Disconfirmation bias
Also known as: Motivated skepticism
We scrutinize evidence against our beliefs far harder than evidence that flatters them.
What it means
Disconfirmation bias is the tendency to scrutinize evidence that challenges a belief we hold far more harshly than evidence that supports it. Congenial claims face a lax test (can I believe this?) and are waved through; uncongenial claims face a strict one (must I believe this?), and effortful search usually finds a flaw or alternative reading to justify rejection. The asymmetry grows with expertise and motivation, because more knowledge supplies more raw material for counter-argument. This lopsided evaluation replicates robustly. Its dramatic downstream cousin — the idea that feeding both sides the same mixed evidence drives opposing camps further apart — is far shakier: early demonstrations leaned on people's self-reports of change, and when later work measured attitudes directly the polarization mostly vanished. The safe reading is narrow: we judge congenial evidence leniently, but shared briefings need not leave committed opponents more divided.
Why it happens
Reasoning is effortful and directional. We rarely audit a belief we already hold and like; we audit the challenge to it. The asymmetry works through the question we silently ask of a claim. For congenial evidence the test is lax: can I believe this? A single plausible reason to accept is enough. For uncongenial evidence the test is strict: must I believe this? We keep searching memory for an escape hatch, and because almost any finding has some flaw or alternative reading, effortful search usually finds one. The extra cognitive work is spent almost entirely on the unwelcome side. This is why the bias grows with expertise and motivation rather than shrinking: more knowledge supplies more raw material for counter-argument, and a belief bound to identity raises the felt cost of conceding.
What the evidence shows
Ditto and Lopez (1992) gave people a saliva test for a fictitious enzyme. Those told the result was unfavorable took longer to accept it as final, were more likely to retest, and blamed disruptions to their routine for a false reading; favorable results were accepted at face value. Taber and Lodge (2006) had partisans rate arguments on gun control and affirmative action: congenial arguments were judged strong and waved through, while opposing arguments drew active counter-argument, and the effect was largest among the most knowledgeable. Kahan and colleagues (2017) sharpened the point: on a politically charged data problem, the most numerate subjects were the most polarized, deploying their quantitative skill to explain away numbers that cut against their side.
Where the strong claims weaken
The asymmetric evaluation itself replicates well. The dramatic downstream claim, that feeding both camps the same mixed evidence pushes them further apart, is shakier. Lord, Ross and Lepper (1979) reported this polarization on the death penalty, but the effect rested heavily on participants' own reports of how much their views had moved. When later work measured attitudes directly, before and after, rather than asking people to describe the change, the polarization mostly vanished. Kuhn and Lao (1996) and others found biased evaluation without reliable net movement apart. So the safe reading is narrow: people judge congenial evidence more leniently, but that does not guarantee committed opponents leave a shared briefing more divided than they arrived.
Using it in practice
The practical lesson is that debate among the committed sharpens rebuttal skills more than it changes minds, so exchanging arguments is a poor tool for persuasion or for settling a disagreement. Two habits help at the individual level. First, apply one standard of proof to both sides: before dismissing an unwelcome study, ask whether you would accept the same design if it favored you. Second, use consider-the-opposite prompts, deliberately generating reasons the disliked conclusion might be right, which narrows the asymmetry in controlled tests. For decisions that matter, separate the two judgments a reader tends to fuse: rate how well a piece of evidence is made before, and apart from, whether you like where it points.
Examples
Shown the same mixed studies on a contested policy, supporters and opponents each rate the studies favoring their side as sound and pick apart the others, leaving both more convinced than before.
A manager who rates an employee highly reads the glowing review as proof and the poor one as a hard quarter with difficult clients. Same evidence, two standards of proof.
A supporter waves through the refereeing decision that favours their team, then demands five replays and a rule-book citation for the one that goes against it.
A reviewer skims the paper confirming their theory and praises its clean design, then combs the one that challenges it for confounds, small samples, and unshared data before recommending rejection.
Told a favourite food is harmful, a reader questions the study's funding and sample size; told coffee extends life, they share the headline without a glance at the method.
First described in Lord, Ross & Lepper (1979); Taber & Lodge (2006).
Key references
- Kahan, D. M., Peters, E., Dawson, E. C., & Slovic, P. (2017). Motivated numeracy and enlightened self-government. Behavioural Public Policy, 1(1), 54-86. doi.org/10.1017/bpp.2016.2
- Taber, C. S., & Lodge, M. (2006). Motivated skepticism in the evaluation of political beliefs. American Journal of Political Science, 50(3), 755-769. doi.org/10.1111/j.1540-5907.2006.00214.x
- Kuhn, D., & Lao, J. (1996). Effects of evidence on attitudes: Is polarization the norm? Psychological Science, 7(2), 115-120. doi.org/10.1111/j.1467-9280.1996.tb00340.x
- Ditto, P. H., & Lopez, D. F. (1992). Motivated skepticism: Use of differential decision criteria for preferred and nonpreferred conclusions. Journal of Personality and Social Psychology, 63(4), 568-584. doi.org/10.1037/0022-3514.63.4.568
- Lord, C. G., Ross, L., & Lepper, M. R. (1979). Biased assimilation and attitude polarization: The effects of prior theories on subsequently considered evidence. Journal of Personality and Social Psychology, 37(11), 2098-2109. doi.org/10.1037/0022-3514.37.11.2098