The Misleading-Charts Workshop

P14.chart-craft.05 · Audience: guest, it-ml, language-pro · Prerequisites: Scales, Axes & Legends

Real LLM grading for this pageLLM grading (this page):

You can lie with a chart without faking a single number. That is the unsettling lesson this workshop teaches from the inside. Every trick here uses real data, clean arithmetic and a perfectly ordinary chart type — and still leaves the reader believing something false. A truncated axis turns a 5% wobble into a cliff. A cherry-picked window makes a rising series look like a crash. An area-sized icon blows a modest change out of all proportion. In statistical literacy you learned to spot these as a reader (P1 owns the reader's seat). Here you sit in the maker's seat: you build each distortion on purpose, put a number on exactly how much it lies, and then repair it — because the fastest way to become incapable of being fooled is to have made the lie yourself.

Step 1 / 5The maker's seat — honesty is a choice, not a default

A chart does not become honest by using true numbers. Every lie in this workshop is built on data that is entirely correct; the deceit lives in the encoding, the framing and the window — choices the maker makes after the arithmetic is done. This is why "the data is accurate" is no defence, and why the maker carries a responsibility the numbers cannot discharge: the same honest figures can be drawn to inform or to mislead, and which one happens is up to you.

ⓘ Concept: Real numbers do not make an honest chart
truedata+misleadingencoding/framing/windowafalseconclusionhonestyischosenatthemomentofdrawing,notguaranteedbythenumberstrue data + misleading encoding/framing/window → a false conclusion · honesty is chosen at the moment of drawing, not guaranteed by the numbers

Why it matters — It is comforting to think that if you never fabricate a value you cannot mislead, and it is false. The lie factor, the cherry-picked window and the area trick all operate downstream of the data, on the geometry and the framing, so a chart can be arithmetically impeccable and still plant a conclusion the numbers do not support. Recognising this moves honesty from a property of the dataset to a property of the presentation — something you actively choose at the moment of drawing, not something you inherit from clean data. The whole workshop is built to make that choice visible: build the lie, measure it, repair it, so the honest version is the one you can now produce deliberately.

Ask the mentor about this module

Ask a question about this content. The mentor explains and grounds its answer in what you are studying; asking is recorded as a learning signal, not a grade.

Ctrl/Cmd + Enter to send

🎓 Practice ladder

4 graded rungs · ~40 min

Now build the lies and their repairs yourself. Each rung is a three-panel workspace: instructions on the left, a code editor in the middle, output and test results on the right. Run checks the visible tests; Submit grades against hidden charts. The explorer rung computes a lie factor from a baseline; the practitioner rungs repair a truncated chart and expose a cherry-picked window; the senior rung fixes an area/pictogram trick by scaling area to value.

Rung 1 — Measure the lie factor (explorer)

Loading exercise…

Rung 2 — Repair the truncated axis

Loading exercise…

Rung 3 — Expose the cherry-picked window

Loading exercise…

Rung 4 — Fix the area trick (senior)

Loading exercise…

Try it yourself

A scratch console for this page's ideas — ungraded, nothing you run here is recorded.

Scratch console

A scratch console with the scientific stack (pandas, numpy, scikit-learn). Runs on the server — no network, resource-limited and measured.

Output appears here.