Lesson 1: The Building Blocks of All Charts

Learning Objective: By the end of this lesson, you will be able to:

  • Grasp how the human visual system functions and its role in interpreting charts, including common cognitive biases that affect our perception of visual information.
  • Identify the key building blocks of the most common data visualizations including heatmap tables, bar charts, pie charts, line charts, histograms, scatter plots, and maps.
  • Understand how the four common data scale types map to visual traits like length, size, and color to enhance clarity and accuracy in data presentation.
  • Critically assess the effectiveness of different visual representations of data.

Knowledge Check

Click the accordions below to expand the question and answer.

Question
This image depicts two horizontal lines labeled "A" and "B." On line A, there is a yellow circle positioned towards the right end of the line. On line B, there is a gray circle located closer to the center.

What concept is being demonstrated in this image, where the yellow circle on line A stands out compared to the gray circle on line B?

a. Magnitude Channel

b. Identity Channel

c. Optical Illusion

d. Pop-out Effect

Answer

d. Pop-out Effect

The Pop-Out Effect

The “pop-out effect” happens when a particular visual element in a data visualization instantly grabs attention because of a distinctive feature, such as its color, shape, or size. This effect enables our visual system to quickly and easily recognize the standout element, directing our focus to it almost instinctively. When used effectively in data visualization, the pop-out effect helps ensure that critical information is immediately noticed, allowing viewers to concentrate on the most important data with minimal effort.

Basics of Visual Perception

As light enters our eyes, our brain processes it into images, allowing us to recognize patterns, shapes, and colors. However, our visual system, while highly sophisticated, is not infallible. It can be easily deceived by optical illusions and overly complex visuals, leading to misinterpretations of what we see.

For example, when we look at this solid-colored rectangle placed against a gradient background, our brain might interpret it as also having a gradient, even though it’s uniformly colored. This is because our perception is influenced by the surrounding colors and shapes, causing the rectangle to appear as though it changes in tone from one side to the other. Such visual tricks highlight the limitations of our perception and remind us that what we see is not always an accurate reflection of reality.

This image features a grey gradient background transitioning from a darker shade on the left to a lighter shade on the right. Overlaid centrally on the gradient is a horizontal rectangular bar with a solid light gray color.
A horizontal rectangular bar with a solid light gray color.

These visual processing quirks can pose challenges, especially in data visualization, where clarity and accuracy are paramount. If the visual elements of a chart or graph are too complex or misleading, they can lead to incorrect conclusions. This underscores the importance of designing visuals that align with our brain’s natural processing capabilities. Recognizing these limitations in our visual perception is the first step toward mastering data interpretation.

The Building Blocks of Charts

Every chart, regardless of its complexity, is built from three essential components: variables that fall into different scale types, the visual marks that represent data, and the encodings that modify these marks. When these elements come together, they form a complete chart. 

This image is a graphical representation of the components that make up a chart. The top left header reads "Variables of Different Scale Types." The other three column headers are labeled Marks, Encodings, and Chart. The first column contains two labels: "CATEGORICAL" in salmon pink and "QUANTITATIVE" in blue. The second column shows different types of marks such as a dot, a squiggly line, and a square. The third column shows encoding methods with dotted outlines and colored circles in different sizes. The fourth column displays a bar chart. The layout implies a formula where combining variable types, marks, and encodings results in the creation of a chart.

Click through the tabs below to explore these components in more detail.

Variables are the characteristics tied to each data point in a chart. These variables fall into two main categories: categorical and quantitative.

  • Categorical variables can be either nominal or ordinal. Nominal variables classify data into groups without any specific order—like different types of fruit. For example, whether an apple is listed before or after a banana doesn’t affect the meaning, as there’s no inherent ranking between them. On the other hand, ordinal variables have a defined sequence, such as medal rankings. Listing them as silver, bronze, and gold wouldn’t make sense because their order is significant.
  • Quantitative variables, on the other hand, are numerical and can be further divided into interval and ratio scales. Interval scales maintain equal intervals between values but lack a true zero point, as seen with temperature in degrees Fahrenheit. Here, zero doesn’t signify the absence of temperature but is merely a reference point. Ratio scales, however, have both consistent intervals and an absolute zero, making them ideal for measuring quantities like length or counting items.
This image is a visual guide to different types of data scales. At the top, "CATEGORICAL" is labeled in a salmon pink banner, branching into "NOMINAL" and "ORDINAL" categories in the same color. Nominal data is represented by icons of an apple, a banana, and an orange, while ordinal data is shown with gold, silver, and bronze medals numbered 1, 2, and 3. Below, "QUANTITATIVE" is labeled in a blue banner, further divided into "INTERVAL" and "RATIO" categories. Interval data is illustrated by a thermometer for temperature, and ratio data by a yellow ruler. The layout uses connecting lines to indicate relationships and classifications among these data scale types.

Marks are the basic shapes or symbols—such as points, lines, or areas—that represent data within a chart. They are the building blocks that transform raw data into a visual format that can be easily interpreted.

For instance, when you look at a scatter plot, the individual dots represent the data points; in a bar chart, it’s the bars that convey the information. The specific type of mark chosen for a chart depends on the nature of the data being presented and what the chart aims to communicate.

This image illustrates how different graphical marks translate into specific chart types. On the left, three basic marks are shown: a blue dot, a light blue squiggly line, and a blue square. Each mark points to a corresponding chart type on the right through an arrow. The blue dot leads to a scatter plot filled with multiple blue dots, the squiggly line transitions into a line chart with a rising curve, and the blue square corresponds to a bar chart with three blue bars of varying heights. The design clearly demonstrates the concept of using simple visual elements to represent data in charts.

Encodings are the ways in which data is visually represented through the modification of marks in a chart. They are the visual attributes—such as position, color, size, or shape—that are applied to marks to convey additional information. For example, in a bar chart, the height of the bars (an encoding) represents the value of the data. Encodings are crucial because they determine how effectively a chart communicates its underlying data. By choosing the right encoding methods, you can make your visualizations clearer, more accurate, and easier for viewers to interpret.

This image displays examples of visual encodings for data. At the top, the word "Encodings" is followed by three illustrations: the first shows a dotted outline forming a square with two solid blue dots positioned at the corners, and the second depicts four overlapping circles in shades of blue, salmon, and yellow. Below, another example of encoding is shown with a sequence of overlapping blue circles, varying in size and overlap to create a sense of depth or hierarchy. The visuals serve to demonstrate how simple shapes and colors can be used to represent different data relationships or values.

Knowledge Check

Click the accordions below to expand the question and answer.

Question

Fill in the blanks: 

In a chart, _____ are the visual elements like points, lines, or areas that represent data, while _____ are the modifications to these elements, such as color, size, or position, that convey additional information.

a. marks, encodings

b. encodings, marks

Answer

a. marks, encodings

How to Pick an Effective Encoding

This idea of marks is part of a broader framework outlined by Tamara Munzner, a renowned expert in the field of data visualization. Munzner’s work delves into how different marks and their associated encodings can vary in their effectiveness, depending on how they align with our natural visual perception.

Munzner categorizes the ways data can be encoded into two primary types: magnitude channels and identity channels.

  • Magnitude Channels are used to represent ordered data, like quantities. These channels include attributes such as position on a common scale, length, and area. For example, using length to represent a value in a bar chart is a magnitude channel, as it visually conveys the magnitude of the data.
  • Identity Channels are used for categorical data, helping to differentiate between categories without implying order. Examples of identity channels include color, shape, and spatial region. In a pie chart, different slices might be colored differently to represent various categories, relying on color as an identity channel.
This image is a summary diagram illustrating the expressiveness and effectiveness of different encoding channels in data visualization, based on Tamara Munzner's framework. It is divided into two main categories: Magnitude Channels (Ordered Attributes) on the left and Identity Channels (Categorical Attributes) on the right.

Munzner’s framework also introduces the concept of Effectiveness Ranks, which help us understand that not all marks and encodings are created equal. Some methods are inherently more effective at conveying information in a way that aligns with how our brains naturally process visuals. For instance, using position along a common scale is considered one of the most effective ways to encode data because it allows for precise comparisons. On the other hand, using color saturation to represent quantitative differences is less effective because it’s harder for our visual system to accurately assess those differences.

In the four images below, you’ll see different encoding methods in action, ranked from least to most effective, from left to right. The first image uses color saturation to represent two data points, one of the least effective methods, making it hard to determine how much larger value A is than value B. The second image uses 2D area, which makes the comparison slightly easier but still not precise. The third image uses position along a common scale, allowing us to easily see that value A is about three times larger than value B. The fourth image combines all three methods—color saturation, 2D area, and position along a common scale—emphasizing the difference between value A and value B.

This image illustrates three different methods - color saturation, size, and position along a common scale. There are four images total. The first represents "Color saturation" and shows two circles labeled 'A' and 'B'. Circle 'A' has a deeper blue color than circle 'B'. The second section labeled "2D area" also displays two circles of the same color with 'A' being significantly larger than 'B'. The third section labeled "Position along a common scale" depicts two horizontal lines with circles on them; circle 'A' is positioned further to the right on one line, than 'B' on the second line. The final section combines all three encoding methods: color saturation, 2D area, and position along a common scale, where circle 'A' is larger, more saturated, and positioned further right compared to circle 'B'. This demonstrates the increasing effectiveness of each encoding method.

As we continue to explore different types of charts throughout this course, Munzner’s effectiveness ranks will serve as a crucial guide. By understanding which marks and encodings are most effective, you’ll be better equipped to select the appropriate visual elements for your data, ensuring that your charts are both clear and impactful. This knowledge will help you create visualizations that not only look good but also communicate information accurately and efficiently.

Knowledge Check

Click the accordions below to expand the question and answer.

Question

Which of the following best describes the difference between magnitude channels and identity channels according to Tamara Munzner’s framework?

a. Magnitude channels represent categorical data, while identity channels represent ordered data like quantities.

b. Magnitude channels are used for ordered attributes such as quantities, while identity channels are used to differentiate between categories without implying order.

c. Magnitude channels focus on color and shape, whereas identity channels focus on size and position.

d. Magnitude channels are less effective at conveying information than identity channels.

Answer

b. Magnitude channels are used for ordered attributes such as quantities, while identity channels are used to differentiate between categories without implying order.

Key Terms & Definitions

Hover over each card to flip it over and reveal its definition.

Categorical Variables

Variables that categorize data without implying any numerical value. They can be nominal, where there is no inherent order (e.g., types of fruit), or ordinal, where there is a specific order (e.g., medal rankings).

Effectiveness Ranks

A concept from Tamara Munzner’s framework that ranks different encoding methods based on their effectiveness in helping viewers understand data. Some encodings, like position on a common scale, are more effective at conveying information than others, such as color saturation.

Encodings

The visual modifications applied to marks, such as color, size, position, or shape, that convey additional information about the data.

Identity Channels

Encoding methods used to represent categorical data, helping to differentiate between categories without implying order. Examples include color, shape, and spatial region.

Interval Scale Variables

A type of quantitative variable where the intervals between values are consistent, but there is no true zero point (e.g., temperature in degrees Fahrenheit).

Magnitude Channels

Encoding methods used to represent ordered data, such as quantities. These include attributes like position on a common scale, length, and area.

Marks

The visual elements in a chart, such as points, lines, or areas, that represent data. Marks are the building blocks of any data visualization.

Nominal Scale Variables

Nominal variables are categorical variables that classify data into distinct groups or categories without implying any order or rank. Examples include types of fruit, colors, or countries. Each category is mutually exclusive, and there is no inherent sequence among them.

Ordinal Scale Variables 

Categorical variables that not only classify data into distinct groups but also imply a specific order or rank among the categories. However, the intervals between the categories are not necessarily equal. Examples include education levels (e.g., high school, bachelor’s, master’s) or satisfaction ratings (e.g., dissatisfied, neutral, satisfied).

Pop-Out Effect

A phenomenon where a specific visual element within a data visualization stands out immediately due to a unique property, such as color, shape, or size, making it easier to identify and focus on key information.

Quantitative Variables

Variables that measure quantities and can be further divided into interval and ratio scales. Interval scales have consistent intervals but no true zero (e.g., temperature in degrees Fahrenheit), while ratio scales have consistent intervals and a true zero point (e.g., length or count).

Ratio Scale Variables

A type of quantitative variable that has consistent intervals and a true zero point, allowing for meaningful comparisons of absolute quantities (e.g., length, weight).

Variables

The attributes or characteristics associated with each data point in a dataset. Variables can be categorical (nominal or ordinal) or quantitative (interval or ratio).

Further Learning

If you’d like to further your learning of the topics covered in this lesson, here are some resources for you to explore:

Lesson Content
Previous Quiz
Back to Course
Next Quiz