Dating apps give you exactly what you asked for. You set filters: height, income, education, proximity. The algorithm surfaces profiles that match. You swipe on the ones who look right. You match. You chat. You meet.
And then, far too often, nothing happens.
Not because the person was misrepresented. Not because the conversation was bad. Because the thing you were optimising for — stated preferences, the traits you say you want — turns out to be a remarkably poor predictor of actual attraction.
This is not speculation. It is one of the most replicated findings in mating science, and it has direct implications for how anyone serious about dating in Bangkok should be spending their time.
Key Takeaways
A study of 186,000 real dating decisions found that stated preferences — the traits people say they want — predict near-zero actual attraction.
Dating apps optimize for stated preferences, which turns out to be the wrong variable; this explains why well-matched profiles so often fail in person.
About 60% of strong romantic relationships began with mediocre first impressions — apps systematically filter these out with initial swipe behavior.
In-person interaction is the only environment where actual attraction predictors emerge: rapport, physical chemistry, presence, emotional availability.
The rational response is to use apps as one channel among several, not the primary one, and to weight in-person formats more heavily than their perceived efficiency suggests.
The 4,000-Date Study
Behavioural research on attraction found a consistent gap between what people say they want and what actually produces chemistry. Across more than 4,000 speed dates, researchers measured stated preferences before the event — the traits participants said they prioritized in a partner — and then compared those preferences against who they actually chose at the end of the night.
The correlation was close to zero.
People who said physical attractiveness was their top priority did not show stronger attraction to conventionally attractive partners than people who rated it as unimportant. People who said they required a high-earning partner were not more drawn to high earners when they actually met them. The traits individuals reported as essential prerequisites for attraction had almost no predictive power over who they left wanting to see again.
This finding holds across cultures, age groups, and study formats. It is one of the most robustly replicated results in relationship science. And it is the exact opposite of what dating apps are designed around.
Dating apps in Bangkok, the best dating apps in Thailand, and every major platform globally are built on a stated-preference architecture. You specify what you want; the algorithm filters for it. The entire model assumes that if you could simply meet enough people who match your checklist, attraction would follow. The research says it will not — at least not reliably.
What Stated Preferences Actually Predict
Stated preferences predict one thing well: what you will say your preferences are next time someone asks.
Research on the relationship between stated preferences and actual dating behaviour finds a correlation coefficient in the range of r = 0.10–0.13 — roughly 1% of explained variance in real attraction outcomes. That is not zero, but it is close enough to zero to be operationally useless as a matching criterion. By contrast, the variance accounted for by factors present only in real interaction — shared humour, physical chemistry, conversational rhythm — is substantially larger and cannot be captured in a profile.
The chart above illustrates the gap. In survey conditions, men rate physical attractiveness as a dominant criterion; women rate earning potential and status heavily. These gaps look large and stable. Put the same people in a room — even briefly, in a structured format like speed dating — and both patterns collapse to roughly the same low correlation across all traits. The gender difference that looks definitive on paper becomes nearly indistinguishable in practice.
This matters for anyone relying on a Thailand dating app or any other online platform: the filtering mechanism is optimising for a signal that has near-zero real-world predictive power.
The Gender Gap That Vanishes in Person
The collapse of the gender preference gap deserves its own attention, because it cuts against one of the most commonly cited justifications for profile-based filtering.
The conventional account goes: men are primarily visual, so they maximise for physical attractiveness; women prioritise resources and status, so they maximise for income and ambition. Dating apps should therefore filter on these different criteria for each gender.
Behavioural research challenges this framing. The gender differences in stated preferences are real. But they are largely an artefact of the survey context — asking people to rank abstract traits without the social and sensory information that actual attraction runs on. In face-to-face interaction, research on revealed preferences finds that men and women show roughly equivalent and similarly low correlations between their stated preferences and their actual attraction ratings.
The implication: the stated-preference gender gap is not a description of how attraction works. It is a description of how people think about attraction in the abstract. Thai dating apps, like every other app, are calibrated to the abstract model. The real process works differently.
Where Attraction Really Comes From
If stated preferences do not drive attraction, what does?
Research on relationship chemistry identifies what might be called emergent attraction: the pull that develops through actual interaction, shaped by real-time feedback between two specific people. Chemistry is not a property of person A or person B individually. It is a property of the A-B interaction. It cannot be predicted from a profile of either person in isolation.
Several factors drive emergent attraction that apps cannot screen for:
Micro-responsiveness: how each person tracks and responds to what the other says and does, moment to moment. Behavioural research on early relationship formation shows this is one of the strongest predictors of whether initial interest converts into genuine connection.
Physical presence signals: vocal tone, pacing, eye contact, proximity comfort — the full set of embodied cues that communicate interest and safety. None of these appear in a dating profile or a text-based conversation.
Narrative resonance: whether the stories and frameworks each person uses to make sense of their life feel compatible to the other. This requires actual conversation, not a bio.
Shared context: the specific situation — where you met, what was happening, what you had just talked about — shapes chemistry in ways that pre-meeting profile compatibility cannot replicate.
Dating apps can arrange a meeting. They cannot generate chemistry, because chemistry is not in the profile. It is in the encounter.
How Most Relationships Actually Start
There is a supplementary finding from the same body of research that complicates the standard dating-app mental model further.
Approximately 60% of long-term romantic partnerships begin as friendships, not as direct romantic pursuit. The attraction in these relationships emerged over time, through repeated contact and accumulating shared experience — not from an initial spark that both parties recognised immediately. The person who becomes a partner was not necessarily someone who would have passed a stated-preference filter on first encounter.
Dating apps are structurally optimised for the other 40%: the cases where romantic intent is present from first contact and initial impression is the primary filter. For the majority pathway — the friends-first route where chemistry accumulates gradually — the app model provides no mechanism. The format is wrong for the process.
This does not mean apps are useless. It means they serve a specific use case, and only that use case. For anyone whose actual relationship history shows they tend to fall for people they first knew as friends or acquaintances, the app model is a poor fit for their psychology, regardless of which Thai dating app they use.
What This Means for Dating in Bangkok
Bangkok's international dating scene creates a specific version of this problem.
The city has a high concentration of educated, professionally mobile singles with limited social infrastructure. Expats arrive without pre-existing friend groups. The friendship-first pathway that accounts for 60% of successful relationship formation elsewhere requires time and context that Bangkok's transience makes harder to accumulate. Many people default to dating apps not because they prefer the format but because the alternative — building the social context for organic meeting — takes longer than a three-year contract allows.
This makes format choice more consequential, not less. A Thai dating app puts you in stated-preference mode by design: you assess profiles, filter on criteria, decide in advance. Speed dating in Bangkok inverts the model: you meet a person, experience the actual interaction, and decide after. The latter maps far more closely to how emergent attraction works.
Behavioural research on attraction found that speed dating produces more accurate preference-matching than profile-based selection, precisely because participants are responding to revealed chemistry rather than stated criteria. The format does not guarantee attraction. But it eliminates the stated-preference distortion and gives the actual process — the A-B interaction — a chance to run.
For singles in Bangkok who have been diligent about their dating apps and found the returns disappointing, the research offers a clear explanation: you have been optimising for the wrong signal. The solution is not to optimise harder. It is to change the format.
Frequently Asked Questions
Do dating apps work at all if stated preferences don't predict attraction?
They work for arranging meetings. They do not work as matching systems. The research shows that profile-based filtering does not reliably predict in-person chemistry, but apps can still produce enough meetings that real attraction has a chance to occur by chance. The problem is efficiency: the filter is nearly random relative to the actual outcome, which means the meeting rate required to find a match is much higher than users typically expect.
If chemistry is emergent, how do I know what I'm actually looking for?
You mostly don't, until you're in the room with someone. Behavioural research on stated preferences consistently finds that what people report wanting has low correlation with who they actually choose. The practical implication is to optimise for exposure to a larger number of real interactions rather than for increasingly precise profile filtering.
Is the friends-first pathway possible in Bangkok's expat scene?
Yes, but it requires deliberate social infrastructure. Repeated-contact environments — shared workplaces, interest groups, recurring social events — create the conditions for emergent chemistry to develop. For expats without existing networks, joining recurring groups or events is the functional equivalent of the organic social context that produces friends-first relationships in more rooted communities.
Why does speed dating in Bangkok outperform app-based matching?
Speed dating forces revealed preferences. You respond to an actual person in real time, with real sensory and social information, rather than an abstracted profile. Research on mating decisions finds that in-person brief encounters produce more accurate mutual-selection outcomes than profile-based filtering. LoveLTR's Bangkok dating events are designed around this logic: structured formats that maximise quality face-to-face interactions per unit of time.
What should I actually look for on a date if my checklist is unreliable?
Responsiveness: does this person track what you say and respond to it specifically, or do they stay on script? Comfort: does the conversation feel like work or does it move naturally? Curiosity: are they interested in understanding you, or performing their own profile? These emergent signals are better predictors of long-term compatibility than any criteria you brought in advance.
Conclusion
The research is unusually clear on this point: dating apps are built on a model of attraction that does not reflect how attraction actually works. Stated preferences have near-zero predictive power for real chemistry. The gender gaps that justify differential filtering collapse in face-to-face interaction. The majority of successful relationships begin through a pathway that app-based meeting cannot replicate.
None of this makes apps useless. It makes their limitations legible. The singles in Bangkok who are getting the most from their time are not those filtering harder on their dating apps. They are those who have matched their format to the actual process: more real interactions, less profile optimisation.
Browse LoveLTR's Bangkok dating events for structured formats that put the interaction — not the profile — at the centre of selection.