Connectivity makes a line read as one continuous process

What it is

Uniform connectedness makes parts joined by a continuous path tend to appear as one unit. In a line chart, segments organize discrete observations into one series and support judgments of direction, turning points, and rate of change. They can also invite an inference that change continued between samples. The “continuous process” in the title is a common reading, not a data guarantee: a connection proves neither observation between samples nor linear interpolation or causality.

Why it happens

A continuous path supplies a membership cue for visual tracking, so neighboring samples become successive states of one series. Palmer and Rock introduced uniform connectedness as the tendency to organize uniformly connected regions as units; later work shows that it does not invariably dominate proximity or similarity. Zacks and Tversky's graphic-communication experiments found an association between line displays and trend descriptions, but its strength depended on task, data concept, and communicative situation. Object count, perceived depth, crossings, occlusion, and line discriminability further alter the effect. Perceptual membership in one path must therefore be distinguished from data-level order, sampling, and series identity.

Studying it

Compare points, solid lines, broken lines, dashed segments, and explicit gaps for the same data while manipulating sampling interval, series crossings, stroke, colour, and object count. Separate trend, point-reading, path-tracing, and gap-detection tasks; measure errors, response time, path switches, and inferences about unobserved intervals. Contrast temporal, ordinal, and nominal horizontal variables and ask participants what the connection means. Missing-data studies should cross actual missingness, representation, and task because a break or imputation that helps one judgment need not help another.

Where it stops holding

An ordered axis establishes adjacency but does not by itself justify continuous interpolation; connecting unordered categories invents a process unless a sequence relation is explicitly defined. Cohort changes, revised measurement definitions, long gaps, and estimated segments may all break the premise of one continuing process. Reading direction and time direction should not be conflated: target locales may use different scripts or temporal layouts, so axis direction and order must be explicit. Nonvisual readers cannot recover connections, gaps, or estimated status from the path alone.

Applying it

• Connect points only when horizontal order and adjacency are meaningful. Distinguish unobserved intervals, definition changes, and estimates with breaks, line styles, and direct explanation.
• Position irregular samples at their actual axis intervals and expose sample times; equal visual spacing should not imply equal temporal spacing.
• Keep series identity traceable through crossings and small displays, and provide textual or structured representations of time direction, observed values, estimates, and gaps.
• Test trend, point-value, gap-detection, and path-following tasks. If readers interpret a gap as observed change, revise the connection and explanation rather than relying on a remembered exception.

Related

• Same group: U1.13.1 Proximity and similarity decide the grouping readers see · U1.13.3 Common region can replace the legend for belonging
• Adjacent: U2.03.1 Line charts suit trends over ordered variables · U1.10.5 Redundant channels that disagree semantically give contradictory cues
• Search terms: uniform connectedness · line chart continuity · trend inference · missing-data display