Strip Plot

When to use a strip plot, what it shows, and how it differs from box plots and violin plots.

A strip plot shows every individual observation as a point, arranged by group along the x-axis. Because nothing is aggregated or summarized, every value in the dataset is visible.

When to use a strip plot

Strip plots are most effective when:

  • Each group has 8 or fewer observations
  • Showing every data point matters — for example, when reviewers or readers need to see the raw values
  • You want to make it clear how sparse the data is

For larger samples, points overlap and the plot becomes difficult to read. A box plot or violin plot communicates the distribution more clearly when observations number in the dozens or more.

The "8 or fewer" rule is a display heuristic for readability, not a statistical cutoff. Some datasets with slightly more points may still read well, while others with tied or repeated values can become crowded earlier.

What the figure shows

Each point represents one observation. Points within the same group are spread horizontally using a small, fixed offset to reduce overlap — this is called jitter. The jitter is applied deterministically based on point order, so the same data always produces the same layout.

Groups are colored by group identity, following the same color palette used in box plots and other group comparison figures.

The strip plot does not display a summary line (mean or median). It shows only the individual points. If you need to communicate both the raw data and a summary statistic, use a box plot with individual points overlaid.

If many observations share the same or very similar values, some points can still visually overlap even with jitter. In that situation, the strip plot may understate how many repeated values are present.

When Licklider suggests a strip plot

Licklider strongly prefers strip plots when there are 8 or fewer observations per group. This is the most common case where showing individual points is both meaningful and visually clear.

That preference is intentionally conservative: it is meant to keep raw points readable by default, not to imply that a strip plot becomes invalid at n = 9.

For slightly larger samples, a box plot with points is often suggested as an alternative.

What this page does not cover