Data Presentation

Data does not speak for itself -- it needs a translator. As an intelligence analyst, your job is to transform raw numbers, frequencies, and patterns into visual stories that decision-makers can absorb in seconds. In this lesson, you will learn how to choose the right chart type, avoid the cardinal sins of data presentation, and design visuals that clarify rather than confuse.

The single most important decision in data presentation is choosing the right visual format. The wrong chart type can actively mislead your reader, even when the underlying data is accurate.

The Chart Selection Framework

Ask yourself one question: what relationship am I trying to show? Your answer determines the chart type:

• Comparison between categories: Bar chart (vertical or horizontal). Use horizontal bars when category labels are long.

• Change over time: Line chart. Use multiple lines to compare trends across groups. Never use a bar chart for time series if you have more than six time periods.

• Part-to-whole relationship: Stacked bar chart or, sparingly, a pie chart. Pie charts work only when you have 2-4 slices and the differences are large.

• Correlation between two variables: Scatter plot. Add a trend line if the relationship is meaningful.

• Geographic distribution: Map (choropleth or dot map). Essential for intelligence products involving location-based threats.

• Flow or process: Sankey diagram or flowchart. Good for showing funding flows, organizational structures, or attack pathways.

Using a 3D pie chart to show seven categories is a war crime in data visualization. Edward Tufte would like a word with you.

When to Use a Table Instead

Charts are not always the answer. Use a table when:

• The reader needs exact values (financial figures, coordinates, dates).
• You are comparing many attributes across a few items (e.g., a capability comparison matrix).
• The data set is small (fewer than 10 data points) -- a chart adds visual overhead without adding clarity.
• You need to show mixed data types (numbers, text, dates) side by side.

When in doubt, ask: "Does my reader need the pattern or the precise numbers?" If they need the pattern, use a chart. If they need the numbers, use a table. If they need both, use a chart with a supporting data table below it.

Even the right chart type can mislead if the design is sloppy. Color choices, axis scales, and labeling decisions all shape how the reader interprets your data -- for better or worse.

Color Principles

• Use color to encode meaning, not to decorate. Red for threats or negative trends, green for positive outcomes, blue for neutral data. Be consistent across your entire report.

• Limit your palette. Three to five colors maximum. If your chart looks like a bag of Skittles, you have too many categories -- group some together.

• Consider color blindness. About 8% of men have red-green color deficiency. Use patterns, labels, or a colorblind-safe palette as a backup.

• Avoid traffic-light defaults. Red-yellow-green is tempting but overused and inaccessible. A sequential color scale (light-to-dark) often communicates magnitude more effectively.

Axis and Scale Integrity

Manipulating axes is the fastest way to lie with data -- intentionally or not.

Misleading

A bar chart showing attack frequency with a Y-axis starting at 95 instead of 0, making a rise from 97 to 103 incidents look like a 600% increase.

Honest

The same data with a Y-axis starting at 0, showing the rise in proper proportion -- a modest 6% increase. If the increase matters, annotate it; do not distort the scale to manufacture drama.

Truncating the Y-axis is the data visualization version of "technically correct but deeply misleading" -- the kind of thing that gets analysts a quiet talking-to from the quality review team.

A chart without context is just a pretty picture. Annotations transform a visual from "interesting" to "actionable" by telling the reader exactly what they should notice and why it matters.

What to Annotate

• Key inflection points: Where a trend changes direction. Label what caused the change -- "Ceasefire declared 14 Jun" or "New sanctions effective."

• Outliers: If one data point breaks the pattern, explain it. An unexplained outlier invites speculation.

• Thresholds: Add a reference line for meaningful benchmarks -- "Historical average," "Red line capacity," "Treaty limit."

• Source and date: Every visual needs a source citation and the date range of the data. No exceptions.

The Title Is Your Headline

A good chart title states the takeaway, not the topic. Compare:

Descriptive Title

"Monthly IED Incidents, 2024-2025"

Analytical Title

"IED Incidents Doubled After Ceasefire Collapse (Sep 2024)"

Use the "five-second rule": show your chart to a colleague for five seconds, then take it away. Ask them what the main message was. If they cannot tell you, your annotation and design need work.

Edward Tufte coined the term "chart junk" to describe all the visual clutter that adds no information: decorative gridlines, 3D effects, gradient fills, unnecessary legends, and clip art. In intelligence reporting, chart junk is not just ugly -- it is dangerous, because it distracts from the signal.

The Seven Deadly Sins of Intel Data Viz

• Sin 1: 3D effects. They distort proportions and make values harder to read. Always use 2D.

• Sin 2: Dual-axis charts. Two different Y-axes on the same chart invite misinterpretation. Use two separate charts instead.

• Sin 3: Rainbow color palettes. More than five colors turns your chart into abstract art. Group or filter categories.

• Sin 4: Pie charts for precision. Humans are terrible at judging angles. If the difference between slices matters, use a bar chart.

• Sin 5: Missing units. "200" means nothing without "200 incidents," "$200 million," or "200 kilometers." Always include units.

• Sin 6: Cherry-picked timeframes. Showing only the months that support your narrative destroys credibility when the full dataset tells a different story.

• Sin 7: No source attribution. Unsourced data is unverifiable data. Always cite where the numbers came from.

Apply the "data-ink ratio" test: what percentage of the ink (or pixels) in your chart represents actual data? Remove everything that does not. Gridlines, borders, background fills, and decorative elements should be minimized or eliminated.