Publishable Figs Overview Distribution & Group Comparison ↳ Box Plot ↳ Violin Plot ↳ Strip / Jitter Plot ↳ Bar Chart (Mean ± SEM) ↳ Histogram ↳ Dot Plot (Estimation) ↳ 2D Density Contour Correlation & Regression ↳ Scatter Plot ↳ Regression Plot ↳ Bubble Chart ↳ Confidence Ellipse ↳ Bland-Altman Plot Multivariate & Dimensionality ↳ Heatmap ↳ Parallel Coordinates ↳ PCA Biplot ↳ Hierarchical Clustering Clinical & Biomedical ↳ Kaplan-Meier Curve ↳ ROC Curve ↳ Volcano Plot ↳ Forest Plot Composition & Trends ↳ Pie / Donut Chart ↳ Line Chart (Time Series) ↳ Stacked Bar Chart

Box Plot

The box plot summarizes five statistics: minimum whisker, Q1, median, Q3, and maximum whisker, with individual outlier points beyond 1.5テ悠QR. It is efficient, space-saving, and appropriate for n >= 10. Below n = 10, the quartile boundaries become unreliable and individual data points should be shown instead.

STEP 1 —When to Use

Use a box plot when: n >= 10 per group; the data may be non-normally distributed; you need to compare multiple groups in a small space; outliers are scientifically meaningful and should be individually visible.

Do not use a box plot when: n < 10 (the quartiles are unreliable —use a strip plot instead); you need to show the full distribution shape (violin plot is preferred); the data are bimodal and the box would obscure this (the median of a bimodal distribution can fall in a low-density region).

[Image placeholder: Box plot with 4 groups (n=15 per group), whiskers correctly labeled as 1.5テ悠QR, outlier points shown individually. Alongside: the same data as a strip plot for n=6 per group, showing why quartiles are unreliable at small n.]
Box plot appropriate for n=15 per group (left). Strip plot preferred for n=6 per group (right) where quartile estimates are unstable.

STEP 2 —Journal Requirement

  • Whisker definition must be stated in the figure legend: "Boxes show median and IQR; whiskers extend to 1.5テ悠QR; individual points beyond whiskers shown as outliers."
  • n per group must be stated in the legend or on the axis.
  • Statistical test and p-value or effect size with CI must accompany any inferential comparison.
  • Nature Methods and Cell require individual data points overlaid for n < 20, even on box plots.

STEP 3 —Licklider Implementation

Input

  • Continuous outcome variable and group variable from Variable & ID Mapping
  • n per group (auto-calculated from data)
  • Statistical comparison configuration (from Test Configuration Guard)

Output

  • Box: Q1窶轍3 (IQR), median line, notch option (95% CI for median, for visual significance testing)
  • Whiskers: 1.5テ悠QR (default, labeled), or min/max (user-selectable with label)
  • Outliers: individual points beyond whiskers, styled distinctly
  • Individual data overlay: auto-added with jitter for n < 20 (Nature/Cell compliance)
  • p-value brackets or significance stars positioned automatically above the relevant group pairs
  • Effect size + CI displayed in the legend or below the figure

Style Options

  • Notched box (median CI visible)
  • Paired box (connecting lines between matched observations)
  • Fill color per group (journal palette or custom; accessibility-checked)
  • Journal presets: Nature / NEJM / Science / Default
[Image placeholder: Licklider box plot with n=18 per group. Shows: notched boxes, individual data points overlaid with jitter, significance bracket with p-value and Cohen's d, n labeled below each box, and legend with whisker definition.]
Licklider box plot: notched boxes, individual data overlay (n=18), significance bracket with effect size, and auto-generated legend text.

STEP 4 —Draft Output (Draft / Needs review)

Data are presented as box plots showing median and interquartile range (IQR);
whiskers extend to 1.5テ悠QR; points beyond whiskers are shown individually.
Individual data points are overlaid (n = 18 per group). Statistical comparison
by Mann-Whitney U test; p = 0.003, Cohen's d = 0.81 (95% CI [0.34, 1.27]).

Example journal-style caption. Mann–Whitney is shown for illustration; rank-based group tests are not verified end-to-end in the current product — see Non-Parametric Alternatives.