a hint of pine or burning firewood—made everything feel more real. I kept forgetting I was in a lab.” VR participant, female, 26

5.6.2 Theme 2: cultural-affective connection transcends content exposure

This theme captures affective engagement, when participants reported feeling moved by or empathic toward the cultural content rather than merely understanding it factually, and appeared in 24 of 30 VR interviews (80%). Open codes included affective resonance, sense of belonging, empathy toward the community, and cross-cultural connection. The theme directly relates to the cultural identification construct (β = 0.50 predicting heritage support intentions) and was particularly salient in NPC interactions and the concluding blessing scene. Participants' affective descriptions substantially exceeded anticipated cognitive gains and exhibited high consistency across coding, warranting designation as an independent thematic dimension.

Representative quotations:

“I could almost feel the heat of the fire and the energy of everyone dancing. It was a joyous atmosphere—like being at a real celebration, not a lesson.” VR participant, male, 23

“At the end, when the old priest offered a blessing for me, I felt incredibly honored—almost as if the community had accepted me. I know it wasn't real, but I truly felt welcomed.” VR participant, female, 22

“I'd never encountered Qiang culture before. But after the experience, I started to care about it. I want to know if the real festival still happens, and whether I might be able to attend.” VR participant, male, 27

5.6.3 Theme 3: limited engagement in the control condition

This theme was derived from codes about engagement, attentional state, and affective investment among control participants and appeared in all 10 control interviews. A highly consistent narrative pattern emerged: content was deemed informative but lacking affective impetus, sustained attention was difficult to maintain, and no participant reported a substantive attitude shift. This pattern echoes the limited quantitative improvement observed (control post-test M = 3.5) and suggests that equivalence in content quantity does not entail parity in affective engagement. In other words, the format of the medium independently moderates emotional involvement.

Representative Quotations:

“The slides gave quite detailed information. I learned about the different parts of the festival, but honestly, my mind wandered a bit toward the end. Knowing about something and actually feeling it are different.” Control participant, female, 23

“I felt like I was completing a study task. The content wasn't boring exactly, but I didn't feel especially engaged. I just watched it and remembered some facts.” Control participant, male, 25

“I know more about Qiang culture now, but I'm not sure it changed how I feel about it. Maybe you need a more immersive way to really connect.” Control participant, female, 20

5.6.4 Theme 4: opportunities for experience design improvement

Approximately one-third of VR interviewees offered constructive feedback across four areas: movement freedom (desire for less guided, more exploratory navigation), onboarding (requests for longer acclimatization for first-time users), content depth (desire for more ritual historical background), and linguistic authenticity (suggestions to incorporate Qiang vocabulary). Post-test questionnaires indicated that two VR participants (2%) reported mild motion discomfort, suggesting room for locomotion design improvement.

Representative quotations:

“The experience was great, but I wished I could wander around the village more freely—explore on my own instead of always being guided to the next task.” VR participant, male, 28, experienced VR user

“The controls took a little getting used to, but once I did, it was fine. Maybe a slightly longer tutorial for people who've never used VR before.” VR participant, female, 19, VR novice

“I noticed most of the narration was in Mandarin. It would feel even more authentic if you could add some Qiang words and phrases—even just simple greetings.” VR participant, male, 24

The qualitative findings corroborate that immersive VR created a meaningful learning experience conveying not only ritual knowledge but also the warmth, spirituality, and communal character of Qiang culture, which is an understanding that participants consistently described as difficult to replicate through reading or viewing alone, aligning with experiential learning theory (Kolb, 1984).

6 Discussion

6.1 Theoretical reflections

6.1.1 The non-proportional relationship between knowledge acquisition and affective identification

The present study demonstrates that the effect size for knowledge gains in the VR condition (Cohen’s d = 1.3) was substantially larger than the effect size for attitude change (η² = 0.19). It warrants clarification that this magnitude of effect is greater than relative to typical benchmarks in social science research, and its provenance merits circumspect consideration. The control group’s pre-test mean score was only 7.5 of 15 (approximately 50% accuracy), indicating that participants had virtually no prior knowledge of Qiang New Year culture. This low baseline objectively amplifies the post-intervention between-group disparity. In other words, the current effect size partly reflects knowledge gain “from zero,” rather than a stable estimate of VR’s marginal advantage over alternative media. In populations with richer prior knowledge or in contexts with higher pre-test scores, this effect size would be expected to contract. Furthermore, the fact that both conditions were exposed to identical content and that between-group differences stemmed solely from the mode of presentation, a design feature that confines the interpretation of the effect size to medium-specific properties, also implies that generalization of the findings must account for interactions between content type and medium fit. PLS-SEM path analysis further indicates that the primary predictive pathway for heritage support intentions operates via cultural identification, rather than directly from knowledge scores. Taken together, these two findings converge on a single direction: in the context of heritage VR, knowledge acquisition and affective identification constitute two distinguishable outcomes, and their mechanisms of influence upon behavioral intentions to support heritage preservation are not identical. This structural distinction carries certain conceptual implications for heritage education practice. Educational interventions designed with knowledge transmission as their core objective, whether slideshows, documentaries, or museum interpretive panels, may be effective in enhancing cognitive cultural knowledge. However, if the goal is to stimulate behavioral intentions for preservation, the activation of affective identification may need to be explicitly incorporated as an independent design target, rather than being treated as a natural byproduct of knowledge gain. Naturally, the strength of this inference is constrained by the single-case design of the present study, and whether it can be replicated across other cultural heritage contexts awaits subsequent empirical examination.

6.1.2 Knowledge gains in the control condition and the separability of medium effects

Following exposure to the content-equivalent slideshow, the control group’s cultural knowledge scores increased significantly from M = 7.5 to M = 9.8, representing an improvement of approximately 15 percentage points. This finding suggests that the slideshow, as an information transmission instrument, is by no means ineffective in terms of pure knowledge efficiency. The core knowledge content about the Qiang New Year festival has a degree of medium-independent transferability. This constitutes an important methodological confirmation of the present study’s design. The relative advantage of VR manifests across two asymmetrical dimensions: a substantially larger magnitude of knowledge gain (accuracy rising from 50% to 86%), and the near absence of significant change in cultural attitudes and identification in the control condition. This asymmetry gives rise to a more finely grained research question: under conditions of constant content quality, what are the respective contributions of embodied interactivity, multisensory stimulation, spatial presence, and narrative structure to depth of learning and affective bonding? The design of the present study cannot disaggregate these sub-components: the difference observed between the VR and control conditions represents the composite effect of multiple factors rather than the independent contribution of any single variable. Addressing this question will require more refined multi-condition experimental designs; the data presented here merely provide preliminary grounds for posing it.

6.1.3 Activation of identification among cultural outsiders

All participants in this study were cultural outsiders with no prior familiarity with Qiang culture. Nevertheless, in follow-up interviews (n = 40; 30 from the VR condition, 10 from the control condition), the asymmetric design reflects the study’s primary focus on VR-mediated experience, with control interviews serving as contrastive context rather than parallel analysis, consistent with embedded mixed-methods convention; Creswell and Plano Clark (2018), following dual independent coding (κ = 0.83), most VR-condition interviewees were classified under the category “described experiences of affective connection.” It should be noted that this proportion derives from coders' categorical judgments of interview transcripts rather than from participants' direct self-report. The two differ in meaning, in that the former is contingent upon the coding framework’s definitions, and the latter is derived from the participants' subjective articulations. Moreover, interviews were conducted immediately following the intervention, when affective priming effects may not yet have dissipated, and the sample size was limited. The stability of the pattern described above awaits further examination with larger samples and delayed measurement. Even so, the convergent presentation of this pattern across both quantitative and qualitative data merits serious consideration. It provides preliminary indication that, under specific design conditions, including embodied ritual participation, community-validated cultural authenticity, and narrative presence driven by social NPCs, VR experiences may be capable of activating affective responses among cultural outsiders that transcend purely cognitive engagement. This bears conceptual resemblance to effects documented in the “perspective-taking” literature (Bertrand et al., 2018), yet its systematic extension into the dimension of cross-cultural identification remains a research domain awaiting empirical construction.

6.1.4 Preliminary evaluation of the ECPM framework

PLS-SEM path results provided preliminary support for the four hypotheses of ECPM, particularly the core proposition that Cultural Immersion exerts an indirect effect on heritage support intentions via cultural identification (indirect effect β = 0.24, 95% CI [0.14, 0.35]). It bears reiterating that PLS-SEM path coefficients reflect conditional associations rather than experimental causal relationships. ECPM, as a provisional conceptual integration derived from a single case, requires that the stability of its structural paths be validated through subsequent research spanning diverse cultural contexts and heritage typologies. At this stage, ECPM is more appropriately positioned as a testable conceptual framework rather than a fully validated theory. This finding resonates with recent work extending TAM for immersive learning environments by incorporating affective and sensory dimensions (Makransky and Lilleholt, 2018; Radianti et al., 2020), lending convergent validity to ECPM’s emphasis on cultural immersion as a core predictor of engagement. Additionally, Champion (2021) has argued that designing virtual heritage places requires fundamentally rethinking how communities are involved in the creation process, which is a principle reflected in the community co-design methodology of the present study.

6.2 Practical implications

6.2.1 Cultural authenticity as a design variable

The most immediate practical implication of this study is that cultural authenticity should not be regarded as an aesthetic concern or an ethical bonus but rather as a core design variable that influences experiential outcomes. The predictive power of the composite Cultural Immersion construct (comprising IP and PCA) on cultural identification (β = 0.48) is comparable to and in some cases, greater than, that of technology acceptance indicators. PCA, as a core dimension of this composite construct, indicates that users' subjective judgment of the credibility of cultural content directly influences the depth of affective connection. (Its contribution to the depth of affective bonding is further detailed in the disaggregated path coefficients presented in Supplementary Appendix B.) This implies that, in resource-constrained development contexts, prioritizing budget allocation toward community participation and content verification, rather than the singular pursuit of visual fidelity, may constitute a more efficacious strategic orientation.

6.2.2 Community co-creation as a methodological necessity

The practical experience of this study indicates that community co-design is not merely an ethically commendable practice; it is a methodological necessity for ensuring cultural authenticity. The subtle deviations originating from the source community identified during the fieldwork phase (e.g., NPC portrayals failing to reflect the equitable participation of men and women), the repeated insistence by community elders that “meaning takes precedence over procedure,” and the explicit expectations articulated by youth representatives for a celebratory atmosphere rather than a museum-exhibit aesthetic, would be virtually impossible to obtain through literature review or expert consultation alone. Integrating knowledge holders into the earliest stages of the development pipeline, rather than merely subjecting the finished product to review, constitutes the most replicable methodological feature of this study. This aligns with the community-centered design principles advocated in the heritage informatics literature (Smith and Akagawa, 2009), yet the present study translates these principles into a concrete iterative workflow, thereby furnishing an actionable reference.

6.2.3 VR as a supplementary tool for heritage education

The findings of this study offer preliminary empirical grounding for the deployment of VR in settings such as museums, schools, and cultural centers. Prudent articulation is required: the effects demonstrated by VR in the context of this study were observed under controlled laboratory conditions and supervised by trained research personnel, with a participant sample consisting exclusively of university students. Direct extrapolation of these effects to authentic deployment contexts must consider factors such as equipment adaptability, variability in usage environments, and differential acceptance across demographic groups (e.g., older adults, children). The appropriate positioning of VR is as a supplement to and catalyst for physical cultural engagement, not a replacement for it. It may open an alternative pathway for populations who, due to geographic or temporal constraints, are unable to be physically present, granting them access to endangered cultural heritage. However, the terminus of this pathway should point toward, rather than supplant, authentic cultural contact.

6.3 Limitations

The present study is subject to five limitations that together demarcate the trustworthy boundaries of its conclusions. First, the single-control-group design constrains the precision with which medium effects can be resolved. While the study demonstrates with relative clarity that the VR experience outperformed the content-equivalent slideshow across three outcome dimensions, it cannot systematically differentiate the independent contributions of embodied interactivity, multisensory stimulation, spatial presence, and narrative structure to VR’s advantage. The addition of a 2D video condition or a non-embodied 360° VR condition would facilitate more refined inferences in future research. Second, the absence of delayed post-test data renders the durability of effects unknown. It is impossible to ascertain whether knowledge gains and attitude changes persist for weeks or months or whether expressed preservation intentions translate into observable behavior. This constitutes the most significant methodological lacuna of the present study. Third, the single-case design raises concerns about external validity. The study is grounded in the specific cultural context of the Qiang New Year and a sample of university students from Sichuan Province (ages 18–30, 58% female, all with tertiary education). Whether conclusions generalize to other heritage types, demographic groups, or technological platforms remains undetermined. Fourth, self-report instruments entail inherent limitations. Cultural attitudes, identification, and preservation intentions were measured via Likert-type scales susceptible to social desirability effects. Coupling self-report measures with behavioral outcome indicators such as tracking post-study engagement in culturally related activities would substantially enhance validity. Fifth, the laboratory setting raises concerns about ecological validity. Conducted on high-specification hardware using standardized procedures, the study presents a palpable gap relative to authentic deployment contexts such as museums, schools, or community centers. Whether experiential effects persist under lower equipment specifications and noisier real-world environments requires systematic evaluation.

7 Conclusion and future directions

This study addressed a pressing practical question: can embodied immersive technology activate, among cultural outsiders, an affective bond transcending information reception, and can this bond be linked to behavioral preservation intentions? Based on the Qiang New Year festival case, the findings offer a preliminary affirmative response that must be understood with due recognition of its boundaries.

The study’s substantive contributions accumulate across three dimensions. First, under a content-equivalent control design, it provides an interpretable empirical pattern demonstrating that the embodied and multisensory properties of the medium are systematically associated with differential outcomes in knowledge gain and affective identification. Second, via PLS-SEM, it provides preliminary validation of the structural pathway wherein cultural identification mediates the relationship between immersive experience and preservation intentions, offering a testable conceptual framework for subsequent research. Third, it documents a co-design pathway that deeply integrates the source community throughout VR development, providing a methodologically attested reference for other endangered ICH digital transmission practices.

A caveat warrants reiteration: ECPM constitutes a provisional conceptual integration from a single case, not a fully validated theory. Systematic testing across cultures, heritage types, and demographic groups is a necessary prerequisite for these preliminary findings to develop into robust knowledge.

7.1 Future directions

Three interrelated directions merit prioritized advancement.

First, the priority of cross-contextual testing should be elevated. Applying embodied VR ethnography to qualitatively distinct ICH types, including oral linguistic rituals, bodily skill traditional crafts, or collective-memory festivals, is both a necessary step for testing ECPM’s boundary conditions and a foundational undertaking for accumulating comparative case knowledge.

Second, longitudinal tracking designs represent the most urgently needed methodological supplement. Repeated measurement of knowledge retention, attitude persistence, and behavioral intention actualization at 4- to 8-week delays would answer the question the present study cannot: do affective effects of immersive experience translate into durable change, or do they decay to the point of indistinguishability from the control condition?

Third, multi-condition fine-grained designs will decompose the internal structure of medium effects. Adding a non-embodied 360° VR condition or an audio-only immersive condition would enable systematic isolation of the respective contributions of embodied interactivity, visuospatial presence, and narrative structures, refining the coarse conclusion that “VR is generally effective” into granular design-informing knowledge.

Within the specific context of the Qiang New Year festival, the embodied VR experience exhibited measurable advantages in knowledge acquisition and cultural attitudes relative to a content-equivalent slideshow, and the mediating role of cultural identification received preliminary structural support. The value of these findings lies in providing a testable and replicable research framework. Subjecting ECPM to empirical scrutiny across broader cultural contexts and demographic populations remains the necessary path toward robust knowledge.

Generative AI statement

The author(s) declared 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/frvir.2026.1823519/full#supplementary-material.

Supplementary material Supplementary material at the source

Supplementary material Supplementary material at the source

Footnotes

Edited by: Sara Gonizzi Barsanti, Università della Campania Luigi Vanvitelli, Italy

Reviewed by: Yunfeng Shu, Zhejiang University, China

Herlan Suherlan, Politeknik Pariwisata NHI Bandung, Indonesia