Problems with Charts
Problems with Scales and Axes | Optical
Effects
Problems with Scales and Axes
Improper use of scales and axes may lead to misinterpretations of charts.
- Do not exaggerate differences by shifting the X axis to a higher
(or lower for negative values) level.
Example: If the X axis is shifted to
a value of 15, the difference between 16 and 17 seems to be very large, although
it is less than 10%.
- Do not minimize differences by choosing a range that is too large
for the actual values.
Example: Do not use a range so that the
values only fill only 20% or less of the chart.
- Do not use a scale that leads to wrong interpretations.
Example: A logarithmic scale may minimize
relevant differences.
- Do not use different scales on charts that are to be compared; different
slopes for identical changes may lead to wrong conclusions.
Optical Effects
Optical effects may lead to distortions, and thus to misinterpreations of charts.
- Care for distortions in areas or volumes. Do not vary both
width and height.
Example: Be careful with 3D charts where
3D borders may add to the perceived area.
Example: Do not increase width and height
simultaneously with the data values - while size increases linearly, area
is squared (see figure 1).
Example: Be careful with pictographs
that use images of different sizes to represent quantities (analog to simultaneously
increasing width and height).
- Do not use brightness to exaggerate areas.
Example: If the largest area is lightest
and the smallest area darkest, the perceived differences between the areas
are increased (see figure 2).
- 3D graphs are fancy, but mostly misleading, especially if pie charts
are used (see figure 3 and 4).
- Avoid patterns and textures, e.g. striped patterns, that
cause optical illusions or perspective effects that distort areas.
- Do not combine distortions.
Example: Do not combine brightness variations
with area distortions like simultaneously varying width and height (see figure
2).
Examples
Figure 1: Areas quadruple, if lengths double; the effect
may even be worse with realistic images.
Figure 2: Changing width and height simultaneously as
in figure 1, as well as varying brightness, may exaggerate the increase even
more.
Figure 3: A 3D column chart makes it hard to compare
actual lengths; moreover, perspective may let columns in the background appear
larger than columns of the same size in the foreground (size constancy)
Figure 4: Perspective makes it impossible to reliably
compare the sections of a pie; furthermore, 3D borders may add to the area
of segments. Also angles are much harder to estimate than distances.
top