The chart library behind the data stories

Line charts for showing how a quantity changes over time. Six variants from generic to survey-flavored: a simple (year, value) series with optional confidence intervals, the GSS demographic-split chart with presidential-term reference areas, an indexed mode that rebases every series to 100 at a baseline year, a metadata-panel layout, a small-multiples grid, and an Observable Plot rendering.

FRED-style economic time series with recession-band shading. Three flavors covering the common use cases: dual-axis comparison for metrics on different scales, index-rebased comparison for relative growth, and economic-shaded with a brush selector for time-range filtering.

Each series' value at two time points joined by a single line, with endpoint confidence intervals, automatic top-mover labeling with collision-managed endpoint labels, and per-series focus. Use it to show direction and magnitude of change at a glance — movement, not a ranking.

State-level bar charts for ranking US states on a single metric. Two implementations: a sortable Recharts version with search and expand affordances, and an Observable Plot version with built-in sorting. Both built around BRFSS state-level data but accept any 50-row state dataset.

Choropleth maps in Observable Plot using the US Albers projection. The primary variant covers US states with built-in TopoJSON; alternates handle generic geographies (any TopoJSON or county-level resolution), European countries, and ZIP-code resolution. All variants are responsive to container width and accept standard TopoJSON / GeoJSON inputs.

A US county choropleth rendered as inline SVG from a caller-supplied TopoJSON, with theme diverging or sequential ramps, state borders, and a hover tooltip. Use it for print-quality county maps in articles; for zoomable, high-cardinality maps reach for the MapLibre engine instead.

The distribution family: histogram, violin, KDE density (single / overlay / combined), and split bar. Use a histogram when bin counts matter, a density plot when the shape matters, a violin when you want both shape and summary, and a split bar to highlight a subgroup difference within each category.

A principal-component biplot: observation scores as a quiet cloud plus variable loadings as labeled, focusable arrows with collision-managed leader labels. Use it to read which variables drive each component and how observations spread along them.

A ridgeline chart of per-group distributions with an optional mirrored violin mode, median rules, and direct labels. Region focus (hover, click, or keyboard) dims the other groups. Use it to compare the shape of a distribution across a handful of groups without a legend.

A tabbed dashboard for one BRFSS health measure broken down by demographic axis — age, education, income, race/ethnicity — with a one-click toggle between dot, line, and bar renderings of the same levels. Use it when the question is 'who is most affected?': the tabs let a reader walk one prevalence estimate across every demographic lens without leaving the figure.

Three chart shapes (line, bar, dot) for showing one outcome across five demographic axes — Age, Education, Income, Race, Gender — from a single dataset shape. Includes the multi-panel composite for showing the same outcome on all five axes side-by-side with consistent scales.

One row per category with a connector between two endpoint dots — the visual gap between two series (men vs women, then vs now), with value labels and an optional ratio note. Use it when the distance between two values is the finding.

A two-panel composite for visualizing GSS split-ballot wording experiments. The top panel shows pooled-across-years estimates by a moderator (party, race, age) as a paired-dot "dumbbell" plot — one dot per wording, connected by a gap label in percentage points. The bottom panel is the 1984–2024 time-trend with both wordings overlaid and design-adjusted 95% CIs. Built around the canonical "natfare" (welfare) vs "natfarey" (assistance to the poor) experiment, but the data shape generalizes to any GSS NAT___ / NAT___Y pair.

Scatterplots with OLS regression line, 95% confidence band, and R² display. Four flavors: a generic Recharts version built for European Social Survey data, plus three Observable Plot variants tuned for health-survey two-variable plots (basic, faceted by a third variable, regression overlay).

Odds ratios for several predictors compared across strata, faceted by term and colored by stratum, on a log axis with an OR = 1 reference. Use it to read how an effect changes across subgroups (e.g. a model refit within each age band).

A square correlation matrix ordered by hierarchical clustering so correlated blocks sit together, with topic-colored labels and a diverging-ramp legend. An optional second strip shows how each measure tracks demographic context variables. Use it to survey the structure of many pairwise relationships at once before committing to a single scatter.

A 0-100 composite score as a semicircular gauge with an end knob, the arc colored from worse to better through the theme's sentiment roles. Reach for it when a single headline number needs an at-a-glance read of how good or bad it is, not a full chart.