Lecture 7

Color Part 3 - Design Considerations

Agenda

1. Discussion of Reading

2. Activity (Attention)

3. Visual Processing

4. Activity (Preattentive Attribute)

5. Gestalt Principles of Perception + Activity

6. Explain Homework

ICEBREAKER

Most recent life-hack or life-skill you learned

Looking Ahead

4/7: Maps, pt 1

4/14: Maps, pt 2

4/21: Final Project Preparation

Final Project Description

Description: For the final project you will create a report in a similar style to the NYT articles we have viewed in class. Using data of your choosing, your report will include at least 3 different kinds graphics (scatterplot, histogram, map, etc.).

For each visualization there should be:

1. A description of what the graphic reveals about the data

2. A description of certain design decisions you made in order to produce the graphic(i.e. type of plot, color scheme, theme elements, etc.)

DURING class on Tuesday 4/28 you will give a brief 3-5 minute presentation of your report via google slides.

Final Project Elements

• Final Project Proposal

  • Type: Google doc pdf
  • To: BCourses
  • Due: 4/20, 11:59 PM

• Presentation:

  • Type: google slides
  • Due: BCourses
  • Due: 4/27, 11:59 PM

• Coding Materials:

  • Type: QMD file + csv of your data
  • To: BCourses
  • Due: 4/27, 11:59 PM

Dataset Guidelines

Make sure your data falls within the project guidelines:

• You may not use a data set that has already been used in the course:

  • Ex: Iris, mtcars, storms, etc.

• You may use multiple data sets if they are related:

  • Ex. If you are looking at wealth across the world you could use a data set for country GDP, and a separate dataset for poverty rates. 

• You may use your own data from an app, device, or website

Suggestions for Finding Data Sets:

• Github repository

• Microsoft Tableau Suggestions

Datasets Continued

• Contact us by 4/14 if you need help finding a data set!

• Note: You may need to manipulate the data set you download before using it

If your data is not in a format you can easily work with contact us by 4/14 so we can help you.

Final Project Proposal

• Due: 11/20

• Description: In 4-5 sentences provide a brief overview of your project. You should include the topic of your project, what dataset(s) you are using, and a rough plan of the three different visualizations you will create. You should also let us know if you would like additional resources for certain aspects of your project. (Ex. You want to include a visualization that we have not covered in class such as an interactive map).

• File type: Google doc pdf

• oSubmit to: BCourses

Questions?

Discussion of Reading

Discuss in Groups of 2–3

Discuss the reading graphic:

1. Review: What data is shown? How is color used to encode ordinal, binary, or nominal data?

2. Praise: What’s one thing you liked about how the graphic portrays data?

3. Critique: What’s one change that would make it better?

Today’s Learning Goals

1. Explain why preattentive attributes guide the viewer’s eye before conscious thought
2. Apply Gestalt principles to design cleaner, more effective visualizations

Part 1: Visual Processing

Find the Difference

Look at each pair of images. What makes it easy or hard to spot the change?

Ron Rensink’s Examples

Ron Rensink’s Examples

Ron Rensink’s Examples

Prof. Sanchez’s Example

What Just Happened?

• Change blindness: our brains don’t store a perfect snapshot of the world
• We only process what we’re focused on or what jumps out
• Small changes in unattended regions are completely invisible
• This demonstrates the power and limits of human visual perception

Tip

Takeaway for dataviz: If something is important, you can’t rely on your audience to find it. You must direct their attention to it.

How the Visual System Works

Visual processing has two components:

Eyes — image receptors

• Capture raw pixel information
• No interpretation yet

Brain — image processor + interpreter

• Imposes structure and meaning
• Fills in gaps, groups objects, detects patterns
• Decides what to pay attention to

Visual Memory: Three Stages

Memory Type	Analogy	Duration
Iconic memory	Buffer / temp storage	< 0.5 sec
Short-term (working)	RAM	~20 sec
Long-term	Hard disk	Permanent

Good visualizations work with these limits — not against them.

Part 2: Preattentive Processing

Iconic Memory = Preattentive Processing

Iconic memory is a waiting room where each snapshot waits to be passed to short-term memory.

• Processing is rapid (under 250ms)
• It is parallel so you see the whole image at once, not pixel by pixel
• It is unconscious, happens before you “decide” to look
• It relies on contrast between features

The Power of Preattentive Processing

Can you count the 6s?

The Power of Preattentive Processing

Now try:

What changed? A single visual attribute, color, made the target pop out without counting.

This is preattentive processing at work.

Two Types of Visual Search

Preattentive (fast)	Attentive (slow)
Parallel — whole image at once	Serial — scan one element at a time
< 250ms regardless of # items	Time increases with number of items
“Pop-out” effect	Effortful, exhausting

The Preattentive Attributes

These visual features are processed before conscious attention:

Form

• Shape
• Size
• Orientation
• Line length / width
• Enclosure / grouping
• Added marks (fill, texture)

Color

• Hue (the “color” itself)
• Intensity / saturation
• Luminance / brightness

Spatial

• 2D position
• Depth cues (stereoscopic)
• Motion

Activity: Spot the Preattentive Attribute

Which attribute makes the target pop out instantly? Which requires search?

Try each example below and think about why it’s fast or slow.

Design Rule #1

Use preattentive attributes to direct attention

If something is the most important part of your chart, an outlier, a threshold, a key category, encode it with a preattentive attribute (usually color or size) so it pops out without effort.

Don’t make your audience search for the point.

Design Rule #2

Don’t overload the preattentive channels

Using too many colors, sizes, or shapes forces your viewer into attentive (slow) search mode, exactly what you were trying to avoid.

Rule of thumb: highlight one thing with a strong preattentive signal. Keep everything else neutral (gray is your friend).

Memory and Visualization

Understanding how our brains pay attention, and how our memory system handles incoming information, goes a long way toward making better visualizations.

Key question to ask yourself every time: “Am I making my audience work harder than necessary?”

Part 3: Gestalt Principles

What is Gestalt Theory?

Gestalt (German: “whole” or “form”)

• Developed in Germany in the 1920s
• A theory of form processing:how we organize visual input into meaningful wholes
• The whole is different from the sum of its parts
• We don’t see isolated pixels but rather objects, groups, and relationships

Good design works with these principles

The 8 Main Gestalt Principles

Principle	The Brain Assumes…
Proximity	Things close together belong together
Similarity	Things that look alike are related
Continuity	Lines and curves continue in the smoothest path
Closure	Incomplete shapes are seen as complete
Connectivity	Objects joined by lines are related
Enclosure	Objects sharing a boundary form a group
Symmetry	Symmetric arrangements feel stable and grouped
Figure & Ground	We separate foreground from background

Gestalt in practice: FiveThirtyEight

We’ll walk through several published charts and identify the Gestalt principles at work.

Why Democrats Struggle to Mobilize a Religious Left

Principle: Connectivity

In dataviz: Lines in a line chart connect points across time. The connection implies continuity and relationship even though the data points are discrete.

Principle: Proximity

In dataviz: Grouped bar charts use proximity to show that bars within a group belong together (same time period, same category). Spacing between groups signals separation.

Principle: Similarity

In dataviz: A consistent color for “Democrats” across all charts in a report uses similarity to signal the same category.

Bryce Harper May Already Be Past His Prime

Principle: Continuity

The eye follows the smoothest path through a series of points so we perceive a trend, not individual values, and this is why line charts are so good at showing trajectories

How Unpopular is Donald Trump?

Chunking

Chunking is the process of grouping individual pieces of information into meaningful wholes to improve retention.

• Working memory can hold ~4 “chunks” at a time
• Good charts chunk information visually (through spacing, color, enclosure, or labels)

The 5 Corners of the 2020 Democratic Primary

Principle: Enclosure

Objects sharing a common region are perceived as belonging together.

In dataviz: Shaded regions, background panels, and borders create groups without explicit labels. The boundary does the cognitive work for you.

Principle: Figure & Ground

We instinctively separate a “figure” (foreground object) from its “ground” (background).

In dataviz: A white chart background is the “ground.” Your data marks are the “figure.” Gridlines should recede into the background, not compete with your data.

High-contrast accent colors push elements into the foreground.

Putting It All Together

A checklist for every chart you make

• Is the most important element highlighted with a preattentive attribute?
• Is everything else visually muted (gray, thin, light)?
• Are related elements close together (proximity)?
• Are same-category elements visually consistent (similarity)?
• Are groups enclosed or separated with whitespace or borders (enclosure)?
• Do your lines create a clear figure against a receding background (figure/ground)?
• Are you working with these principles?

In-Class Activity

Gestalt Principles in the Wild

With a partner, identify the Gestalt principles in the graphics on the worksheet.

For each chart, note:

1. Which principle(s) are at work
2. How the designer used it (spacing, color, enclosure, etc.)
3. Whether it helps or hurts the visualization

Homework

HW 4: Recreating a Screen Time Bar Chart

• Reshape and plot your own screen time data using ggplot2
• Apply what you know about color and preattentive attributes to make a clean, readable chart

Reading for next week:

“We Still Don’t Believe How Much Things Cost” — WSJ Link

Final Project:

• Start looking for datasets now
• Proposal due 4/20
• Email us if you need help finding or cleaning data