Raincloud Plot Maker for Research Data

What does this raincloud plot maker do?

A raincloud plot stacks three views of the same data: a half-violin density "cloud", a box plot, and the raw observations as jittered "rain". This raincloud plot maker computes each layer from your uploaded values, so readers can see the shape of each distribution, its median and spread, and every data point without switching figures. Bar charts with error bars hide skew, bimodality, and outliers; a raincloud plot shows them, which is why many journals and reviewers now encourage it for small and medium samples.

How the cloud, box, and rain are computed

• The cloud is a Gaussian kernel density estimate with Silverman bandwidth, 0.9 × min(SD, IQR/1.34) × n^-1/5, evaluated on a 24-point grid shared by all groups.
• Each cloud is scaled to the same peak height, so this half violin plot compares shapes and locations rather than sample sizes.
• The box spans Q1 to Q3 with a median line; whiskers reach the most extreme values within 1.5 × IQR, and values beyond that are flagged as outliers.
• The rain shows raw values. For groups with more than about 40 observations, an evenly spaced sample across the sorted data is drawn, always including the minimum and maximum.

How to read a raincloud plot

• A tall, narrow cloud means values cluster tightly; a wide or flat cloud means high variability.
• Two humps in a cloud suggest a bimodal distribution, such as responders and non-responders mixed in one group.
• Compare medians with the box plots, and check the rain for gaps, clusters, or single extreme values that drive the mean.
• With fewer than about 10 points per group, trust the raw points more than the smooth cloud.

Tips and common mistakes

• Use long format. If your groups are separate columns, reshape them into one group column and one value column first.
• Keep to 8 groups or fewer; each group needs at least 3 values, and 15 or more gives a more stable density.
• The plot is horizontal by default. Write "vertical" for groups along the x-axis, "order by median" to sort groups, or "show mean" to add mean diamonds.
• Density clouds extend slightly past the data range because of kernel smoothing. Report n and the test you used in the caption, since the plot itself shows no p-values.

When to use a raincloud plot

• Use it instead of a bar chart with error bars when reviewers or co-authors want to see the underlying data.
• It suits reaction times, physiological measurements, questionnaire scores, and growth data with roughly 5 to 500 observations per group.
• With thousands of points per group the rain turns into a solid band, so a violin plot or box plot alone is often cleaner.
• For repeated measures, a raincloud shows each timepoint distribution but not individual trajectories; add a paired line plot if within-subject change matters.