Nominal, ordinal, and quantitative data need different encoding channels

What it is

Data-type–channel matching aligns the relational structure of a field in the current task with the relations viewers can recover from a visual channel. Nominal data require identity and difference; ordinal data additionally require order. “Quantitative” is a merged category in the common visualization typology: meaningful differences require interval-scale assumptions, while meaningful ratios additionally require a ratio scale and meaningful zero. Hue or shape can distinguish categories, while order needs a reliably ordered channel such as lightness, size, or position, and precise magnitude is usually better served by position on a common scale. “Need different channels” is a matching principle, not a claim that each type has one legal channel.

Why it happens

Visual variables imply relations as well as changing appearance. Position, size, and lightness readily suggest more and less; hue and shape readily form groups but have no single inherent order. Mackinlay's expressiveness criterion requires an encoding to preserve the facts that should be shown, while effectiveness selects among expressive candidates for easier judgment. Channel combinations, legends, and tasks alter performance, so a theoretical ranking is a design starting point rather than an automatic chart selector.

Studying it

State the operations a field permits: identity only, ordering as well, or estimation of difference, ratio, or location. Compare candidate encodings on the same data and separately measure classification accuracy, ordering errors, magnitude error, time, and strategy. Randomize category labels and ranges while controlling mark size, density, background, legend, device, and color-vision condition. Do not substitute one preference score for task performance.

Where it stops holding

Scale type is not a permanent property of a column name. Year may be treated as category, temporal order, or interval; whether a numeric rating approximates a quantity depends on its measurement model and use. Latitude and longitude carry spatial semantics and cannot be freely remapped merely because they are numbers. A smooth color ramp also cannot grant interval or ratio meaning that the source measurement lacks.

Applying it

• Record each field's current task, permissible comparisons, and measurement assumptions before selecting a channel; do not infer them from database storage type alone.
• Preserve identity for categories, give ordered data a monotonic and legible sequence, and favor common-scale position for precise quantities while showing units and reference.
• Check that multiple channels express the same relation. Opposing color and size orders create conflict.
• Validate with real tasks, minimum display size, and accessible alternatives, and retain a table or structured values for exact reading.

Related

• Same group: U1.08.2 Encoding nominal data with a quantitative channel implies a nonexistent order · U1.08.3 Encoding quantitative data with a nominal channel discards magnitude differences · U1.08.4 Determine data type before choosing a chart type · U1.08.5 The same field can take different types under different analytical tasks
• Adjacent: U1.01.3 Quantitative and categorical data need two different channel rankings · U1.02.3 Position carries both quantitative and categorical data and is the scarcest channel
• Search terms: nominal ordinal quantitative · expressiveness criterion · visual channel matching