Comparing and ranking

Comparing and ranking is the type of analysis when your primary concern is to evaluate individual data points and establish a ranking. A key issue with these charts is how to order categories:

  • The alphabetical order of categories is rarely the right solution, and should you should avoid it.
  • Preserve the implicit sequence in ordinal variables, like names of months or weeks.
  • When there is no order (like country names), order categories by the quantitative value.

The generic idea of “ranking” has many variations, and you should try to find the one that best conveys your message. For example:

  • Simple comparison: you have a single series and rank the categories.
  • Actual vs. expected: here, you have two series, and you compare not only values but also compare each value with a reference. Actual versus budget is a typical example.
  • Reference: plotting a reference line or marker helps the reader make sense of the data. For example, add a line that shows the national unemployment rate and use bars to display the regional values.
  • Deviation: calculates positive and negative variation to a standard reference, like zero or 100%. The values appear to the left or the right of the quantitative horizontal axis (or bottom/top of a vertical axis)
  • Range: instead of entire columns, compares the range of values. A typical example is the gender wage gap, with a marker indicating the average.
  • Emphasis: use color to emphasize a category of interest (your country in a list of countries)

Common geometries

The bar (vertical or horizontal) is the most common geometry when comparing and ranking. According to research, people tend to evaluate heights and tops, which justifies a reasonable consensus that we must display the whole bar, which means that the scale should start at zero. When variations are relatively small (variation of Gross Domestic Product), showing the change rate is more frequent than the absolute values.

Because bars take up a lot of space, the chart gets cluttered, and the analysis degrades when adding multiple series. Dots (dot plots, lollipops) are good alternatives, making the chart lighter and easier to read. Some argue that you can avoid starting the scale at zero with dot plots, but if you feel the need to do that, check first if you can use other metrics instead.