Cherry-Picking Data

Selection bias happens when the data you measure is not representative. Surveying only gym members about exercise will overstate how much the public works out. The sample looks valid, but it was chosen in a way that guarantees a misleading result.

Graphs can lie too. Truncating the y-axis makes a tiny change look dramatic. Stretching the x-axis flattens a sharp trend. Misleading scales are the most common trick in news graphics, business decks, and political ads.