The Same Series on Three Different Vertical Scales

Panel (c) runs its vertical axis from zero to thirty per cent, far beyond anything the series reaches, so the line is squashed into the bottom third and the swings look small. Panel (d) runs from three to ten, close around the data, so the same swings fill the panel. Panel (e) changes the data instead of the axis, plotting five-year averages, which removes the year-to-year movement altogether.

None of the three is wrong. All plot the real unemployment rate. But a reader shown only panel (c) would come away believing unemployment barely moves, and a reader shown only panel (d) would believe it lurches about constantly. The sliders set the top of the first panel's scale and the length of the averaging window in the third.