Only charts sharing a scale can be compared on magnitude

What it is

Scale comparability across views requires compatible units, domains, transforms, directions, and visual ranges when readers directly compare magnitude through mark height, position, or size. “Only” in the title applies to that direct visual comparison. Readers can still compare different scales by reading explicit ticks or values and converting them, but the task has become separate lookup and calculation rather than comparison of graphical magnitude.

Why it happens

A shared mapping makes visual distance transferable between views. With per-chart auto-ranging, equal heights can denote different values and unequal heights the same value. Consistency does not require identical tick labels; it requires the same data-to-pixel or data-to-visual-value function. Bars also need a meaningful zero when length conveys proportion, whereas point and line views can share a disclosed nonzero domain. Log scales can be shared, but equal distances encode ratios rather than raw differences.

Studying it

Using the same data, compare shared scales, independent scales, explicitly labeled independent scales, and direct values on cross-view difference, ratio, order, and shape tasks. Measure error, time, axis checking, and the rate of treating local height as global magnitude; vary co-presence, plot size, and transform. Qu and Hullman formalized same-field same-XY or colour-scale constraints while observing exceptions for data range, layout, and available encodings, supporting evaluation of both the constraint and the reason to override it.

Where it stops holding

Independent domains can reveal within-group change for small-magnitude groups and suit normalized shape or local anomaly tasks, provided absolute amplitude is not implied. When magnitudes span orders, a shared linear domain may flatten small values; use a common log or index scale, a standardized view, a difference view, or separate the tasks. A shared scale is also insufficient for identity, evidential, or causal comparability: definitions, time windows, and populations must align.

Applying it

• Register measure, unit, domain, transform, direction, baseline, and pixel range for each intended cross-chart comparison, and generate them from one scale specification.
• Share scales and plot dimensions for direct magnitude comparison. If that is impossible, prominently label independent scales and use separation, direct values, or a dedicated comparison view.
• Permit free scales for local-shape tasks, but state normalization or independent domains in headings and accessible descriptions rather than relying on ticks alone.
• Test sentinel values for identical rendered positions and rerun the cross-chart task in responsive, exported, and assistive-technology representations.

Related

• Same group: U1.11.1 The same entity keeps the same colour and position across a chart series · U1.11.2 Encoding that jumps between charts forces readers to relearn the legend · U1.11.4 Pin the mapping in the data layer; never configure it chart by chart · U1.11.5 Extend category sets by appending, never by reassigning
• Adjacent: U1.02.5 Position judgment depends on alignment; axis misalignment cancels the advantage · U2.15.2 All facets must share the same scale to be comparable
• Search terms: shared scale · cross-view comparison · scale consistency