Magnitude and distribution

Turns quantity into a length, a position, an angle, or an area, so how much, how spread, and how it compares become visible.

You tell a friend your afternoons are bad. She asks how bad, and you say about a five out of 10. That answer ends the conversation, because there is nothing left to ask about a five. Then you log 60 weekday afternoons, and the five turns out to be the average of two different kinds of day: 24 that sat near a three, and 32 that sat near an eight. Four afternoons were actually a five.

Magnitude and distribution is the family most people already call a chart. It works by turning a quantity into something your eye can measure without counting: the length of a bar, the height of a point, the width of an angle, the area of a band. Once a number has a physical size on the page, comparing two numbers becomes comparing two sizes, and eyes are fast at that. Everything else in the family is a choice about which size to use.

Its nine renderers split into two jobs, and the split is the thing worth learning. Bar, line, area, pie, and isotype answer how much: how much of this, how much over time, how much of the whole. Histogram, bell curve, box plot, and scatter answer how spread: where the cases pile up, how far apart they get, whether two measures move together. People reach for the first group by habit when the live question sits in the second, which is how one average ends up standing in for a set of days that have nothing in common but a mean.

Use it when you have numbers and a question about size or spread. It is weak at explaining. A chart can show that late lunches and dead afternoons show up together, and it has no way to say which one moves the other. It is also silent about anything you did not measure, and what you did not measure is often the reason the number looks the way it does.