Alone in a crowd of sheep: asymmetric perceptions of conformity and their roots in an introspection illusion

Journal of Personality and Social Psychology 92(4): 585–595, doi 10.1037/0022-3514.92.4.585. Read as the authors' manuscript ("Forthcoming"), deposited by Jonah Berger in the University of Pennsylvania repository (Marketing Papers 292, hdl 20.500.14332/39462; "not the copy of record"). The published version was not read. Read by researcher R5a, 5 October 2026. Provenance: papers/change/pronin2007_crowd_of_sheep.provenance.json.

What was read

• Read in full: all 1,093 lines of pdftotext -layout output of the 25-page manuscript: abstract, introduction, Studies 1–5, general discussion, references, author note, footnotes 1–3, and the two figure pages.
• Not read: the bar charts as images. Their axis text was extracted, and the values used below are stated in the text.

What they did and found (verified)

• Study 1 (44 Princeton undergraduates). Own susceptibility to 16 everyday influences (gestures, dress, drinking, risk-taking, media on political views and others) against the average student, on 1–7 with 4 = the same.
  • M = 3.69, t(42) = 4.62.
  • Denial was larger for undesirable influences (mean within-person b = .22) and for influences that coders rated as operating without awareness (b = .21), each independently.
• Study 2 (40 iPod owners). How much social influence drove their own purchase against peers' purchases, on 1–9 with 5 = the same.
  • M = 3.30.
  • The same whether influence was framed as desirable (3.50) or undesirable (3.10).
• Study 3 (44 Princeton students). Each voted on 18 fictitious university proposals after being told whether "the Princeton panel" of students supported each one.
  • Actual conformity: agreed with the panel on 10.55 of 18 against 9 by chance (t(42) = 4.74).
  • Influence by the panel: own 3.07 against 3.61 for an alleged peer with the same number of agreements (1–7).
  • Influence by own thoughts: 5.68 against 5.09.
  • For themselves they reported weighting their thoughts over their votes, but not for the peer (interaction F = 66.76). This weighting fully mediated the self–other difference.
  • They recalled agreeing less often than they had: own 9.91 against the peer's 10.98, when both were 10.55. Recall tracked the truth less well for self (r = .46) than for the peer (r = .66).
• Study 4 (64 Stanford students). Three California propositions, each presented with the parties' positions.
  • They voted with their party on 2.38 of 3 against 1.5 by chance.
  • Self-rated party influence 3.97, against 4.84 from yoked observers who saw the same thoughts and votes.
  • Actors' self-ratings tracked how many party-related thoughts they had listed (r = .58); observers' did not (r = .06).
• Study 5 (92 Stanford students). Asked what it would mean to conform in a scenario, they chose an introspective definition 63% of the time about themselves and 38% about others.
• Footnote 3. A further study (49 self, 52 peer) found the same thoughts-over-behaviour asymmetry when people explained a conformity episode they admitted to.
• The authors' account. People judge their own conformity by introspection (did I think about fitting in?) and others' by behaviour. Much social influence operates without awareness, so it escapes the self-assessment.

Limits

• Small samples of elite US students.
• Mostly ratings relative to peers, in low-stakes domains.
• Only Studies 3–4 compare self-ratings with actual conforming behaviour.
• Inconsistencies found. Degrees of freedom do not match the ns:
  • Study 1: t(42) with n = 44;
  • Study 2: t(38) with n = 40, and two subgroup tests at t(18);
  • Study 3: t(42) with n = 44;
  • Study 5: F(1, 91) printed beside F(1, 90).

What it means for Kurisutina (inference)

• People report being less influenced than they were, and than equally influenced peers, and they misremember their own conforming record. Interview self-reports of susceptibility are biased low. Following P6, the person's susceptibility settings must come from behaviour or measured probes, not from what they say about themselves.
• A faithful replica would deny its own conformity while showing it in behaviour. Its account of why it changed would come from its "thoughts" (reasons it can name), not from a record of being pushed. A replica that says "you influenced me" may be less person-like. This is a testable signature, and it puts the replica's self-account in tension with the provenance log the machinery keeps.