Effective Data Communication
Making a chart is easy. Making one that communicates clearly takes deliberate choices — the right type, honest axes, good color, and a narrative that carries the reader from data to insight.
Choosing the Right Chart
Three factors determine the right chart: data type (continuous, categorical, time-series), number of variables, and the message you want to convey.
Chart Selection
- Bar chart — compare categories (start at zero)
- Histogram — distribution of one continuous variable
- Scatter plot — relationship between two continuous variables
- Line plot — one axis is time
- Heatmap — many variables or a time × category matrix
Tufte’s Rule
Show the data, and nothing but the data. Edward Tufte’s core insight: every element of a chart should earn its place by encoding information. If removing it makes the chart clearer, it should be removed.
Data–Ink Ratio
The data-ink ratio is the fraction of a graphic’s ink devoted to non-redundant data. Maximize it by erasing chart borders, heavy grid lines, decorative fills, and redundant labels. Each removal makes the data stand out more.
Misleading Axes
Bar charts must start at zero — bar length is the encoding, and length is always measured from the baseline. Truncating the y-axis can make a 4% difference look like a 5× difference.
- Dual axes — can make any two lines appear to track each other
- Cherry-picked time ranges — starting at a low to show maximum growth
- Reversed y-axis — plotting downward what everyone expects to go upward
Cherry-Picking
Cherry-picking means showing only the data that supports your conclusion while hiding contradictory evidence. It is invisible to the viewer — making it one of the most pervasive forms of visual dishonesty.
Color Palettes
Match the palette to the data type:
- Sequential — single hue, light→dark; for ordered quantities with a zero
- Diverging — two hues from a midpoint; for correlations or anomalies
- Categorical — distinct hues, no implied order; for nominal groups (≤8–10)
Colorblind-Safe Charts
About 8% of men have color vision deficiency. A chart that relies on red vs. green is unreadable to a significant fraction of your audience.
- Use Okabe-Ito palette for categorical data
- Use viridis / cividis for sequential data
- Encode with color + shape — never color alone
Storytelling with Data
A data story has three parts:
- Context — what question are we answering?
- Data — one clear message per chart
- Insight — the “so what?” stated explicitly
What you learned
Effective data communication combines the right chart type, Tufte’s data-ink principles, honest axes, colorblind-safe palettes, and a narrative that drives from data to decision.