Analyze Misleading Graphs

A graph is supposed to make data easier to read at a glance, but the same numbers can be drawn in ways that mislead the reader. The author can shrink the y-axis, change the spacing, tilt the chart in 3D, or leave out part of the timeline so the picture matches a story they want to tell. In this lesson you will learn the most common tricks and how to look past them to the real numbers.

Trick 1 - Truncated y-axis

• A truncated y-axis is one that does NOT start at zero. The bars are still drawn correctly above the start value, but the eye reads bar HEIGHT as bar SIZE.
• Example: three bars at 99, 100, and 101 look almost identical when the y-axis runs 0 to 110, but they look like a giant staircase when the y-axis runs 98 to 102.
• Always read the smallest number on the y-axis FIRST. If it is not zero, ask yourself whether the differences are really that big.

Trick 2 - Inconsistent intervals

• A trustworthy axis uses equal spacing: 0, 10, 20, 30, 40 or Jan, Feb, Mar, Apr.
• A misleading chart might jump from 10 to 25 to 40 without telling you, or place 2018, 2019, 2022 at equal distances and skip 2020 and 2021. The eye sees a steady trend that is not really there.
• Run your finger along the axis and check that each step covers the same amount.

Trick 3 - 3D effects and tilted charts

• Pie charts that are tilted backward make the slice closest to the viewer look bigger than it is, even when the percentages are equal.
• 3D bars cast shadows and have angled tops, so the eye is no longer sure where the bar ends.
• When in doubt, ignore the picture and read the printed values or percentages.

Trick 4 - Cherry-picked range

• Showing only the slice of time that supports a claim is called cherry-picking the range.
• A drug that lowered cholesterol for 6 months but rose again for the next 18 months looks like a success story when the chart only shows the first 6 months.
• Ask: what happened BEFORE this window? What happened AFTER? Is the period long enough to draw a conclusion?