How Figures Mislead

An early draft of the annual report shows nitrate 'skyrocketing', the bar for this year looks three times taller than last year's. You check the axis: it starts at 8 mg/L, not 0. The actual increase is real but modest; the chart made it look dramatic.

Small design choices, a truncated y-axis, a dual axis with mismatched scales, cherry-picked date ranges, or 3D effects that distort area, can make an honest dataset look far more dramatic (or far flatter) than it really is. None of these require faking any numbers; they only require choosing how to draw them.

It is like a photograph that zooms in so tightly on someone's face that a small smile looks like an enormous grin. Nothing about the photo was faked, the camera just chose to crop out the context that would have shown how small the expression actually was.

Truncated axes are one of the most common and most criticized visualization practices in both scientific publishing and media reporting on science, precisely because they're technically 'not lying' about any individual number while still misleading readers about magnitude. Reviewers and readers are increasingly trained to check axis ranges as a first step.

Explain to someone why 'nothing was technically falsified' doesn't mean a chart can't still mislead.