Comparison
What the research says about confronting someone
September 13, 2026 · 3 min read
Five findings that keep getting cited in conversations about speaking up. They come from different literatures and measure different things, and each one is routinely stretched a step past what it tested. Each entry below gives the study, the result, and the claim it does not support. Background for The Trade She Never Accepted.
Czopp, Monteith & Mark (2006) — confrontation temperature
The study: Three experiments in Journal of Personality and Social Psychology. Alexander Czopp, Margo Monteith and Aimee Mark confronted participants about stereotyped responses they had just given. The first experiment varied how harshly the confrontation was worded.
What it found: Both the mild and the hostile versions reduced stereotyped responding afterwards. The third experiment, which did not vary tone, found a confrontation also lowered self-reported prejudice on a broader attitude measure. Harshness raised participants’ anger and irritation toward the confronter.
What it does not say: That gentleness persuades better. The correction landed either way. Hostility was a cost, not a lever.
Legault, Gutsell & Inzlicht (2011) — motivation for change
The study: Lisa Legault, Jennifer Gutsell and Michael Inzlicht, in Psychological Science, compared anti-prejudice messaging that emphasised external requirements to comply against messaging that engaged participants’ own reasons.
What it found: The controlling version produced more prejudice than running no intervention at all. The autonomy-supporting version reduced it.
What it does not say: That pressure never changes behavior. It says pressure framed as an external demand can move the measured attitude backwards — a specific and counterintuitive result, not a general rule about influence.
Hornsey & Imani (2004) — who is allowed to criticize
The study: Matthew Hornsey and Armin Imani ran three experiments in Personality and Social Psychology Bulletin, holding the criticism constant and varying whether the critic was a member of the criticized group.
What it found: Insider criticism drew substantially less defensiveness. The effect ran through whether listeners believed the critic meant to be constructive. Experience with the group did nothing for outsiders, and an insider’s lack of it hurt.
What it does not say: That outsiders should stay quiet. It measures how criticism is received, not who has standing to give it.
Gaertner, Dovidio et al. (1989) — one group instead of two
The study: Samuel Gaertner, John Dovidio and colleagues assigned 360 participants to two groups, then manipulated whether they thought of themselves as two groups or as one larger one.
What it found: The one-group condition lowered bias between the former subgroups — the effect they named recategorization.
What it does not say: Anything about keeping both identities at once. The dual-identity refinement came from later work; this paper tested the single superordinate group.
Gottman — contempt as a predictor
The study: Longitudinal observation of couples by John Gottman, producing the four behaviors he named the Four Horsemen. Full breakdown: Gottman’s Four Horsemen and their antidotes.
What it found: Contempt is the single strongest behavioral predictor among the four. It differs from criticism in target: criticism disputes what someone did, contempt disputes whether they rate a hearing.
What it does not say: That one contemptuous moment predicts an outcome. The research is about sustained pattern.
How to read them together
They answer different questions, and the overlap is narrower than it looks. Czopp measured whether temperature changes the result; Legault measured whether the framing of a demand changes the attitude; Hornsey measured whether the source changes the reception; Gaertner measured whether the group boundary itself can be moved; Gottman measured what predicts an outcome over years.
Put side by side, none of them recommends saying less. What they converge on is narrower: the parts of a confrontation that do the work and the parts that only feel like they do are separable, and the separation is not where most people assume. Two applications: Does being harsh make criticism more effective? and, when the group is your own, how to criticize it without sounding superior.
From the essay:Read the full piece →
Want more like this? Subscribe to the newsletter.