Skip to content

Anscombe's Quartet

Four hand-built bivariate datasets that share nearly identical low-order summary statistics yet have radically different scatter geometries, demonstrating that summaries are lossy projections and that one must visualize before granting any parametric summary authority over the data.

Core Idea

Anscombe's quartet (1973) is four small bivariate datasets hand-constructed to share nearly identical summary statistics — means and variances of x and y, correlation, regression slope and intercept, residual sum of squares — while having radically different scatter geometries: a linear cloud, a smooth curve, a linear cloud with an outlier, and a degenerate structure fixed by one high-leverage point. It makes vivid that summary statistics are lossy projections of distributional shape: low-order summaries define equivalence classes of distributions that can contain qualitatively distinct generators.

Scope of Application

The quartet is a constructed demonstration object whose home is descriptive statistics, data visualization, and statistical instruction — its role is to motivate visualization as a complement to summary statistics.

  • Statistical pedagogy — reproduced in essentially every applied-statistics text since Tukey's EDA.
  • Software defaults — shipped as R's datasets::anscombe and seaborn's load_dataset('anscombe').
  • Exploratory-data-analysis methodology — arguing the visualize-before-model workflow.
  • Pedagogical descendants — the Datasaurus Dozen (2017), including a literal dinosaur scatter.

Clarity

The quartet converts the abstract caution "summary statistics are lossy" into a fact one cannot un-see, sharpening the distinction between a summary and the distribution that generated it. It licenses the question "what else could share these exact summaries, and would I notice?" and localizes the prescription: inspect the scatter and residuals before granting a parametric summary authority, since no better summary can recover projected-away structure.

Manages Complexity

An unbounded worry — every way a summary might mislead — collapses into one object: the equivalence class of distributions consistent with the observed summaries, and what varies freely within it. The analyst tracks one dimension of vigilance and a fixed rule (visualize before trusting), with a simple branch: well-behaved region means the proxy stands, curvature or leverage means it is suspect.

Abstract Reasoning

The quartet licenses an anti-inference discipline holding the generating geometry underdetermined by the numbers; a diagnostic scatter-and-residual read classifying which class member is in hand; and a boundary-drawing move fixing which remedy is admissible (looking, not better summaries) and separating the lesson from association-reversal and target-gaming look-alikes.

Knowledge Transfer

As a constructed demonstration object the quartet transfers within statistics as an exhibit-plus-discipline — the lesson and diagnostic propagate across pedagogy, software defaults, and EDA, all one substrate, with the Datasaurus its lineal descendant. Beyond statistics what generalizes is the structural fact it exemplifies, carried by parents compression, dimensionality_reduction, and model_misspecification/validity — kept distinct from Goodhart's law and Simpson's paradox.

Relationships to Other Abstractions

Local relationship map for Anscombe's QuartetParents appear above the current abstraction, mutual partners to the right, and children below. Node labels state whether each abstraction is prime or domain-specific; colors identify relation types.Anscombe's QuartetDOMAINPrime abstraction: Dimensionality Reduction — is a decomposition ofDimensionalityReductionPRIME

Current abstraction Anscombe's Quartet Domain-specific

Parents (1) — more general patterns this builds on

  • Anscombe's Quartet is a decomposition of Dimensionality Reduction Prime

    Anscombe's Quartet is a constructed demonstration of dimensionality reduction collapsing visibly different data geometries onto the same low-dimensional summary vector.

Hierarchy paths (4) — routes to 3 parentless roots

Neighborhood in Abstraction Space

Anscombe's Quartet sits in a sparse region of the domain-specific corpus (71st percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.

Family — Statistical Paradoxes & Distributional Structure (11 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-07-12