Mobile travel reviews contain fine-grained emotions that aggregate sentiment misses. This study aims to examine how discrete emotional cues – micro-delight (joy + surprise) and anticipatory anxiety (fear + sadness) – shape tourist satisfaction, and whether tour type (guided vs self-guided) moderates these effects.
A total of 3,678 English reviews were collected from guided (GetYourGuide, Viator) and self-guided (VoiceMap, GPSmyCity) platforms. BERT-based emotion extraction quantified discrete emotions, PCA on sentence embeddings controlled for latent semantic content and ordinal logistic regression tested the hypotheses, supported by robustness checks and model diagnostics.
Micro-delight positively predicts satisfaction (ß = 2.115, p < 0.001), while anticipatory anxiety negatively predicts satisfaction (ß = −1.006, p < 0.001). Tour type moderation was limited: including tour-type interaction terms improved model fit (χ2(2) = 19.49, p < 0.001), but only the micro-delight × tour type interaction was statistically significant, whereas the anxiety × tour type interaction was not (p = 0.479). Tour type was significant as a control (ß = −0.692, p < 0.001), indicating lower baseline satisfaction for self-guided tours.
Hospitality managers should prioritize reducing anticipatory anxiety as a universal strategy, while investing in delight-inducing features that yield comparable returns across guided and self-guided tour formats. For self-guided tour operators, reducing wayfinding anxiety and amplifying delight through personalized content offer a clear path to improve satisfaction.
The study provides large-scale evidence that discrete emotional cues predict satisfaction beyond aggregate sentiment, supporting emotional granularity in mobile review contexts. The findings provide limited evidence that tour type alters the relationship between discrete emotions and satisfaction, suggesting that emotional management strategies may be applied broadly across tour formats. The methodological pipeline demonstrates how BERT-based emotion extraction can be aligned with theory-driven hypothesis testing in tourism research.
