Maze of conveniences: re-examining hotel reviews and social media data

Purpose: This investigation explores the increasingly prevalent role of data-driven approaches in hospitality and tourism research, which predominantly relies on expansive datasets from singular sources of social media data. The focus of this study is on uncovering distinct variations across different online travel agencies (OTAs) regarding key analytical metrics, including review topics, sentiment and review length on consumer ratings and their implications for research and practice. Design/methodology/approach: Employing a comparative analysis method, this study analyzed over 11,000 user reviews from a premier boutique hotel in China, gathered from three leading Chinese OTAs. It integrated diverse social media analytics techniques, such as Word Cloud analysis, latent Dirichlet allocation (LDA), traditional regression and PLS structural equation modeling multi-group analysis (MGA) to evaluate review characteristics among the OTAs. Findings: The results reveal significant discrepancies in review topics, sentiments, review lengths and their effects on consumer ratings across the OTAs examined. These findings highlight the risks of relying on a solitary data source for making generalizations. To improve the data-driven research’s validity and reliability in hospitality and tourism, the study advises the adoption of multi-source data for a more holistic understanding of consumer sentiments and behaviors. Originality/value: By pointing out the methodological limitations of present hospitality and tourism research, this paper emphasizes the challenges of a single-source data-driven analytical approach. It suggests several avenues for enhancing the reliability and validity of future research in the field, marking a significant contribution to methodological advancements in hospitality and tourism studies.