Gallery
Visualisations
Every graphic here is drawn as SVG in the browser, keeps its own data table, and can be operated entirely with a keyboard. Nothing is a bitmap, so figures stay sharp when printed and readable when text size is increased. Each entry says what the graphic is good for and what it hides.
Timeline graphics
A timeline is for showing when something happened relative to something else. It is a poor choice for showing how much.
Placing an interrupted series next to a stable one shows what an interruption looks like. It does not show that one caused the other, and the units are different, so the vertical distance between the lines means nothing.
Canopy changes over decades; diversion changes with a collection contract. Two measures on the same axis can imply a comparison that the underlying processes do not support.
Comparison diagrams
Comparison forms answer one of two questions: how does this rank, or how did this change? Choosing the wrong one is the most common charting error we see.
A slope chart shows direction and relative change clearly. It hides everything that happened between the two dates, including reversals.
The same data as the slope chart, arranged to answer a different question. Notice that Nunavut looks like a failure here and like the largest improvement there. Both readings are accurate.
Grouping works when the measures share a unit and a scale. If they do not, two separate charts are almost always clearer than a second axis.
Stacking is right when the parts genuinely sum to a whole. Only the bottom band can be compared precisely across categories; the middle bands are hard to read and that is a property of the form, not a flaw in the data.
Proportion and quantity
Showing a share of a whole, and showing a count as something a reader can actually picture.
A donut works for four or five categories at most. Beyond that the segments become too similar to compare and a ranked bar chart is better.
A dot matrix turns a percentage into a count a reader can verify by eye. It is slower to read than a bar and better at making a proportion feel concrete.
Tile maps
A tile map trades geographic accuracy for readability. Every jurisdiction gets the same amount of space, so Prince Edward Island is as visible as Ontario.
The trade-off is real: a tile map cannot show that a territory is larger than several provinces combined, and it should not be used where physical distance is part of the story.
Educational charts
Graphics built to teach one idea, usually about how a measure behaves rather than about its current value.
The same construction activity, reported two ways. Adjustment removes the part of the swing that repeats every year so that a genuine change is visible. It also means the adjusted figure for a single month is an estimate, not a count.
A linked trip counts a journey once. An unlinked boarding counts each vehicle. A fare transaction counts each tap. All three are correct measures of different things, and reports rarely say which one they used.
Reuse, formats and accessibility
Reuse
Graphics may be reproduced with attribution to Whab and a link to the page, provided they are not altered in a way that changes what they say. Source material stays under its original licence.
Formats
Because every graphic is SVG, printing at any size stays sharp. Opening the data table gives you the values behind the picture, which is usually what a reuse actually needs.
Accessibility
Each graphic carries a text description, each data point is reachable by keyboard and announced by screen readers, and no meaning is carried by colour alone. See the accessibility statement.