Jaehee Gim, Sung in Kim, Soona Park, Heewon Kim
5 min
As solo travel continues to grow, researchers have sought to understand how these travelers differ from those traveling with companions. While previous studies have explored this, they often lacked theoretical grounding or produced inconsistent results. This paper investigates how solo travelers evaluate hotels by examining the impact of star ratings, value-for-money, and service quality on their online review scores, while also analyzing their tendency to leave extreme (polarized) ratings.
Using data from over 81,000 TripAdvisor reviews for hotels in New York and Bangkok, the authors employed multilevel modeling to account for hotel-level heterogeneity. They grounded their hypotheses in several social psychology and economic theories: the Elaboration Likelihood Model (to explain the use of star ratings as heuristic cues), the economic theory of utility (to explain value-for-money perceptions), self-presentation theory (to explain the importance of service quality), and group polarization theory (to explain the tendency toward extreme ratings).
The study reveals that solo travelers are more sensitive to hotel star ratings—specifically for 5-star properties—than companioned travelers, suggesting they rely more on these as a heuristic shortcut. Furthermore, solo travelers place a higher weight on value-for-money when determining their overall satisfaction. Conversely, they weigh service quality less heavily than companioned travelers, likely because they lack the social pressure of impression management that occurs in group settings. Finally, the researchers found that solo travelers are significantly less likely to leave the lowest or highest possible review scores, suggesting a more moderate evaluation style compared to the polarized feedback often seen in group travel.
These findings provide actionable insights for hotel managers and online travel agencies. By understanding that solo travelers prioritize value and are less influenced by service-related social dynamics, hotels can tailor their marketing and pricing transparency to better attract this segment. Additionally, because solo travelers are less likely to leave extreme reviews, managers should treat any extreme feedback from this group as particularly significant, as it may represent a more deliberate and considered assessment of their stay.
Despite the rise in solo travelers, differences in online review behaviors between solo and companioned travelers remain underexplored. The current research, using the Elaboration Likelihood Model, self-presentation theory, and group polarization theory, finds that solo travelers give higher review scores based on hotel star ratings and value-for-money, but they weigh hotel service quality less than group travelers even after accounting for hotel heterogeneity through multilevel modeling. Additionally, solo travelers’ review scores are less polarized than companioned travelers. These findings offer valuable recommendations for hotel management and targeted marketing strategies aimed at solo travelers.
Alex: That has real implications for how platforms aggregate ratings. But I want to push on the identification strategy. The dataset is observational, which means traveler type isn't randomly assigned. What's the confound risk here?
Sam: It's the paper's most significant constraint, and the authors are candid about it. TripAdvisor doesn't provide consistent demographic data—no age, gender, nationality. Those variables could easily be doing work that the social psychology frameworks are being credited for. If solo travelers on TripAdvisor skew toward a particular cultural background with systematically different rating norms, you could recover the same pattern without any mentalization mechanism being involved at all.
Alex: So the theoretical story is coherent, but the identification is bounded by what the platform actually records.
Sam: Right. And the two-city sample—New York and Bangkok—is a reasonable start for cross-cultural validity, but it's still a convenience sample defined by data availability rather than theoretical sampling. The generalizability of the mentalization hypothesis specifically is an open question.
Alex: Does the paper suggest a path toward tightening that?
Sam: They point toward NLP on the review text itself. If mentalization is genuinely driving the effect for companioned travelers, you'd expect their reviews to contain more social language—references to shared experience, other people's reactions, relational framing. Correlating those linguistic markers with companionship status would let you observe the mechanism directly rather than inferring it from metadata. That's a meaningful step, though it still doesn't solve the demographic confound without better platform data.
Alex: So the contribution here is less a definitive causal estimate and more a well-specified framework that names traveler type as a variable the field has been systematically ignoring.
Sam: That's the right read. The multilevel design is solid for what it claims—isolating a systematic difference in rating behavior by traveler type, controlling for hotel-level factors. What it can't do is rule out that unmeasured demographics are the actual driver. For researchers working with review data, the practical takeaway is that traveler type isn't just a segmentation tag for marketing. It's a cognitive filter that changes what the outcome variable represents. Pooling across traveler types without accounting for that is a form of measurement error.
Alex: And one that's probably present in a large share of the hospitality literature that treats star ratings as interchangeable signals.
Sam: Precisely. The next step for the field is either richer platform data or experimental designs that can actually manipulate social context. Until then, this paper gives a clear theoretical basis for why the confound exists, even if it can't fully resolve it.
Alex: A well-framed problem is still a genuine contribution. Thanks for walking through the mechanics, Sam.
Sam: Thanks for listening to ResearchPod.