Charm Upgrade

Learn · Psychology library

Confirmation bias

People tend to seek and favour evidence that fits what they already believe; that this shapes hopeful reads of a person is untested.

PlausiblePlausible: One study, or indirect evidence from a nearby question. This is the weakest level among the claims below.

What the research shows

A meta-analysis found a moderate average preference for information that agrees with one's view, clearest when people search for information. In lab interviews, people given a hypothesis about someone chose questions likely to confirm it, and the answers tended to back it up, partly through question wording and a tendency to say yes. A review concludes that people reach the conclusions they want when they can build a justification that looks reasonable to them. Separately, people who want more sexual involvement tend to see more sexual interest in others. What you can take from it: before trusting a hopeful read, ask what you would expect to see if you were wrong.12345

What dating marketing says it shows

A typical sales line, written for this page

Dating marketing rarely names this effect. It shows up in the advice instead, when lists of signs of interest invite you to collect evidence for what you hope. A typical sales line promises: "Spot these signs and you will know where you stand."

What doesn't follow

No cited study tests confirmation bias in judging romantic interest directly, so the application is an inference, and the bias says nothing about whether a particular read is right.

Limits

Attitudes on issues and trait impressions, mostly in lab tasks with student samples; the interview studies used supplied hypotheses and, originally, female participants only. The over-perception findings come from scenario and speed-dating ratings. No large registered replication was found for the lab findings.

Sources

Each numbered mark above opens the evidence behind that statement: the label, what it does not show and where it comes from.

  1. Supported with caveatsOn average people prefer information that fits what they already think, and the preference is clearest when they search for information, less so when they interpret or recall it.

    Why this label

    Several studies agree, with stated limits such as student samples or lab settings. Demonstrates: a meta-analysis of choices between agreeing and challenging information found a moderate average preference for agreeing information (d = 0.36), weaker when people held their view less strongly or the information was poor, and a reversed preference when challenging information helped a current goal; a narrative review covers many forms of the bias; one study (N = 199) found it strongest in search, in interpretation only on polarizing topics, and absent in recall. Marketing framing: none common.

    What it does not show

    • Evidence concerns attitudes and beliefs on issues, not impressions of a romantic interest
    • Moderate average effect with variation across conditions; Nickerson's review is narrative, not systematic
    • Laboratory and survey tasks, mostly student or convenience samples
    • Single study for the stage-by-stage pattern (N = 199); no large registered replication found for these findings

    Who it applies to

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

    Sources

  2. PlausibleIn lab studies, people given a hypothesis about someone chose questions likely to confirm it, and the answers they got tended to back it up, partly because of how the questions were worded and a tendency to answer yes.

    Why this label

    One study, or indirect evidence from a nearby question. Demonstrates: Snyder and Swann's four studies had female participants pick interview questions to test a given trait and found they preferred confirming ones, which drew behaviour that confirmed the hypothesis; one later study found asking hypothesis-consistent questions and a tendency to say yes together produced confirming data and conclusions in line with it. Marketing framing: none common.

    What it does not show

    • Staged lab interviews with student participants; hypotheses were supplied, not formed naturally
    • Original study used female participants only
    • The later study suggests part of the effect is question form and agreeable answering, not only motive
    • No large registered replication found; single-lab studies

    Who it applies to

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

    Sources

  3. Supported with caveatsPeople tend to reach the conclusions they want to reach, but only when they can build a justification that looks reasonable to themselves.

    Why this label

    Several studies agree, with stated limits such as student samples or lab settings. Demonstrates: Kunda's review concludes there is considerable evidence that desired conclusions bias how beliefs are accessed, built and evaluated, constrained by the need for plausible justification. Marketing framing: none common.

    What it does not show

    • Narrative review from 1990, not a meta-analysis; mostly pre-2000 lab studies
    • Mostly student samples and self-relevant judgements, not romantic cues specifically
    • Effect sizes not summarised in the abstract; later replication record not reviewed here

    Who it applies to

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

    Source

  4. Supported with caveatsPeople who want more sexual involvement themselves tend to see more sexual interest in others, so one's own wishes can colour how friendliness is read.

    Why this label

    Several studies agree, with stated limits such as student samples or lab settings. Lenton and colleagues found projection of one's own casual-sex desire onto others, stronger when the target's motive was ambiguous. Lee and colleagues found that sex differences in misperception shrank once such projection and sociosexual orientation were accounted for.

    What it does not show

    • Two papers using study-setting ratings (scenario or speed dating)
    • Self-report measures
    • Associations, not proof of a causal mechanism
    • Does not say who is right in a particular encounter; checking directly and kindly is what resolves it

    Who it applies to

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

    Sources

  5. PlausibleThe confirmation-bias research was done on attitudes and trait impressions, so applying it to hopeful readings of friendly or ambiguous cues is an inference, not a tested result.

    Why this label

    One study, or indirect evidence from a nearby question. Demonstrates: the reviewed studies show search and interpretation tilting toward a held belief, and separate work shows people who want more sexual involvement tend to see more sexual interest in others (see romantic-vs-platonic-signals-c6), and that people overrate how well they read others (see nonverbal-inference-limits-c5); no cited study tests confirmation bias in judging romantic interest directly. Marketing framing: lists of 'signs' of interest presented as reliable readings, for example H.O.T. A.P.E. (see dating-frameworks-c1).

    What it does not show

    • Extrapolation across contexts: no direct test of confirmation bias for romantic cues was found
    • Population-level tendencies say nothing about whether one person's reading in a given moment is right or wrong
    • Linked claims carry their own limits (self-report, study-setting ratings)

    Who it applies to

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

    Sources

Used in