Extreme values shrink the remaining symbols below legibility

What it is

A proportional symbol's maximum size is usually set by the largest data value. A high outlier can consume the available size range and shrink most other symbols into nearly identical dots: their values still differ, but the differences cannot be seen. This is size-scale compression, not absent data.

Why it happens

Circle area must scale with value, while radius grows with the square root of value. To prevent the largest circle from covering the map, the renderer compresses the full radius range. In a long-tailed distribution, differences among the second-largest and typical values can fall below a pixel or visual threshold. Quietly clipping the maximum or scaling radius linearly makes small values visible but breaks the area-to-value mapping.

Where it stops holding

An outlier can be the anomaly the task is meant to reveal, so it cannot simply disappear for visual convenience. Log or power transforms improve small-value visibility only when the legend and annotation disclose them. If the task is fine comparison among low values, the scale limit signals that a table, facet, or magnified companion view is a better form.

Applying it

• Compare the maximum to the median and second-highest value; flag compression when the largest mark drives the rest towards the minimum clickable size.
• Keep an untransformed overview for the outlier, then add a labeled local view excluding it or a facet with its own range.
• Verify at target display size: ask readers to rank five typical-range symbols. Near-random ranking means use magnification or another encoding.

Related

• Same group: U12.04.1 Proportional symbols encode values as symbol area placed at geographic locations · U12.04.2 Symbols must map value to area and carry a size legend · U12.04.3 Symbols must be drawn in size order when they overlap · U12.04.4 Proportional symbols fit absolute quantities; choropleths fit rates
• Nearby: U2.06.5 If precise comparison of the third variable is needed, facet or make another chart · U3.03.1 Logarithmic scales fit data spanning orders of magnitude
• Search terms: outlier · size scale · proportional symbol map