Charm Upgrade

Learn · Attraction without mythology

What tends to predict liking in first meetings

Lesson 2 of 7 in this module · about 7 minutes · Romantic communication

The idea

Several factors tend to go with liking in first meetings, each with its limits, but these averages predict poorly how one particular person will respond 1. Five follow.

Physical attractiveness. A meta-analysis of 97 samples found that attractiveness went with higher romantic evaluations, a moderate to strong link, about equal for both sexes. Earning prospects had only a weak link for either 2. Speed-dating data showed the same pattern, although some researchers dispute whether men and women value attractiveness equally in a long-term partner 3. Even a strong link leaves most of the variation in liking unexplained, and ratings come from other people and are not fixed traits 3. The research mostly uses Western student samples and photographs, descriptions or brief meetings 2. It offers no basis for judging yourself or anyone else by looks.

Similarity, as the other person sees it. For strangers with little or no interaction, similarity measured on questionnaires predicts attraction, but as people interact more, the similarity that predicts liking is the similarity they perceive 4. The data are correlational, so perceived similarity may partly follow from liking someone 4.

Sensing that the other person likes you. People who sense that a particular person likes them tend to report somewhat more interest in that person 5. The speed-dating evidence is limited, and perceived interest and one’s own interest were measured at the same time, so which drives the other is not settled. As an average, it cannot tell you whether any one person likes you 5.

Sharing about yourself. A meta-analysis found that people who disclose more personal information tend to be liked more, disclose more to those they like, and like those they have disclosed to 6. The links varied with study design, type of disclosure and the speaker’s gender, and most studies were lab and student studies from before 1994 6.

Time in a responsive conversation. In two experiments, strangers who interacted for longer liked each other more, explained by feeling the other person was responsive, comfort and feeling one knew the partner 7. These were same-sex pairs of strangers in a lab, not romantic settings, and the finding is Plausible, not established 7. More time does not guarantee liking. It is not a reason to press for contact with a person who is not interested 7.

Why it matters

A good average is a poor guide to one person. In two speed-dating studies, more than 100 answers about traits and preferences, collected before the event, predicted part of who tends to like or be liked in general. They predicted none of how much one person liked one particular partner 1. A seven-month study of singles found only modest support for specific compatibility and none for matching on ideal preferences 1 8.

Unpredictable from these measures does not mean random, since other measures might do better 1. The speed dates lasted four minutes, and the samples were mostly heterosexual students or community members 8. These are population results, so they cannot tell anyone how a given other person will respond 1.

What it looks like

Example

At a friend’s party you talk with Leila, 29. She asks about your work and listens. You mention your allotment, and she says she grows tomatoes too. You ask, “What are you growing this year?” and stay with her answer. You forecast nothing about how she feels. You watch what she does next.

The same moment, handled the other way

Counterexample

Same party. You tick through the factors in your head: she laughed, you share two hobbies, the hour passed fast. You decide she is bound to like you and stop listening to her answers.

The factors are averages, so they cannot forecast one person.

The one thing to remember

The rule

Use these factors to decide what you do, never to forecast how one person will respond.

Listen responsively, share something real, find out in conversation what you actually share, and give it time. Then watch what she does. That is about this conversation and this person, and no average replaces it.

Try it now

Item 1 of this lesson’s set: “Looks decide everything”.

Open the exercise set

In the real world (optional)

If you like, think of someone you once liked quickly. Keep it in your head, and ask whether what stayed with you was the qualities they had or something you did together.

What this doesn’t tell you

It does not tell you why you liked one person, or whether she liked you. It does not rank people, and it gives no reason to think about your own looks as a score. Much of the evidence comes from students and from brief meetings arranged by researchers.

Why do we believe this?

Each numbered mark in the text opens the evidence behind that sentence. Every claim in this course carries one of five labels.

Well supported
A meta-analysis or several independent studies agree, with no serious failure to replicate.
Supported with caveats
Several studies agree, with stated limits such as student samples or lab settings.
Plausible
One study, or indirect evidence from a nearby question.
Speculative
An idea or argument that hasn't been directly tested.
Marketing claim
Something a seller says about their product, without published evidence.
  1. Supported with caveatsBefore two people meet, their questionnaire answers about their own traits and preferences predict who tends to like or be liked in general, but not how much they will like one particular partner.

    Why this label

    Several studies agree, with stated limits such as student samples or lab settings. In two speed-dating studies, machine-learning models predicted part of each person's general tendency to desire or be desired, but none of the unique desire for a specific partner. A preregistered seven-month study of singles found only modest support for specific perceiver-by-target compatibility and none for matching on ideal preferences.

    What it does not show

    • Two speed-dating samples of romantically unattached people, plus a seven-month study of singles with a different design; largely the same research group.
    • Unpredictable from the measures used does not mean random; other measures could predict compatibility.
    • Self-report predictors collected before meeting; short four-minute dates.
    • Population-level: it cannot tell any individual how a given person will respond to them.

    Who it applies to

    A pattern across groups of people, not a statement about any one person.

    Sources

  2. Well supportedWhen people evaluate potential partners (photographs, descriptions or first meetings), physical attractiveness raised romantic evaluations about equally for both sexes, and earning prospects had only a small effect for either.

    Why this label

    A meta-analysis or several independent studies agree, with no serious failure to replicate. A meta-analysis of 97 samples found attractiveness predicted romantic evaluations moderately to strongly and earning prospects weakly, with sex differences in these links small and not significant, even though stated preferences differ by sex.

    What it does not show

    • Mostly evaluations of hypothetical targets, photographs or first meetings; mostly Western samples.
    • Short-term evaluations, not long-run relationship outcomes.
    • Disagreement exists on how best to test preference matching; not assessed here.

    Who it applies to

    A pattern across groups of people, not a statement about any one person.

    Source

  3. Supported with caveatsIn first meetings, a partner's physical attractiveness is among the strongest predictors of romantic interest for both men and women.

    Why this label

    Several studies agree, with stated limits such as student samples or lab settings. A meta-analysis found a moderate-to-strong link between attractiveness and romantic evaluations for both sexes, with sex differences small. Speed-dating data showed the same equal pattern for real people met in person, even though people stated different priorities.

    What it does not show

    • Mostly student samples; speed-dating events paired opposite-sex participants; attractiveness was typically rated by others.
    • Even a strong correlation leaves most of the variation unexplained, and attractiveness ratings are not fixed traits of a person.
    • The sex-difference finding is contested: a published rejoinder (Meltzer et al., 2014) argues men still value attractiveness more in a long-term mate; that source is not included here.
    • Applies to initial attraction, not to who builds a good relationship.
    • The meta-analysis spans the attraction and relationships literatures, not only first meetings; first-meeting support comes from the speed-dating study.

    Who it applies to

    A pattern across groups of people, not a statement about any one person.

    Sources

  4. Supported with caveatsThe more people actually interact, the more the similarity that predicts liking is similarity they perceive rather than similarity measured on questionnaires.

    Why this label

    Several studies agree, with stated limits such as student samples or lab settings. A meta-analysis found that measured similarity predicted attraction for strangers with no or brief interaction but not in existing relationships, while perceived similarity predicted attraction throughout. A speed-dating study found the same pattern for romantic liking.

    What it does not show

    • The meta-analysis is dominated by lab studies of strangers and information on paper; abstract-level reading only.
    • Perceived similarity may partly result from liking someone rather than cause it; the data are correlational.
    • The speed-dating test is one study of undergraduates at one event series.
    • Perceived similarity on specific traits was a much weaker predictor than a general sense of similarity.
    • Actual similarity still predicted attraction in short-interaction studies in the meta-analysis, with a smaller effect; it lost significance only in existing relationships.

    Who it applies to

    A pattern across groups of people, not a statement about any one person.

    Sources

  5. Supported with caveatsPeople who perceive that a particular person likes them tend to report somewhat more interest in that person, although the speed-dating evidence for this is limited and the studies cannot show which comes first.

    Why this label

    Several studies agree, with stated limits such as student samples or lab settings. In speed-dating, desire that was unique to a pair showed some reciprocity, and in a seven-month study of singles, perceiving that a potential partner was interested was one of the robust positive predictors of one's own interest.

    What it does not show

    • The speed-dating report is a short report with 156 undergraduates; its full text was read in the S7.1 check, which confirmed the pattern; the reciprocity correlation it reports is small (r = .14). The r = -.41 in the same report belongs to a different finding, that desiring many people predicted being desired less (c3).
    • All three sources are correlational: perceived interest and one's own interest are measured by self-report at the same time, so they do not show that being liked causes liking in return.
    • Luo and Zhang describe only 'some support' for reciprocity, so the effect is not uniform across samples.
    • These are population averages; they do not tell you whether a given person likes you.
    • The Luo and Eastwick 2022 abstracts support only 'some' reciprocity and perceived interest as a predictor.

    Who it applies to

    A pattern across groups of people, not a statement about any one person.

    Sources

  6. Well supportedPeople who share more personal information tend to be liked more by listeners, and people tend to like those they have shared with.

    Why this label

    A meta-analysis or several independent studies agree, with no serious failure to replicate. A 1994 meta-analysis found disclosure and liking linked in three directions: intimate disclosers are liked more, people disclose more to those they like, and disclosing to someone increases liking for them. The relation depended on study design and disclosure type.

    What it does not show

    • Based on an abstract-level reading; effect sizes not extracted.
    • Mostly lab and student studies from before 1994; little on dating-specific contexts.
    • Moderated by paradigm, type of disclosure and discloser gender, so the link is not uniform.
    • Correlational and experimental studies are mixed in the pooled effect.

    Who it applies to

    A pattern across groups of people, not a statement about any one person.

    Source

  7. PlausibleIn two experiments, strangers who interacted for longer liked each other more, which runs against the idea that familiarity breeds contempt.

    Why this label

    One study, or indirect evidence from a nearby question. More interaction time led to more attraction, and the effect was explained by perceived responsiveness, comfort during the interaction and feeling one knows the partner. Familiarity helped because the interaction itself gave people more evidence of care and ease.

    What it does not show

    • Two experiments in one paper, pairs of previously unacquainted same-sex strangers in a lab, not romantic contexts.
    • Sample details were not checked beyond the abstract.
    • More time with someone does not guarantee liking; this does not support pressing for contact with a person who is not interested.

    Who it applies to

    A pattern across groups of people, not a statement about any one person.

    Source

  8. Supported with caveatsKnowing someone's traits and stated preferences in advance does not reliably predict whom they will click with: such measures predict general tendencies better than desire for a specific person.

    Why this label

    Several studies agree, with stated limits such as student samples or lab settings. Speed-dating models could not predict desire for specific partners from pre-date self-reports, and a preregistered 7-month study found only modest compatibility effects, with perceived partner attributes, attachment features toward that partner (such as seeking closeness) and perceived interest standing out.

    What it does not show

    • Speed-dating is a 4-minute, mostly heterosexual, student or community setting.
    • Machine-learning null results can reflect measurement limits.
    • The later study was observational and self-reported.
    • Neither abstract tests attachment style as a predictor of whom someone clicks with, and the 2022 study found attachment features felt toward a specific partner among the robust predictors, so the claim says nothing about attachment style.

    Who it applies to

    A pattern across groups of people, not a statement about any one person.

    Sources

More practice

The rest of the set, Evidence check: what predicts liking, and what it cannot predict: 4 items, about 4 minutes. Decide what the evidence supports, and how well. Open the exercise set.

Go further

Outside talks, videos and pages on the topics this lesson cites. Nothing from them loads until you follow a link, and a link is not an endorsement: read the note under each one.

  • Vanessa Van Edwards: 10 Quirky Questions to Ask On Your Next Date

    Video · leaves this site for youtube.com

    Example open-ended, curiosity-led questions to ask on a date, relevant to responsiveness and to restarting a conversation that has stalled.

    Keep in mind. Light entertainment format; questions are prompts, not a technique that guarantees liking.