Cherry Picking

Selectively presenting confirming evidence while suppressing disconfirming evidence from the same available population, so the offered sample gives an impression the full distribution would not support — locating the dishonesty in the selection process, not the individual data points.

Version

v5 · 2026-09-28 · History · 3 corrections

Domain-specific #

151

Domain group

Humanities

Origin domain

Rhetoric

Subdomain

Evidence Evaluation and Fallacies → Rhetoric

Aliases

Suppressed Evidence, Fallacy of Incomplete Evidence, Incomplete Evidence Fallacy, Card Stacking

Core Idea

Cherry picking (the fallacy of incomplete evidence) is selectively presenting confirming evidence while suppressing disconfirming evidence from the same available population, so the offered sample gives an impression the full distribution would not support. The structural commitment is a specific locus of dishonesty: not at the individual data points, each of which may be accurate, but at the selection process that chose them. It is therefore undetectable by local fact-checking; it becomes visible only against the population the sample was drawn from.

How would you explain it like I'm…

Only the Good Cherries

Imagine a bowl of cherries where some are yummy and some are rotten. If you only show your friend the yummy ones, they'll think the whole bowl is great, even though it isn't. Cherry picking is when someone only shows the facts that help their side and hides the ones that don't. Each fact can be true, but the picture is still unfair.

Picking Only Helpful Facts

Cherry picking is when someone argues for something by showing only the evidence that supports them and leaving out the evidence against them. Each fact they show might be completely true. The trick is in which facts they chose to show. That's why checking each fact one at a time won't catch it; you have to ask what they left out. For example, if a study measured five results and reported only the two that looked good, that's cherry picking.

Selective Evidence Fallacy

Cherry picking, also called the fallacy of incomplete or suppressed evidence, is presenting evidence that supports your conclusion while hiding evidence against it from the same pool, so the audience gets a false impression of what the full evidence shows. What's dishonest isn't any single fact, which may be accurate, but the selection process: the choice of what to show depends on what you want to conclude, yet the sample is presented as if it were representative. Because of that, fact-checking individual claims can't detect it; you have to compare the sample with the full set or ask what the left-out part would show. In science, it appears as selective outcome reporting, such as a trial measuring five pre-registered outcomes and publishing only the two that were significant. Remedies include pre-registration, systematic reviews and requirements to disclose adverse evidence.

 

Cherry picking (the fallacy of incomplete or suppressed evidence) is selectively presenting confirming evidence while suppressing disconfirming evidence from the same available population, so that the offered sample creates an impression the full distribution of evidence would not support. The fault lies in the selection process rather than in individual data points, each of which may be accurate: the arguer samples in a way correlated with the preferred conclusion and presents the sample as representative. The fallacy is thus invisible to local fact-checking and becomes visible only by comparing the sample with the population it came from or asking what the complementary subset shows. In science it takes the form of selective outcome reporting, for example a trial that pre-registers and measures five outcomes but reports only the two significant ones; each result is real, but the inference is unwarranted because reporting depended on the result. Structural remedies follow: pre-registration makes selective reporting detectable, systematic review counters selective citation, professional-conduct rules requiring disclosure of directly adverse controlling authority counter selective legal argument, and adversarial formats force engagement with counter-evidence.

Scope of Application

Cherry picking lives across reasoning and the fields that present evidence to persuade, wherever a deliberate presenter samples an evidence population in a conclusion-correlated way.

• Reasoning and argumentation — the named fallacy in informal logic ("incomplete evidence," "suppressed evidence").
• Scientific communication and meta-research — selective outcome reporting and the popular framing of publication bias.
• Journalism and political rhetoric — citing supporting anecdotes while omitting countervailing ones.
• Marketing claims — "9 out of 10 dentists" subgroup framings and selectively quoted reviews.
• Legal argumentation — hostile selective use of precedent, against which rules of professional conduct require disclosure of directly adverse controlling authority.

Clarity

Naming cherry picking relocates the locus of dishonesty from the individual claim to the selection process, and that relocation makes the error catchable at all. It defeats the natural defense of verifying each datum, because every offered point can be accurate while the impression is unwarranted. It shifts the question from "is this claim true?" to "was the displayed subset drawn correlated with the conclusion?" and separates the fallacy from confirmation bias, passive selection bias, and the Texas sharpshooter fallacy.

Manages Complexity

An evaluator otherwise faces a miscellaneous list of misleading moves, each defeating the wrong instrument of claim-verification. Cherry picking compresses the list to one regularity by relocating the fault to the selection process, reducing an open-ended audit to one tracked relationship — the correlation between the selection rule and the conclusion. From it the verdict reads off via one counterfactual probe: what would the complementary subset show? The branch structure is binary, and the remedies organize into one family attacking the selection rule.

Abstract Reasoning

The concept licenses a diagnostic move — inferring the fault lies in the selection rule and treating local truthfulness as the wrong instrument, probed by the complementary subset. It licenses an interventionist move — remedies attacking the selection rule, each foreclosing one suppression channel. It draws boundary lines to a clean binary and against confirmation bias, passive selection bias, and the sharpshooter fallacy. And it applies order-of-events reasoning tracing the impression to the selection step.

Knowledge Transfer

Within reasoning and evidence-presenting fields cherry picking transfers as mechanism, because every instance is the same relocation of the fault to the selection process, resolved by one probe and one remedy family across rhetoric, science, law, journalism, and marketing. Beyond evidence-presentation, what recurs is not "cherry picking" but the broader mechanism it specializes — biased sampling distorting an inference — which is Selection Bias (with Confirmation Bias for the cognitive precursor). Those parents reach substrates with no presenter; a planet's orbit cannot cherry-pick, so importing the term there mislabels a passive process as deliberate.

Relationships to Other Abstractions

Current abstraction Cherry Picking Domain-specific

Parents (1) — outgoing typed relations

• Cherry Picking is a kind of Selection Bias Prime

  Cherry picking is selection bias specialized to a deliberate presenter whose conclusion-correlated evidence selection makes the displayed subset misrepresent its source population.

Hierarchy paths (5) — routes to 5 parentless roots

• Cherry Picking → is a kind of Selection Bias → is a kind of Bias

Show alternative paths (4)

Neighborhood in Abstraction Space

Cherry Picking sits in a moderately populated region (49^th percentile for distinctiveness): it has near-neighbors but no dense thicket of look-alikes.

Family — Unclustered & Miscellaneous (2769 abstractions)

Nearest neighbors

• Null Ritual — 0.86
• Two-Sided Message — 0.86
• Quote Laundering — 0.86
• Spotlight Fallacy — 0.86
• Kettle logic — 0.86

Computed from structural-signature embeddings · 2026-10-25