Violin vs Box Plot: Which Should You Use?

Box plots and violin plots answer the same question — how do these groups compare? — but they show different things. Here's how to pick the right one.

Both charts line groups up to compare them. The difference is what they draw: a box plot shows a five-number summary; a violin plot shows the full shape.

The box plot's strength: clean comparison

A box plot reduces each group to median, quartiles, and outliers. That spareness is the point — you can line up a dozen groups and read level, spread, and skew instantly, with no visual clutter. For dashboards and many-group comparisons, it's hard to beat.

The violin's strength: seeing shape

A violin draws a smoothed density, so it reveals what the box hides — above all, bimodality. Two groups can have identical box plots while one is a single hump and the other is two separate clusters. If the shape carries meaning, the violin shows it.

A box plot and a violin plot of the same data side by side — Same data: the box summarizes, the violin shows the full shape.

The quick rule. Many groups, you just need the summary → box plot. Fewer groups, the distribution's shape matters (or you suspect two clusters) → violin. And you don't have to choose: a violin with the quartile box drawn inside gives you both.

Trade-offs

Violins need enough data per group — a density estimate from a handful of points is misleading.
Box plots are more familiar to general audiences.
Violins use a smoothing bandwidth; too smooth hides structure, too rough invents it.

Make either in Stratum: box plot or violin plot.

Download Stratum

Frequently asked questions

What's the difference between a violin plot and a box plot?

A box plot shows a five-number summary (median, quartiles, outliers); a violin plot shows the full smoothed distribution shape. The violin reveals multiple peaks that a box plot hides.

When should I use a box plot vs a violin plot?

Box plot for clean comparison across many groups when you just need summaries; violin when the distribution's shape matters or a group might be bimodal.

Why might two groups have the same box plot but different violins?

Because a box plot only knows quartiles. Two groups with identical quartiles can have very different shapes — one unimodal, one bimodal — which only the violin reveals.

Do violin plots need a lot of data?

More than a box plot — the density estimate is unreliable with very few points per group. With small samples, prefer a box plot or show the individual points.