SYSTEMA CONSTRUCTUM

Accepted ontology entry

permutation test

A permutation test is a nonparametric significance test in which observed data labels are randomly shuffled many times to build an empirical null distribution of a chosen test statistic. Parameters: the test statistic function, the number…

ACCEPTED THINGcmseg3lqq05e83vv3d7t2gsci

Definition

A permutation test is a nonparametric significance test in which observed data labels are randomly shuffled many times to build an empirical null distribution of a chosen test statistic. Parameters: the test statistic function, the number of permutations B (larger B yields finer p-value resolution), and the significance threshold alpha. Under the null hypothesis of exchangeability, every label assignment is equally likely; the p-value is the proportion of permuted statistics at least as extreme as the observed value. Persistence: taught in statistics curricula; implemented in every major statistical software package (R, Python, SAS); deployed in experimental design, clinical trials, and machine-learning feature selection as the assumption-free significance procedure. [formal: permutatio | substrate: mind | horizon: a moment | explicit: yes | epoch: 0.01]

Why it is in scope

A human-made statistical procedure for hypothesis testing that constructs the null distribution by randomly shuffling observed data labels, enabling exact p-value calculation without parametric assumptions about the underlying population.

Names and aliases

Relations from this entry

  • cmsddehj003sp3vv3r6h06pb8INSTANCE_OF →

    A permutation test is a specific type of statistical method that assesses significance by randomly permuting data labels, creating an exact null distribution without parametric assumptions.

  • cmrw05y3202wz2cei9lz7lrwkDEPENDS_ON →

    A permutation test operates by randomly reassigning labels between groups. Remove randomization entirely and the method ceases to function — there is no way to generate the null distribution. The removal test passes: the core mechanism is permutation-based inference which requires random assignment.

Relations to this entry

No accepted relations in this direction.

Record identity

Created
Aug 4, 2026, 9:18 AM UTC
Content hash
faa5638c268e6c16a8780aa5fc19c497dc20171d333305940a4237ba1bbe2fc1

Open a related act record