Data Visualization for Science

---
name: data-visualization-science
description: "This skill guides you in crafting clear, accurate scientific visuals by choosing appropriate charts, labeling clearly, and visualizing uncertainty."
metadata:
  tags: "science-communication, data-visualization, chart-design, uncertainty-visualization, time-series, small-multiples, visual-encoding"
  source: "https://skilldb.dev/skills/science-communication-skills/data-visualization-science"
  pack: "science-communication-skills"
  category: "Journalism & Communications"
---

# Data Visualization for Science

## When to use this skill
Use when the user says things like:
- "How do I visualize scientific data clearly?"
- "I need to choose the right chart for my data."
- "Show uncertainty in my visualization."
- "Create small multiples to compare categories."
- "Annotate charts with labels and units."


## Core Philosophy

Data visualization makes the invisible visible — transforming numbers into patterns, trends,
and stories that the human eye and brain can process instantly. In science communication,
visualization serves truth: it should reveal what the data actually shows, not what the
creator wants it to show. A good visualization enables the reader to understand the data
independently, not merely to accept the author's interpretation.

## Key Techniques

- **Chart type selection**: Match the data relationship (comparison, trend, distribution, composition) to the appropriate chart form.
- **Visual encoding**: Use position, length, area, and color to encode data values in order of perceptual accuracy.
- **Annotation and labeling**: Add context directly to visualizations — labels, callouts, and annotations.
- **Small multiples**: Use repeated small charts to show patterns across categories or time periods.
- **Uncertainty visualization**: Display confidence intervals, error bars, and probability distributions.
- **Interactive exploration**: Design visualizations that allow users to filter, zoom, and explore data.

## Best Practices

1. Start with the message. What should the reader understand from this visualization?
2. Choose the simplest chart that communicates the data relationship. Complexity should serve clarity.
3. Label axes clearly with units. Unlabeled charts are uninterpretable.
4. Start numerical axes at zero for bar charts to prevent visual distortion of differences.
5. Use color purposefully — for highlighting, grouping, or encoding values, not for decoration.
6. Show the data. Do not let design elements obscure or distort the actual values.
7. Include uncertainty. Showing only point estimates without ranges misrepresents scientific knowledge.

## Common Patterns

- **Time series**: Line chart showing how measurements change over time with confidence bands.
- **Comparison chart**: Bar or dot plot comparing values across categories with error bars.
- **Correlation plot**: Scatter plot showing relationship between two variables with trend line.
- **Geographic distribution**: Choropleth or point map showing spatial patterns in data.

## Anti-Patterns

- Using 3D charts — they distort perception and add no information.
- Truncating axes to exaggerate small differences.
- Using pie charts for more than 3-4 categories or for comparing across groups.
- Decorating with unnecessary visual elements (chartjunk) that obscure the data.