ISYS3438 Chap.4 Marks, Channels, Colour and Accessibility
Marks, Channels, Colour and Accessibility
Marks represent data items and channels vary their position, length, angle, area, colour or other properties. This chapter connects attribute role with perceptual accuracy. Ordered magnitude is usually strongest on a common position scale, while categorical identity needs channels that separate without implying rank.
Colour is split into hue, saturation and luminance so categorical, sequential and diverging palettes can be chosen deliberately. Accessibility checks make the encoding readable beyond the analyst's own screen.
What this chapter covers
- 01
points, lines and areas
- 02
position and length
- 03
angle and area
- 04
hue, saturation and luminance
- 05
sequential and diverging palettes
- 06
contrast and redundant cues
Worked example · free
Replace bubbles with an ordered comparison
- +1Place gap percentage on a common horizontal position scale.
- +1Sort occupations by value and label the zero or target reference.
- +1Use colour only when it adds a second defensible meaning and remains accessible.
Key terms
- Visual mark
- A visual mark is a point, line or area representing a data item.
- Magnitude channel
- A magnitude channel is a perceptually ordered property used to show more or less.
- Identity channel
- An identity channel separates categories without an inherent order.
- Diverging palette
- A diverging palette uses two ordered colour directions around a meaningful centre.
Marks, Channels, Colour and Accessibility FAQ
Why is position usually stronger than area?
People compare aligned positions more accurately than circle areas or angles. Area may still be useful for overview or geography, but exact comparisons usually need labels or a companion common-scale view.
When should I use a diverging palette?
Use it when values move in two meaningful directions around a defined centre such as zero, target or average. Do not create a centre merely because a diverging palette looks dramatic.
Is colour accessibility only about colour blindness?
No. It also includes contrast, screen size, projection, printing and whether colour is the sole carrier of a critical state. Labels, shapes and position can provide redundant meaning.
Assessment move
Take three charts and list every mark and channel. Rank the channels by how accurately the intended comparison can be made. Convert one colourful design into a monochrome version; anything that becomes ambiguous needs a redundant cue or a different encoding. Test the final result at the size used in submission. Train channel choice with paired comparisons.
Take one small dataset and encode the same ordered values using aligned position, length, angle, area and colour intensity. Time how accurately you can rank close values and recover exact differences. Explain why aligned position and common-scale length usually support precise comparison, while area and angle work better for coarse pattern.
Do not turn the ranking into a prohibition: geography, density and compact overview can justify weaker channels when the task is not exact retrieval. For every chart, name the mark and every channel that carries data. Separate those encodings from styling such as borders or background. Check for accidental double encoding and for channels that imply an order the variable does not possess.
Shape or hue can distinguish categories, while lightness or position can represent ordered magnitude. If size is used, verify that symbol area rather than radius corresponds to value. A legend cannot repair an encoding whose perceptual logic conflicts with the task. Build colour palettes from purpose.
Use qualitative hues for unordered groups, sequential lightness for low-to-high values and a diverging scheme only when a meaningful centre exists. Test the design in greyscale and with a colour-vision simulator, but also inspect contrast, label size and reliance on hover. Critical status should combine colour with text, shape or position.
Avoid red–green opposition as the only signal and reserve saturated accents for the few elements that deserve attention. Conduct an accessibility walkthrough using keyboard order, zoom, export and spoken description. Write a concise alternative explanation that names the question, key pattern, scale and limitation rather than reading every value.
Check that annotations remain legible on projection and that tooltips do not contain the sole definition or caveat. When a requested brand palette fails contrast or ordering, document the risk and propose a compliant alternative. Accessibility is a design requirement that improves scrutiny for the whole audience, not a final colour correction. Create a small encoding register for the finished view.
For each field, list its analytical type, mark, visual channel, legend wording and accessibility backup. Verify that no channel carries two incompatible meanings and no decorative element resembles a data encoding. Print or export the view, reduce its size and inspect whether labels, contrast and hierarchy survive outside the authoring tool. Ask a peer to describe the pattern without relying on colour names.
If the description changes, revise the palette, direct labels or redundant cue. The register becomes useful evidence in a design justification because it connects perceptual principles to deliberate choices rather than claiming that the chart is clear by inspection.
Maintain one exported accessibility proof with the submission: a greyscale check, contrast result, keyboard or non-hover route and short textual summary of the decisive pattern. Note any constraint that could not be resolved and the mitigation used. This turns inclusive design from a general promise into verifiable choices tied to the actual visual.