Dilution effect
Adding irrelevant facts to relevant ones weakens, rather than sharpens, a judgment.
What it means
The dilution effect is the tendency for non-diagnostic, irrelevant information to weaken the influence of genuinely diagnostic information on a judgment, so that adding worthless detail pulls estimates toward the middle. Named by Nisbett, Zukier and Lemley (1981), it turns up across probability, clinical, forensic and forecasting judgments. The effect looks doubly irrational — as if the mind both fails to ignore worthless information and lets it corrode good evidence — but whether valuable information is genuinely degraded is disputed: a preregistered replication has failed to find the effect at all, and some evidence attributes the gap instead to people overestimating a diagnostic cue when it is shown alone. Accountability and pressure to justify can paradoxically worsen it, because explaining a judgment invites incorporating every mentioned detail. And because it is paradigm-dependent, its real-world reach is uneven rather than systematic.
What the evidence shows
The effect reproduces widely, in probability judgments, consumer ratings, clinical estimates and forecasting tasks, but its strength and even its cause are unsettled. A large preregistered replication of the original grade-prediction paradigm by Highhouse and colleagues in 2022 found no dilution at all, a warning that the size of the effect is paradigm-dependent rather than guaranteed. Its interpretation is contested as well. In 2020 Sanborn and colleagues ran five experiments and concluded that people are roughly accurate when they combine diagnostic and non-diagnostic evidence; the real error sits in the comparison condition, where they overestimate diagnostic evidence shown alone. On that reading the "dilution" is partly an illusion created by inflating single-cue judgments, not the noise corroding good evidence.
When it shrinks or reverses
Non-diagnostic detail does not always dilute. Peters and Rothbart showed the direction depends on typicality: neutral facts that fit the target's category can sharpen a judgment, while facts that make the target seem atypical moderate or even reverse it, evidence that the effect tracks perceived similarity rather than logical relevance. The source matters too, with information from a credible source diluting more than the same content from one the judge distrusts. And awareness is not enough: Kemmelmeier found that instructing people to distinguish useful from useless items left the effect intact, and it vanished only when participants physically removed the non-diagnostic pieces before judging. Knowing which facts are noise does not, by itself, stop them from moving the answer.
Where it shows up
The cost lands hardest where professionals are paid to separate signal from noise. In auditing, Hackenbrack found that padding a file with irrelevant facts pulled fraud-risk assessments toward the middle, softening warranted alarm. Jurors moderate their judgments of a defendant when non-probative biographical detail is presented alongside genuine evidence. Clinicians asked to rate risk grow less decisive once unrelated personal notes are added to a chart. And moral evaluations shift: McHugh and Igou showed that adding trivia about a clearly good or clearly bad person makes observers rate them as more ordinary, with the pull strongest for bad actors. In each case the padding quietly erodes accuracy precisely where a confident, extreme judgment was the correct one.
Using it in practice
Because flagging noise does not neutralize it, the reliable defense is to strip non-diagnostic material out before forming a judgment, not after. Structured formats help: scoring each case only against a fixed, pre-agreed list of predictive criteria keeps stray detail from entering the tally, and some audit studies report that eliciting a frequency estimate rather than a global impression blunts the pull. One common practice backfires, in that demanding a written justification tends to enlarge dilution rather than shrink it, because explaining a decision invites the writer to reach for every fact on the page. The goal is not to gather less information, but to decide in advance which facts are permitted to move the answer.
Examples
Told only that a student studies many hours a week, people predict a high GPA; told the same plus trivia about his hobbies and family, they predict a more average GPA.
A candidate with ten years of directly relevant experience is rated lower once her CV also lists her commute, her hobbies and her school — facts that predict nothing about the job.
A lender told an applicant has never missed a payment forecasts strong repayment; add that he drives a blue car and owns a cat, and the forecast drifts toward average.
An auditor who sees only a suspicious ledger entry rates fraud risk high; surround that entry with routine, irrelevant invoices and the same auditor's rating drifts back toward normal.
A radiologist told only that a scan shows a classic warning sign estimates high risk; add the patient's occupation and travel plans, and the estimate softens toward the mean.
First described in Nisbett, Zukier & Lemley (1981).
Key references
- McHugh, C., & Igou, E. R. (2025). The moral dilution effect: Irrelevant information influences judgments of moral character. Journal of Behavioral Decision Making, 38(4), e70034. doi.org/10.1002/bdm.70034
- Highhouse, S., Freier, L. M., Stevenor, B., Shea, M. A., Childers, M., & Melick, S. R. (2022). Failure to replicate the basic dilution effect in performance prediction. International Journal of Selection and Assessment, 30(2), 195-201. doi.org/10.1111/ijsa.12344
- Sanborn, A. N., Noguchi, T., Tripp, J., & Stewart, N. (2020). A dilution effect without dilution: When missing evidence, not non-diagnostic evidence, is judged inaccurately. Cognition, 196, 104110. doi.org/10.1016/j.cognition.2019.104110
- Kemmelmeier, M. (2004). Separating the wheat from the chaff: Does discriminating between diagnostic and nondiagnostic information eliminate the dilution effect? Journal of Behavioral Decision Making, 17(3), 231-243. doi.org/10.1002/bdm.473
- Peters, E., & Rothbart, M. (2000). Typicality can create, eliminate, and reverse the dilution effect. Personality and Social Psychology Bulletin, 26(2), 177-187. doi.org/10.1177/0146167200264005
- Nisbett, R. E., Zukier, H., & Lemley, R. E. (1981). The dilution effect: Nondiagnostic information weakens the implications of diagnostic information. Cognitive Psychology, 13(2), 248-277. doi.org/10.1016/0010-0285(81)90010-4