Color in Data Visualization

STAT 80B Week 2 - Thursday

Prof. Marcela Alfaro Córdoba

Statistics - UCSC

10 Mar 2026

Introduction

A Tale of Two Maps

Rainbow colors

Sequential colors

Same data. Different colors. Which is easier to read?

Why Color Matters

Bad color choices can:

• Hide important patterns
• Create patterns that aren’t there
• Exclude 8% of your audience (colorblind viewers!)
• Make your graph ugly 😅

Good news: There are simple rules to follow!

Today’s Mission

Learn three types of color scales:

1. Qualitative: For categories (apples vs. oranges)
2. Sequential: For “more” and “less” (cold to hot)
3. Diverging: For opposites (profit vs. loss)

Plus: How to make sure colorblind people can see your work!

Part 1: Categories = Different Colors

Qualitative Scales

The Job: Tell Things Apart

Scenario: You’re showing favorite pets in a survey.

• Dogs
• Cats
• Birds
• Fish

You need colors that are obviously different from each other.

Rules for Qualitative Colors

1. Distinct: Easy to tell apart at a glance
2. Equal: No color screams “LOOK AT ME!” more than others
3. No order implied: Don’t go light → dark (that suggests more/less)

Good examples: Red, blue, green, orange
Bad examples: Light blue, medium blue, dark blue (that’s ordering!)

ColorBrewer is Your Friend

ColorBrewer2.org has tested palettes!

ColorBrewer qualitative palettes

The top section shows qualitative palettes (all equal, all different).

Think-Pair-Share (3 minutes)

You’re making a map of U.S. states colored by region: - West - Midwest
- South - Northeast

Which color combination would you choose?

A. Light green, medium green, dark green, darkest green
B. Red, blue, yellow, purple
C. Red, light red, orange, light orange

Discuss: Which follows the “no implied order” rule? Why are the others problematic?

What We Discovered

• Option A: NO! Suggests ordering (West is “best”?)
• Option B: YES! All distinct, no hierarchy
• Option C: NO! Red/orange are related, implies groups

Remember: Use contrasting colors for categories, not shades of the same color!

Part 2: Showing “More” and “Less”

Sequential Scales

When Numbers Have Direction

Some data naturally goes from low to high:

• Temperature (cold → hot)
• Income (poor → wealthy)
• Test scores (low → high)
• Population density (sparse → crowded)

Now you WANT colors that show order: light → dark

Sequential Color Scales

Light to dark = low to high

Light colors = low values

Light yellow = few people

Dark colors = high values

Dark blue = many people

Your brain instinctively reads this: “More color = more stuff”

One Color vs. Multiple Colors

Single hue:

Light blue → Dark blue

• Simple
• Less contrast
• Very intuitive

Multi-hue:

Yellow → Orange → Red
or
Purple → Blue → Green → Yellow

• More contrast
• Easier to distinguish values
• More interesting

Popular Sequential Palettes

• Viridis: Purple → green → yellow (colorblind-safe!)
• Blues: Light blue → dark blue (classic)
• YlOrRd: Yellow → orange → red (heat map)

All of these work great for showing quantities that go from low to high.

Think-Pair-Share (5 minutes)

You’re mapping average income by county.

Option A: Rainbow colors (red, orange, yellow, green, blue, purple)
Option B: Light yellow → dark blue
Option C: Red, green, blue, purple (random distinct colors)

Discuss: Which option clearly shows “low income to high income”? What’s wrong with the rainbow?

What We Discovered

• Option A (Rainbow): CONFUSING! Is green “more” than yellow? Hard to tell!
• Option B (Light → Dark): PERFECT! Clear direction
• Option C (Random colors): NO! These are for categories, not amounts

Golden rule: If your data has order, your colors should too!

5 Minute Break ☕

Stretch, check your phone, grab water!

Part 3: Showing Opposites

Diverging Scales

When There’s a Meaningful Middle

Some data has a critical center point:

• Temperature relative to freezing (above/below 32°F)
• Profit/loss (above/below $0)
• Political lean (Democrat ← middle → Republican)
• Agreement scales (strongly disagree ← neutral → strongly agree)

You need colors that show TWO directions from the middle.

Diverging Color Scales

Light in the middle, dark at both ends, different colors for each direction.

Dark Blue ← Light Blue ← WHITE → Light Red → Dark Red
   (very cold)    (cold)  (neutral)  (hot)   (very hot)

The pattern: Two sequential scales stuck together at a shared middle!

When to Use Diverging Scales

Perfect for:

• Positive vs. negative numbers
• Above vs. below an average
• Agree vs. disagree
• Gain vs. loss

Key question: Is there a meaningful zero/middle point? If yes → diverging scale!

Popular Diverging Palettes

From ColorBrewer:

• RdBu: Red ↔︎ white ↔︎ Blue (classic, but problematic for colorblind!)
• BrBG: Brown ↔︎ white ↔︎ Blue-Green (better!)
• PiYG: Pink ↔︎ white ↔︎ Yellow-Green (colorblind-safe!)
• PuOr: Purple ↔︎ white ↔︎ Orange (good contrast)

Think-Pair-Share (5 minutes)

You’re showing election results: Percent voting for Candidate A.

• 0% = everyone voted for Candidate B
• 50% = tied
• 100% = everyone voted for Candidate A

Which color scale?

A. Light gray → dark gray (sequential)
B. Red ↔︎ white ↔︎ Blue (diverging)
C. Red, blue, green, yellow (qualitative)

Discuss: Why does the 50% midpoint matter here?

What We Discovered

• Option A: Misses the point! Can’t see who won
• Option B: PERFECT! Clearly shows majority one way or the other
• Option C: Wrong tool—these are for categories

The white middle makes 50-50 splits obvious!

Part 4: The Colorblindness Challenge

Designing for Everyone

The Reality

About 8% of men (1 in 12) have colorblindness.

In a class of 30 students, ~2 can’t see certain color combinations!

Most common: Red-green colorblindness (can’t easily distinguish red from green)

What Colorblind People See

What you see: - Red vs. Green = obvious

What they see: - Both look brownish/yellowish - Maybe can’t tell them apart at all

Simulation of colorblindness

The Red-Green Problem

Never use these combinations:

• Red vs. Green for binary choices
• Traffic light metaphors (red = bad, green = good)
• Red-green diverging scales

Better alternatives:

• Blue vs. Orange
• Purple vs. Green
• Pink vs. Yellow-green

Good News: Some Scales Just Work!

Colorblind-safe by design:

• Viridis family: Purple → yellow (changes in both color AND brightness)
• Okabe-Ito palette: Specifically designed for colorblindness
• Single-hue sequential: Light → dark works for everyone
• Blues/purples: Generally safe

The “Add Redundancy” Trick

Don’t rely on color alone! Add a second cue:

• Shapes: Circle vs. square vs. triangle
• Line types: Solid vs. dashed vs. dotted
• Labels: Directly label important elements
• Patterns: Stripes, dots, crosshatch

Example: In a line graph with 3 groups, use color AND different line types.

Activity Time! (15 minutes)

Real hands-on practice:

1. Open ColorBrewer2.org
2. I’ll show you 2 different plots (scenarios)
3. For each plot:
   • Decide: Qualitative? Sequential? Diverging?
   • Choose a palette from ColorBrewer
   • Check the “colorblind safe” box
   • Write down: Your palette name and why you chose it

Work in pairs. We’ll share answers together afterward!

Scenario 1

Map showing: Average rainfall by state (ranges from 5 inches to 60 inches per year)

Your tasks:

1. What type of scale? (Qualitative / Sequential / Diverging)
2. Go to ColorBrewer → pick that scale type
3. Select “colorblind safe”
4. Choose a palette (write down the name)
5. Why is this a good choice?

Scenario 2

Chart showing: Change in employment from 2023 to 2024 by industry

• Some gained jobs (positive %)
• Some lost jobs (negative %)
• Values range from -15% to +20%

Your tasks:

1. What type of scale? (Qualitative / Sequential / Diverging)
2. Go to ColorBrewer → pick that scale type
3. Select “colorblind safe”
   
4. Choose a palette (write down the name)
5. Why does the zero point matter?

Let’s Discuss!

Scenario 1 - Rainfall:

• Scale type: Sequential (low to high rainfall)
• Good choices: Blues, YlGnBu, YlOrRd
• Why: Shows clear direction from dry to wet

Let’s Discuss!

Scenario 2 - Employment Change:

• Scale type: Diverging (loss vs. gain, with zero in middle)
• Good choices: BrBG, PiYG, PuOr (NOT RdBu!)
• Why: Need to see negative vs. positive clearly, with neutral middle

Closing

What We Learned Today

1. Qualitative scales: Different colors for categories (no order)
2. Sequential scales: Light → dark for low → high values
3. Diverging scales: Two directions from a meaningful middle
4. Colorblindness: 8% of men affected, avoid red-green!
5. Always test at ColorBrewer or with simulators

The Three Questions

Before choosing colors, ask yourself:

1. Am I showing categories OR amounts?
2. If amounts, is there a meaningful middle point?
3. Will colorblind people be able to see this?

Answer these → you’ll pick the right scale every time!

Tutorial Today

We’ll practice in code:

• Applying these color scales to real data
• Switching between scale types
• Testing visualizations for colorblindness
• Changing coordinate systems (recap from Tuesday!)

Assignment Reminder

Due next Tuesday: Concept Map #1

Connect these ideas:

• Data types (quantitative, categorical)
• Visual encodings (position, color, size, shape)
• Scales (linear, log, color scales)
• Coordinate systems (Cartesian, polar, maps)

Hand-drawn, 1 page. See Canvas for details!

Questions?

I’ll be here if you need me!