4. Results
The Kaiser–Meyer–Olkin (KMO) measure of sampling adequacy is .975, indicating excellent sampling adequacy and a high proportion of shared variance among variables. Values above .90 are considered indicative of superb factorability, suggesting that the correlation matrix is appropriate for latent structure detection. Bartlett’s Test of Sphericity is statistically significant (χ2(1128) = 35320.933, p < .001), rejecting the null hypothesis that the correlation matrix is an identity matrix. This result confirms that inter-item correlations are sufficiently large to justify factor extraction. Collectively, these diagnostics support the suitability of the dataset for exploratory factor analysis and subsequent structural modeling.
As shown in
Table 1, six factors with initial eigenvalues greater than 1.0 were retained according to the Kaiser criterion using maximum likelihood extraction. The initial solution indicates that the first factor accounts for 31.95% of the total variance. The second and third factors explain an additional 13.39% and 11.37% of variance, respectively, resulting in a cumulative 56.71% prior to extraction refinement. Following maximum likelihood extraction, the six-factor solution explains 61.79% of the total variance. Although the sixth factor exhibits a post-extraction eigenvalue below 1.0, its retention was supported by scree plot inspection and its theoretical coherence within the conceptual framework. The first extracted factor accounts for 31.15% of the variance, while subsequent factors contribute progressively smaller proportions. In behavioral research, cumulative explained variance exceeding 60% is generally considered satisfactory under maximum likelihood extraction. The results therefore support the adequacy of the six-factor structure and justify progression to confirmatory factor analysis.
Table 2 presents the pattern matrix obtained through oblique rotation, revealing a clearly differentiated six-factor structure corresponding to Experiential Immersion, Pastoral Escape, Educational Enrichment, Living Heritage, Perceived Historical Authenticity, and Behavioral Intention. The loading pattern demonstrates strong factorial clarity, with all primary loadings exceeding .70, indicating substantial shared variance between indicators and their respective latent constructs. Experiential Immersion is defined by loadings ranging from .716 to .796, capturing temporal absorption, vivid engagement, active participation, sensory stimulation, and experiential detachment from everyday routine. The consistency and magnitude of these coefficients indicate a coherent experiential dimension centered on embodied and affective involvement. Pastoral Escape displays equally robust loadings (.783–.814), reflecting psychological detachment from modern life, routine disconnection, and restorative motivations. The tight clustering of these indicators suggests a well-defined escape-oriented construct.
Educational Enrichment emerges as a distinct cognitive dimension, with loadings between .786 and .826. Items related to meaningful learning, lifestyle understanding, heritage appreciation, and intellectual stimulation load strongly and exclusively on this factor, confirming its conceptual independence from affective immersion and escape motivations. Living Heritage is characterized by strong loadings (.754–.845), particularly the “Human Element” indicator (.845), underscoring the centrality of performative interaction and lived demonstration in shaping perceptions of authenticity and engagement. Perceived Historical Authenticity shows highly stable loadings (.761–.792), indicating that architectural integrity, material authenticity, historical credibility, and accuracy converge into a tightly structured evaluative dimension. Finally, Behavioral Intention is defined by consistently high loadings (.749–.775), confirming internal coherence of future-oriented behavioral commitment. Cross-loadings are negligible and remain well below interpretative concern thresholds, indicating clear empirical differentiation among constructs. The absence of substantial secondary loadings indicates that the factors are empirically distinguishable despite theoretical proximity. Overall, the pattern matrix supports a stable multidimensional architecture in which affective, cognitive, experiential, authenticity-based, and behavioral dimensions remain conceptually aligned yet statistically distinct. This factorial clarity provides a robust empirical foundation for subsequent confirmatory factor analysis and structural modeling in a high-impact research context.
The measurement model demonstrates very good overall fit to the data. The chi-square statistic is non-significant (χ2(449) = 472.186, p = .217), and the normed chi-square ratio (χ2/df = 1.052) is well below recommended thresholds, indicating minimal discrepancy between the observed and model-implied covariance matrices. Absolute fit indices support model adequacy: RMR is low (.025), while GFI (.974) and AGFI (.969) exceed conventional cut-off values, indicating strong representation of the data structure. The RMSEA is .007 (90% CI [.000, .013]; PCLOSE = 1.000), indicating close approximate fit and very low residual error. Incremental fit indices are consistently high (NFI = .978, RFI = .976, IFI = .999, TLI = .999, CFI = .999), substantially exceeding the .95 benchmark and demonstrating that the specified model fits the data considerably better than the independence model. Parsimony-adjusted indices (PNFI = .885; PCFI = .904) indicate that good fit is maintained while preserving model parsimony. Information criteria (AIC, BIC, ECVI) are markedly lower for the default model compared to the independence model, further supporting the adequacy of the specified model. Hoelter’s critical N values (1127 at .05; 1177 at .01) indicate adequate sample size for stable model estimation. Overall, the measurement model exhibits a stable and well-specified factorial structure with minimal residual variance and strong absolute and comparative fit.
As shown in
Table 3, all constructs demonstrate strong internal consistency, with Composite Reliability (CR) values ranging from 0.878 to 0.919, well above the recommended threshold of 0.70. This confirms a high level of internal coherence among indicators within each latent construct. Average Variance Extracted (AVE) values range from 0.613 to 0.653, exceeding the 0.50 criterion for convergent validity. These results indicate that each construct accounts for more than half of the variance in its respective indicators. Overall, the values presented in
Table 3 provide evidence of satisfactory reliability and convergent validity.
As shown in
Table 4, the square roots of AVE (displayed on the diagonal) are higher than the corresponding inter-construct correlations in all cases. The highest observed correlation (between Experiential Immersion and Behavioral Intention, r = 0.76) remains below their respective √AVE values (0.802). Similarly, correlations between Perceived Historical Authenticity and Experiential Immersion (0.69), as well as between Perceived Historical Authenticity and Behavioral Intention (0.67), remain below the square roots of AVE for the respective constructs. These results indicate that each construct shares more variance with its own indicators than with other latent variables, supporting adequate discriminant validity according to the Fornell–Larcker criterion.
As shown in
Table 5, the majority of HTMT values fall below the conservative threshold of 0.85, supporting discriminant validity among most latent constructs. The highest value appears between Experiential Immersion and Behavioral Intention (0.95), exceeding the conservative criterion and approaching the liberal cut-off of 0.90, indicating substantial conceptual proximity. Given the theoretical linkage between immersive experience and behavioral outcomes, this elevated association is not unexpected; however, it suggests a strong structural relationship between the constructs. The HTMT value between Experiential Immersion and Perceived Historical Authenticity (0.86), as well as between Perceived Historical Authenticity and Behavioral Intention (0.85), slightly exceed or approach the conservative threshold, reflecting theoretical relatedness within the experiential framework. Overall, while certain construct pairs demonstrate high associations, the pattern of results—considered alongside the Fornell–Larcker criterion and CFA factor structure—indicates that the constructs remain empirically distinguishable within the measurement model.
Taken together, the results reported in
Table 4 and
Table 5 confirm that the latent constructs are empirically distinct despite theoretical proximity among some dimensions. The measurement model therefore provides overall support for discriminant validity, considered in conjunction with the Fornell–Larcker criterion and CFA results, and supports the structural integrity of the proposed factorial framework.
As illustrated in
Figure 1, the structural model depicts standardized path coefficients among the six latent constructs, with all hypothesized relationships estimated using maximum likelihood procedures. The diagram visually confirms the directional structure of the model, highlighting Experiential Immersion as the central mediating mechanism linking motivational antecedents (Pastoral Escape and Living Heritage) to evaluative and behavioral outcomes. The magnitude of standardized coefficients displayed in the figure corresponds directly to the estimates reported in
Table 6, demonstrating coherent alignment between the graphical and tabular representations of the model. The structural model demonstrates satisfactory global fit (χ
2(457) = 476.27, p = .258), indicating no significant discrepancy between the observed and model-implied covariance matrices. All hypothesized direct relationships are statistically significant (p < .001), supporting the proposed structural configuration. Pastoral Escape exerts a strong positive effect on Experiential Immersion (β = 0.623, C.R. = 19.20), indicating that motivations related to detachment from modern routine substantially increase immersive engagement. Living Heritage also significantly predicts Experiential Immersion (β = 0.424, C.R. = 14.89), though its effect is notably weaker than that of escape-based motives. Together, these findings suggest that immersion is primarily driven by experiential detachment rather than solely by heritage exposure.
Living Heritage further demonstrates a direct positive influence on Perceived Historical Authenticity (β = 0.410, C.R. = 13.53). However, Experiential Immersion exerts a stronger effect on authenticity perception (β = 0.512, C.R. = 16.01), indicating that authenticity is reinforced not only by heritage characteristics but also by the depth of experiential engagement. Regarding behavioral outcomes, Experiential Immersion emerges as the dominant predictor of Behavioral Intention (β = 0.571, C.R. = 14.07). Perceived Historical Authenticity also contributes significantly (β = 0.271, C.R. = 7.27), although with a more moderate effect size. Educational Enrichment shows a positive yet comparatively weaker influence on Behavioral Intention (β = 0.100, C.R. = 4.12), suggesting that educational motives play a secondary role in predicting future engagement. Overall, as visually summarized in
Figure 1 and detailed in
Table 6, Behavioral Intention is predominantly driven by Experiential Immersion, which exhibits the strongest standardized effect in the model. Perceived Historical Authenticity functions as a complementary pathway that translates immersive engagement into behavioral commitment, while Educational Enrichment demonstrates a comparatively modest—though statistically significant—contribution. The structural configuration therefore positions immersive engagement as the central mechanism within the motivational architecture.
The mediation hypothesis (H8) proposed that Experiential Immersion mediates the relationship between Pastoral Escape and Behavioral Intention. As reported in
Table 7, the results support this assumption. Pastoral Escape exerts a strong positive effect on Experiential Immersion (β = 0.623, p < .001), while Experiential Immersion significantly predicts Behavioral Intention (β = 0.571, p < .001). Bootstrapping analysis indicates that the indirect effect of Pastoral Escape on Behavioral Intention through Experiential Immersion is statistically significant (β = 0.356), as the 95% bias-corrected confidence interval [0.290, 0.465] does not include zero. These findings suggest that the influence of Pastoral Escape operates primarily through immersive experiential mechanisms rather than through a direct behavioral pathway. Accordingly, H8 is supported. H9 proposed a comparative relationship, suggesting that the effect of Experiential Immersion on Behavioral Intention is stronger than that of Educational Enrichment. As shown in
Table 7, while Educational Enrichment has a positive and statistically significant effect on Behavioral Intention (β = 0.100, p < .001), its magnitude is substantially weaker than the effect of Experiential Immersion (β = 0.571, p < .001). The difference in standardized coefficients indicates that immersion-based motivation constitutes the dominant behavioral driver within the model. Therefore, H9 is supported.
To further examine the mediation structure, H10 tested whether the direct effect of Pastoral Escape on Behavioral Intention becomes non-significant when Experiential Immersion is included in the model. As indicated in
Table 7, once Experiential Immersion is specified as a mediator, the direct path from Pastoral Escape to Behavioral Intention becomes non-significant, while the indirect pathway remains significant. This pattern is consistent with full (indirect-only) mediation. Accordingly, H10 is supported. Taken together, the results reported in
Table 7 indicate that Behavioral Intention is primarily generated through experiential immersion mechanisms. Escape-based motives translate into behavioral commitment through immersive engagement, whereas educational motives play a comparatively secondary role.
As shown in
Table 8, the model demonstrates substantial explanatory power across all endogenous constructs. Experiential Immersion exhibits an R
2 of 0.58, indicating that more than half of its variance is explained by its antecedents. Perceived Historical Authenticity shows an R
2 value of 0.67, suggesting strong predictive capacity of immersive and heritage-related drivers. Behavioral Intention reaches an R
2 of 0.72, indicating that the proposed structural configuration explains a large proportion of variance in the final outcome variable. These values indicate strong explanatory capacity of the overall structural model.