Strip / Jitter Plot

A strip plot (also called a dot chart or jitter plot) shows every individual data point. With small n —typical in animal experiments, cell biology, and pilot clinical studies —this is the only honest representation: there are too few data points for quartile or density estimates to be meaningful. Major journals now require or strongly prefer raw data display for n < 20.

STEP 1 —When to Use

Use a strip/jitter plot when: n < 20 per group (the default in most animal experiments); you want to make all data visible to the reader; individual data points have scientific meaning (e.g., each represents a patient or animal); the data have outliers that should be visible without being labeled as statistical outliers.

Do not use a strip/jitter plot when: n > 100 (overplotting makes individual points invisible —use violin or 2D density); you have so many groups that the plot becomes unreadably wide; points are categorical rather than continuous (use a dot chart with counts instead).

[Image placeholder: Side-by-side comparison: left —bar chart (mean ﾂｱ SEM) for 2 groups with n=7 each, showing clean-looking bars that appear to support the conclusion. Right —same data as strip plot, showing one outlier in each group that reverses the apparent trend when visible. The strip plot reveals what the bar chart hides.]

Bar chart (left) vs. strip plot (right): same data with n=7. The strip plot reveals that the "effect" in the bar chart is driven by a single outlier in each group.

STEP 2 —Journal Requirement

Each point must represent one independent biological unit —not a technical replicate (Replication Type Declaration required).
n per group must be labeled on the axis or in the figure legend.
If a summary statistic (mean, median) is overlaid, the type of error bar (SD, SEM, CI) must be explicitly stated.
ARRIVE 2.0 explicitly recommends raw data display for all animal studies.

STEP 3 —Licklider Implementation

Input

Continuous outcome and group variable
Subject ID for paired data (connecting lines between matched observations)
Jitter amount: auto-calculated to prevent overplotting based on n

Output

Individual points with controlled jitter (seed fixed for reproducibility)
Overlaid mean or median with error bar (type labeled automatically)
Connecting lines for paired/longitudinal data (auto-detected from subject ID)
Color per group with shape per secondary grouping variable (batch, sex, etc.)
n per group labeled below each group axis tick
Significance annotations with p-value, effect size, and CI

Style Options

Beeswarm layout (points arranged like a bee swarm, no overlap)
Sina plot (points within violin boundary —combines strip + violin)
Raincloud plot (half violin + strip + box —Nature Methods style)

[Image placeholder: Licklider strip plot with 2 groups (n=8 each). Shows beeswarm-layout points, overlaid mean ﾂｱ 95% CI, connecting lines between paired observations, n=8 labeled under each group, and significance bracket with p-value and Cohen's d.]

Licklider strip plot: beeswarm layout, mean ﾂｱ CI overlay, paired connection lines, and significance bracket.

STEP 4 —Draft Output (Draft / Needs review)

Individual data points are shown for each group (n = 8 biologically
independent experiments per group). Horizontal bars show the mean;
error bars show the 95% confidence interval. Connecting lines link
paired observations from the same experimental batch. Statistical
comparison by paired t-test: t(7) = 3.9, p = 0.006, Cohen's d = 1.38
(95% CI [0.51, 2.21]).