All in One View

Content from Introduction to Data Visualization

---

Last updated on 2026-10-17 | Edit this page

Overview

Questions

• What is data visualization, and why does it matter for research?
• What can a chart or map show that a table cannot?
• How can a visualization mislead, even when the data are correct?
• How do exploratory and explanatory visualizations differ?
• Which tools are a good fit for beginners, and which come later?

Objectives

• Define data visualization and explain its purpose in research
• Describe the main advantages and risks of visualizing data
• Tell exploratory visualizations apart from explanatory ones
• Recognize common ways charts and maps mislead
• Compare tools for making charts and maps, from spreadsheets to Python and QGIS
• Apply a short set of design principles to your own figures

What is data visualization?

---

Data visualization is the graphical representation of data. Instead of reading rows and columns of numbers, the reader sees positions, lengths, areas, colors and shapes, and uses them to compare values, spot patterns and notice anything unusual.

A good visualization answers a question. “How did unemployment change after 2008?”, “Which counties have the most people over 65?” or “Is income related to commute time?” are all questions a chart or a map can answer faster and more memorably than a table.

Maps are a special kind of visualization: they use location as one of the visual channels. Most of this module is about maps, but the same principles apply to every chart, so we start there.

Why visualize?

---

1. Patterns are easier to see than to read. Our visual system is very good at comparing positions and lengths, noticing clusters, and picking out a value that does not fit. A trend across 50 monthly figures is obvious in a line chart and easy to miss in a table.
2. Visuals reveal what summary statistics hide. Very different datasets can share the same mean, variance and correlation and still look nothing alike when plotted. (Anscombe’s quartet is the classic example; search for it if you have not seen it.)
3. They catch data problems. Impossible values, missing records, duplicated rows and wrong units often show up immediately in a quick plot.
4. They communicate. A clear figure lets colleagues, community partners and policymakers grasp a result without reading your code or your tables.

Exploratory and explanatory visualization

• Exploratory figures are for you. They are quick, often rough, and you make many of them while you get to know your data.
• Explanatory figures are for others. They carry one clear message, with a title that states it, labeled axes, units, a legend and a data source.

Most of the figures in a research project are exploratory. The few that reach a paper, poster or report should be carefully designed explanatory figures.

Advantages and risks

---

Advantages	Risks
Fast comparison of many values	A misleading scale can exaggerate or hide differences
Reveals trends, clusters and outliers	Too much decoration (“chartjunk”) adds noise, not information
Exposes data quality problems	The wrong chart type can suggest relationships that are not there
Communicates across expertise levels	Poor color choices exclude readers with color vision deficiency
Makes results memorable	A polished figure can make weak evidence look convincing

Callout

How a correct chart can still mislead

The two bar charts below show exactly the same numbers. On the left, the vertical axis starts at 70, which makes the differences between countries look enormous. On the right, the axis starts at 0, and the countries turn out to be quite similar.

The same life expectancy data with a truncated axis (left) and a full axis (right)

Bar charts should almost always start at zero, because readers compare bar lengths. Line charts and dot plots can start elsewhere, because readers compare positions, but the axis must be clearly labeled.

Big data and visualization

---

Large datasets make visualization more important, not less:

• Volume. Nobody can read a million rows, but a histogram, density plot or map can show the whole distribution at once.
• Variety. Combining tables, text and locations often calls for layered figures, such as a map with a chart beside it.
• Many variables. With dozens of variables, small multiples (many small, identical charts, one per group) or scatterplot matrices let you compare them systematically.
• Interactivity. Zooming, filtering and hover tooltips let readers explore data that would overwhelm a static figure.

Large datasets also bring practical problems. Plotting millions of points can be slow and produce a solid blob of ink. Sampling, aggregating (for example into hexagons or counties) and transparency are the usual remedies.

Common chart types and what they are for

---

Question	Chart type
How does a value change over time?	Line chart
How do categories compare?	Bar chart (horizontal if labels are long)
How are values distributed?	Histogram, box plot
Are two numeric variables related?	Scatter plot
How do many variables relate to each other?	Correlation heatmap, scatterplot matrix
How does a whole divide into parts?	Stacked bar; pie chart only for 2 or 3 parts
Where does something happen?	Map (choropleth, proportional symbols, dots; see the map episodes)

Tools

---

Tool	Good for	Cost	In this curriculum
Excel or Google Sheets	Quick bar and line charts	Free or licensed	Not covered
Tableau Public	Interactive dashboards without code	Free public version	Not covered
Python (Matplotlib, Seaborn, Plotly, GeoPandas)	Reproducible charts and maps, from quick to publication quality	Free, open source	The notebooks in this episode
QGIS	Desktop mapping and cartographic layouts	Free, open source	The Foundations module’s Getting Started with QGIS episode
R (ggplot2, tmap, sf)	Statistical graphics and maps	Free, open source	Not covered
JavaScript (D3, Leaflet)	Custom interactive web graphics	Free, open source	Not covered

We use Python in Google Colab for charts and simple maps, and QGIS for designed maps, because both are free and widely used in the social sciences.

Principles of effective visualization

---

These principles come from the work of Edward Tufte, William Cleveland, Alberto Cairo and others.

1. Start with the question. Decide what the figure should show before choosing a chart type.
2. Choose the right chart type for that question (see the table above).
3. Label everything. A title that states the message, axis labels with units, a legend and a data source.
4. Be honest with scales. Start bar charts at zero, and say clearly when an axis does not.
5. Remove clutter. Delete anything that does not help the reader: heavy gridlines, 3D effects, background images, redundant labels.
6. Use color with purpose. Color should encode data or draw attention, not decorate. Use colorblind safe palettes such as viridis or ColorBrewer schemes.
7. Prefer small multiples to crowded charts. Several simple panels beat one chart with ten lines.
8. Never put two unrelated scales on one axis. Temperature and rainfall belong in two panels, not on one shared y-axis.
9. Make it accessible. Add alt text, keep good contrast, and do not rely on color alone.

Ethics

---

Visualizations influence decisions about funding, policy and people’s lives.

• Do not cherry pick time windows, places or subsets to support a conclusion.
• State your data source and its limitations.
• Protect privacy. Maps of individual addresses or small areas can identify people.
• Be careful with sensitive topics such as health, crime or immigration. How you classify and color data can stigmatize places and the people who live there.

Practice

---

Work through the notebooks below in Google Colab. Click File > Save a copy in Drive first so you can save your changes.

Challenge

Challenge 1: Exploratory or explanatory?

For each figure, decide whether it is exploratory or explanatory:

1. A quick histogram you make to check whether income values look reasonable.
2. A map of county vaccination rates in a public health department press release.
3. A scatterplot matrix of ten survey variables you look at before choosing a model.

Show me the solution

1. Exploratory: it is for you, to check the data.
2. Explanatory: it carries one message to a general audience, so it needs a clear title, legend, source and careful color choices.
3. Exploratory: it helps you decide what to analyze next.

Challenge

Challenge 2: Spot the problems

A colleague shows you a 3D pie chart with 12 slices in rainbow colors, no title and no labels, showing the share of survey respondents in each of 12 age groups. List at least four problems and suggest a better chart.

Show me the solution

• 3D distorts slice sizes: front slices look bigger than back slices.
• Twelve slices are too many to compare by angle.
• Rainbow colors have no order, even though age groups do, and some colors are hard to tell apart for people with color vision deficiency.
• No title, labels or percentages, so the reader cannot tell what is shown.

A better choice is a bar chart (or histogram) of the 12 age groups in order, in a single color, with a title, labeled axes and the survey source.

Notebooks

---

Level	Notebook	What you will do
Beginner	1a. Making a Good Chart	Build line and bar charts, then fix common problems: missing labels, truncated axes, wrong chart type, too many colors
Intermediate	1b. Choosing the Right Chart	Distributions, scatter plots, small multiples, correlation heatmaps and interactive charts
Advanced	1c. From Charts to Maps	Choropleth maps, normalization, classification methods, colorblind safe palettes and interactive maps

The notebooks use datasets built into Plotly, so they need no download or account. Notebook 1c also offers an optional section for learners with a free Census API key.

Key Points

• Data visualization turns numbers into positions, lengths, areas and colors that people can compare quickly.
• Exploratory figures help you understand your data; explanatory figures communicate one message to others.
• Correct data can still mislead through truncated axes, the wrong chart type or poor color choices.
• Choose the chart type from the question, label everything, and keep scales honest.
• Python and QGIS are free tools that cover charts, maps and cartographic layouts.

Content from Cartography Checklists

---

Last updated on 2026-10-17 | Edit this page

Overview

Questions

• What should I decide before I start making a map?
• How do audience, message and medium shape a map’s design?
• How can a map mislead, even when the data are correct?
• When do I need more data, or a better understanding of my data, before mapping?

Objectives

• Identify the audience, purpose and medium of a map before designing it
• Choose the data attributes that support a single clear message
• Recognize design choices that confuse or mislead readers
• Decide whether additional data or research is needed
• Use a checklist to review a map before sharing it

Why plan before you map?

---

Maps are persuasive. People tend to trust them, and a map can be read in seconds. That makes a well designed map a powerful way to share research, and a poorly designed one a powerful way to mislead.

Most map problems are decided before any software is opened: who the map is for, what it should say, and which data can support that message. This episode is a set of questions to ask, in order, every time you make a map. The next episode covers the design tools (color, classification, projections) you use to answer them.

Good cartography rests on a few core design principles:

• Legibility: everything the reader needs can be read easily.
• Visual contrast: important features stand out from less important ones.
• Figure and ground: the data (figure) separates clearly from the background (ground), such as a basemap.
• Hierarchy: the most important element is noticed first, the least important last.
• Balance: the elements are arranged without clutter or large empty areas.

---

1. Know your audience

---

The audience shapes every other decision.

Ask yourself:

• Are they researchers, policymakers, community members or students?
• How familiar are they with maps and with your topic?
• How much detail can they use?

Audience	Design implications
General public	Few classes, plain language labels, a title that states the conclusion, minimal jargon
Policymakers	One clear message, the places they are responsible for clearly marked, sources stated
Researchers	More detail, precise class breaks, projection and methods stated, uncertainty shown

Callout

Key idea

A map of the same data for scientists and for the public should not look the same.

---

2. Define your message

---

Every map should answer one clear question. If you cannot state the message in one sentence, the reader will not be able to find it either.

Ask yourself:

• What is the single most important takeaway?
• Am I showing a pattern, a comparison, or a change over time?

Avoid:

• Showing many variables at once on one map
• Leaving the reader to work out the point alone

A useful habit is to write the message as the map’s title. The map below states its conclusion in the title, shows only one distinction (8% or more versus below 8%), and greys out everything else so the highlighted counties stand out.

A map with one clear message: counties with unemployment of 8% or more in 2016

---

3. Choose the right data attributes

---

Not all data belongs on your map. Select the variables that support your message and that make sense spatially.

Ask yourself:

• Which variable answers my question?
• Do I need a supporting variable, such as population, to interpret it?
• What type of geometry do I have: points, lines or polygons?
• Is my variable a count (number of people) or a rate (percent, per 1,000, per square mile)? Counts on area maps usually need to be converted to rates; see the next episode.

Matching data to visual variables:

Data	Visual variable	Example
Ordered quantities (low to high)	Lightness of one color	Unemployment rate
Magnitudes at locations	Symbol size	City population
Categories with no order	Different hues or shapes	Land use type

Challenge

Challenge 1: Which variable?

You have county data on average temperature, total precipitation and elevation. Your goal is a map of drought risk. Which variable would you prioritize, and what else might you need?

Show me the solution

Precipitation is the most direct indicator, because drought is a shortage of water. Even better is precipitation compared with normal (for example, percent of the long-term average), since 20 inches a year is normal in some places and a severe drought in others. Temperature is a useful supporting variable because heat increases evaporation. Elevation alone says little about drought.

---

4. Consider how readers will perceive the map

---

Maps are not neutral. Design choices change what readers believe.

Ask yourself:

• Do my colors carry meanings I do not intend (for example, red for “danger”)?
• Does my classification make some places look worse or better than they are?
• Can the main pattern be seen at a glance?
• Can someone with color vision deficiency read it? (See the Accessibility episode.)

Darker or more saturated colors are read as “more”. Keep that order consistent with the data: dark for high values, light for low values.

The map below uses the same unemployment data as the map in section 2, and is very hard to read:

A deliberately poor map of the same unemployment data

What goes wrong here:

• A rainbow palette has no natural order, so readers cannot tell which colors mean high or low.
• Twelve classes are more than anyone can match to a legend, and there is no legend anyway.
• The title (“map1”) says nothing; the axis labels are leftover software defaults.
• Heavy black borders and a bright background compete with the data (poor figure and ground).
• Equal interval classes put almost every county in the first few classes, so most of the map is one dark blue.

Best practice: make the data the figure and everything else the ground. Use a light, muted basemap or none at all, thin light boundaries, and a sequential palette with five to seven classes.

---

5. Choose the right medium

---

Where the map will be seen affects its design.

Medium	Design implications
Web map	Interactive, zoomable; simplify the basemap; add halos to labels so they read over any background
Print	Static, high resolution; check how it looks in black and white; print a test copy
Slides	Simple and bold; large text; one message per slide; viewed from a distance
Social media	Small screens; very few classes; title and source inside the image

Ask yourself: Will readers zoom in? Will it be printed in greyscale? How large will it appear?

---

6. Will your map inform decisions?

---

Some maps are exploratory; others guide real actions such as where to open a clinic or which neighborhoods receive funding.

Decision making maps should:

• Be as accurate as the data allow
• Show uncertainty where possible (for example, margins of error for American Community Survey estimates, or hatching for unreliable values)
• Avoid simplifications that change the conclusion

Examples include flood risk maps used by planners and disease maps used during outbreaks.

Callout

Important

If your map influences decisions, accuracy and clarity matter more than visual appeal.

---

7. Understand what your audience already knows

---

Ask yourself:

• Do readers understand the variable (for example, what “median household income” means)?
• Do units, time periods or scales need explaining?
• Would an annotation help (“highest rate in the state”)?

Tips:

• Include a legend with units and a descriptive title.
• Use plain language where possible.
• Give context: time period, geography and data source.
• Include the essential map elements: a scale bar when distance matters, a north arrow when orientation is not obvious, and a source line.

Aim for maximum information with minimum effort: the reader should get the main idea quickly, without struggling.

---

8. Do you need more data?

---

Incomplete data can lead to misleading maps.

Ask yourself:

• Are there variables that change how this one should be read?
• Is the data recent enough?
• Is the spatial resolution right for the question (state, county, tract)?

For example, a map of unemployment rates can look dramatic in sparsely populated counties where a change of a few hundred jobs moves the rate a lot. Showing population next to it puts the pattern in context:

Unemployment rate and population density side by side

The darkest fifth of counties on the left (unemployment of 6.5% or more) are, on average, less densely populated than other counties and together are home to only about 1 in 10 people in the contiguous United States. Meanwhile, the 41 counties with more than a million residents had a median rate of 4.5%, slightly below the median for all counties (5.0%). Neither map alone tells the whole story. Rates and counts answer different questions: “where is unemployment common?” versus “where do most unemployed people live?”

---

9. Do you understand your data?

---

Before mapping, make sure you understand your dataset.

Ask yourself:

• What exactly does each variable measure, and how was it collected?
• Are there known biases, gaps or margins of error?
• Have you explored the data with summary statistics and quick plots?

If not:

• Do some exploratory analysis (histograms, maps of each variable)
• Read the metadata and documentation
• Ask someone who knows the data
• For advanced work, run a sensitivity check: does the pattern change if you change the classification, the year or the geographic unit?

---

Cartography checklist

---

Before sharing a map, check that:

• I know my target audience
• The map has one clear message, stated in the title
• I chose the most relevant variable, and normalized counts where needed
• The design does not invite misleading interpretations
• The design fits the medium (web, print, slides)
• Accuracy and uncertainty are appropriate for any decisions it will inform
• Units, time period and source are stated
• I have the supporting data needed to interpret the pattern
• I understand how the data were collected and their limitations
• Colors are readable for people with color vision deficiency

Challenge

Challenge 2: Review a map

Apply the checklist to the poorly designed map in section 4. Which items fail?

Show me the solution

• Clear message: fails. The title is “map1” and nothing says what is shown.
• Misleading interpretations: fails. The rainbow palette and equal interval classes hide the pattern and make a few counties look extreme.
• Units, time period, source: fail. There is no legend, year or source.
• Color vision deficiency: fails. Rainbow palettes contain red and green pairs that many readers cannot distinguish.
• Medium: questionable. The bright background would print poorly and distract on screen.

Final thought

---

A good map is not only attractive. It is honest, clear and purposeful: it respects the data, serves its audience, and communicates without distortion.

Discussion

• Think of a map you saw recently. What did it do well? What could be improved?
• How would you present the same data differently for a different audience?
• Share an example of a map that misled you, or one that worked brilliantly, and explain why.

Key Points

• Decide the audience, message and medium before designing a map.
• State the message in the title and show only the data that supports it.
• Rates and counts answer different questions; choose deliberately.
• Design choices such as palette and classification change what readers believe.
• Review every map against a checklist before sharing it.

Content from Fundamentals of Map Design

---

Last updated on 2026-10-17 | Edit this page

Overview

Questions

• What makes a map effective?
• How do visual hierarchy and visual variables guide a reader?
• How should colors be chosen for different kinds of data?
• What are map scale and projections, and why do they matter?
• Which thematic map type suits which data?
• Why should counts usually be converted to rates before mapping?
• How does the classification method change what a map shows?

Objectives

• Identify the core elements of a map layout
• Apply visual hierarchy and visual variables to improve clarity
• Choose sequential, diverging or categorical color schemes appropriately
• Explain map scale and the trade-offs between projections
• Select a thematic map type that fits the data
• Normalize counts into rates where appropriate
• Compare six classification methods and choose one deliberately
• Find the equivalent tools in Python and QGIS

What is a map?

---

A map is a visual representation of spatial data designed to communicate information about locations, patterns and relationships. Like any figure, it is a communication tool, not just a picture.

A good map:

• Has a clear purpose
• Represents the data accurately
• Is easy to read and interpret
• Avoids misleading elements
• Includes the map elements the reader needs

Map elements

The elements of a map layout, labeled on a map of county unemployment

Element	Purpose	Always needed?
Title	States what the map shows (ideally its message)	Yes
Mapped area	The data itself	Yes
Legend	Explains colors, symbols and units	Yes, unless there is a single obvious symbol
Data source	Lets readers check and trust the data	Yes
Scale bar	Shows distance	When distance matters (local and regional maps)
North arrow	Shows orientation	When north is not up or orientation is unclear
Projection note	States the coordinate system	For technical audiences
Neat line	A frame that groups the elements	Optional
Inset map	Shows location in a wider area, or distant parts such as Alaska and Hawaii	When needed

Callout

Key idea

Every element should earn its place. A north arrow on a world map in a familiar projection adds nothing; a missing legend makes a map unreadable.

---

Visual hierarchy

---

Visual hierarchy controls what the reader sees first, second and last.

Tools for creating hierarchy:

• Size: larger elements draw attention first
• Lightness and contrast: strong contrast with the background stands out
• Color saturation: vivid colors stand out against muted ones
• Position: elements near the center or top are noticed early

A typical hierarchy:

1. The data layer: strongest colors and contrast
2. Title and legend: clear but not overpowering
3. Labels: readable, but secondary to the data
4. Basemap, boundaries and other context: muted greys and thin lines

Challenge

Challenge 1: What do you see first?

Look at the map elements figure above. What do you notice first, second and third? Is that the order the mapmaker would want?

Show me the solution

Most readers notice the dark blue counties first (the data), then the large title, then the legend. That is the intended order: the data and its message come first, and the legend explains them. The red annotation labels are added for this lesson and would not appear on a real map.

---

Visual variables

---

Visual variables are the properties of a symbol that can change to represent data. The idea comes from the cartographer Jacques Bertin.

Visual variable	Good for	Example
Position	Location (on maps, this is fixed by geography)	Where a city is
Size	Quantities	City population as circle size
Lightness (value)	Ordered data, low to high	Unemployment rate as light to dark
Hue	Categories	Land use type
Shape	Categories (points)	Airport type
Orientation and texture	Categories, secondary encoding	Hatching for uncertain values

The key rule: use ordered variables (size, lightness) for ordered data, and unordered variables (hue, shape) for categories. Using different hues for low to high values forces the reader to keep looking at the legend.

---

Colors on maps

---

There are three kinds of color scheme, one for each kind of data.

Sequential, diverging and categorical color schemes on three U.S. maps

Scheme	Use for	How it looks	Example palettes
Sequential	Values from low to high	One hue, light to dark	Blues, Greens, viridis, YlGnBu
Diverging	Values above and below a meaningful midpoint (zero, an average, 50%)	Two hues meeting at a neutral light middle	RdBu, BrBG, PuOr
Categorical	Groups with no order	Distinct hues of similar lightness	Set2, Dark2, Tableau 10

Best practices:

• Light colors for low values and dark colors for high values (most readers assume dark means more).
• Use a diverging scheme only when the midpoint means something. The middle map above uses the national median, so it answers “is this county above or below typical?”
• Avoid rainbow palettes for ordered data; they have no natural order and create false boundaries.
• Avoid red and green together; see the Accessibility episode.
• Five to seven classes is usually the most a reader can tell apart.

ColorBrewer is a free tool for choosing map palettes, including colorblind safe ones. The same palettes are built into QGIS and Python.

---

Scale

---

Map scale is the relationship between a distance on the map and the same distance on the ground, for example 1:24,000 (one inch on the map is 24,000 inches, or 2,000 feet, on the ground).

• Large-scale maps show a small area in high detail, such as a city map (1:10,000).
• Small-scale maps show a large area with less detail, such as a world map (1:50,000,000).

The terms feel backwards at first: 1/10,000 is a larger fraction than 1/50,000,000, so the city map has the larger scale.

Scale matters because it determines how much detail can be shown and which patterns are visible. A pattern that is clear across counties may disappear, or reverse, at the tract level.

---

Projections

---

The Earth is round; maps are flat. A projection is a mathematical method for flattening the Earth, and every projection distorts at least one of area, shape, distance or direction.

The world in four projections, with Greenland and the Democratic Republic of the Congo highlighted

Projection family	Preserves	Typical use
Equal area (Albers, Equal Earth, Mollweide)	Relative area	Thematic maps of rates or densities; use these for choropleth maps
Conformal (Mercator, Lambert Conformal Conic)	Local shapes and angles	Navigation, web map tiles, large-scale local maps
Equidistant (Azimuthal Equidistant)	Distance from one point	Distance or flight range maps
Compromise (Robinson, Winkel Tripel)	Nothing exactly, everything approximately	General reference world maps

Web maps (Google Maps, OpenStreetMap) use Web Mercator, which is why Greenland and Russia look enormous online. For thematic maps of the United States, the USA Contiguous Albers Equal Area projection (EPSG:5070) is a common choice; it is used for the U.S. maps in this lesson.

Try The True Size Of to drag countries north and south on a Mercator map and see how much their apparent size changes. Drag Russia down to the equator and compare it with Africa.

Callout

Important

There is no perfect projection, only projections suited to particular purposes. When a map compares areas or densities, use an equal area projection.

This section is about choosing a projection for a map. For coordinate reference systems as part of your data (datums, EPSG codes, and the difference between defining and reprojecting a layer), see the Data Management module’s Coordinate Reference Systems episode.

---

Labels and legends

---

Labels:

• Readable size and font; avoid overlap
• Use hierarchy: larger text for more important places
• Add a light halo or outline when labels sit on top of busy backgrounds

Legends:

• Explain every color and symbol
• Include units (“percent”, “people per square mile”)
• Use readable, rounded class breaks where possible
• Order classes from low to high, matching the color order

---

Thematic map types

---

Choropleth maps

A choropleth map shades areas (states, counties, tracts) by a data value.

Choropleth map of population density by state

Best for: rates, ratios and densities (percent, per capita, per square mile). Avoid: raw counts, because large areas and large populations dominate the map (see Normalization below).

Proportional symbol maps

Proportional symbol maps place a symbol, usually a circle, at each location and size it by a value.

Proportional symbol map of the 60 largest U.S. cities

Best for: counts and magnitudes at points, or counts for areas when you want to avoid the choropleth problem with raw counts. Scale the circle’s area, not its radius, to the value.

Dot density maps

Dot density maps place one dot for a fixed number of things (people, households, cases) inside each area.

Dot density map of Indiana population

Best for: showing how a count is distributed. Remember that dots are placed at random inside each area; they do not show exact locations.

Cartograms

A cartogram resizes areas according to a data value. In a non-contiguous cartogram, each area keeps its shape but is shrunk (or grown) in place, so gaps open between neighbors.

Non-contiguous cartogram of U.S. states scaled by population density

Best for: emphasizing magnitude over geographic area, for example showing that large but sparsely populated states hold few people. Other types, such as contiguous and Dorling cartograms, distort shapes or replace them with circles.

Multivariate maps

A multivariate map shows two or more variables at once, usually with different visual variables.

Multivariate map of Indiana unemployment and population

Best for: exploring relationships between variables. Combine different visual variables (fill lightness for one, symbol size for another) rather than two sets of colors. Keep it to two or three variables, or the map becomes hard to read.

---

Normalization: counts versus rates

---

Normalizing means dividing a count by something that accounts for size: population, households, land area or the total of a group. It turns “how many?” into “how common?”.

Choropleth maps of raw counts mostly show where many people live. The two maps below show the same data:

Hispanic population as a count and as a percent of county population

The count map (left) highlights a few very populous counties. The percent map (right) shows where Hispanic residents make up a large share of the community, a different and usually more useful question.

Common normalizations:

Count	Divide by	Result
People	Land area	Population density (people per sq mile)
People in a group	Total population	Percent of population
Cases, crimes, events	Population	Rate per 1,000 or 100,000
Households without a car	All households	Percent of households

Callout

Rule of thumb

Use rates on choropleth maps. If you need to show counts, use proportional symbols or dot density instead.

---

Static and interactive maps

---

Static maps are fixed images. They suit print and reports, and give you full control of the design. All of the maps in this episode are static.

Interactive (web) maps let readers zoom, pan, toggle layers and click or hover for details. They suit exploration and large datasets, but take more effort to build and to design well.

For an example, open the Workshop 2026 page on this site and scroll down to the interactive map of the workshop location. In Python, a single line, gdf.explore(), turns a GeoDataFrame into an interactive web map (see notebook 1c).

Callout

Guideline

Use interactive maps when readers need to explore. Use static maps when you want to communicate one message clearly.

---

Data classification

---

Most choropleth maps group values into a few classes, each with one color. How you choose the class breaks can change the story the map tells. The six maps below show exactly the same data, Indiana population density by county, classified six ways.

The same Indiana population density data classified six ways

The six methods

Equal interval divides the range of values into classes of equal width (for example 0 to 500, 500 to 1,000). It is easy to read and good for evenly spread data, but with skewed data, as here, almost every area lands in the first class.

Quantile puts the same number of areas in each class (here, about 18 counties per class). Every color is used equally, which is good for ranking (“top fifth of counties”), but very different values can share a class, and similar values can be split apart.

Natural breaks (Jenks) finds breaks that minimize differences within classes and maximize differences between them. It follows the natural clusters in the data and often gives a faithful picture of one map. Because the breaks depend on the data, the same method gives different breaks for different years or places, so maps are hard to compare.

Standard deviation shows how far each value is from the mean (for example, “more than 1 standard deviation above”). It is useful when the question is “how unusual is this place?” and best paired with a diverging palette. With skewed data like this, most areas fall in the middle class.

Pretty breaks uses round, easy to read numbers (0, 1,000, 2,000). Readers like round numbers, but the breaks may not match the shape of the data.

Geometric interval makes each class a constant multiple wider than the last. It suits strongly skewed data, such as population density or income, where most values are small and a few are very large.

You can also set manual breaks at meaningful values, such as the poverty line or a policy threshold, which is often the best choice for maps that inform decisions.

Where the breaks fall

A histogram with the class breaks drawn on it shows why the maps look so different:

Histogram of Indiana county population density with class breaks for three methods

Choosing a method

Method	Best use	Watch out for
Equal interval	Evenly spread data; familiar ranges such as percentages	Skewed data puts nearly everything in one class
Quantile	Rankings and comparisons (“top 20%”)	Similar values split; different values grouped
Natural breaks	Showing clusters in one dataset	Breaks are hard to compare across maps
Standard deviation	How unusual each area is	Readers must understand standard deviation
Pretty breaks	General audiences	Breaks may not suit the data
Geometric interval	Strongly skewed data	Less intuitive class ranges
Manual breaks	Meaningful thresholds	Needs a good reason for each break

Always look at a histogram before choosing. If you will compare several maps (different years or places), use the same breaks on all of them.

Challenge

Challenge 2: Choose a classification

You are mapping U.S. county median household income, which ranges from about $25,000 to $150,000 with most counties between $45,000 and $70,000 and a long tail of wealthy suburban counties.

1. Which method would you choose for a single map in a report, and why?
2. Which would you choose if you were comparing maps from 2010 and 2020?

Show me the solution

1. Natural breaks or quantile both work for a single map. Natural breaks shows the main cluster and the wealthy tail as separate classes. Quantile is good if the message is about rankings (“the top fifth of counties”). Equal interval would put most counties in one or two classes because of the long tail.
2. Use the same manual breaks (for example $40k, $55k, $70k, $90k) on both maps, so a color means the same income in both years. Remember to adjust for inflation if you compare dollar values over time.

Challenge

Challenge 3: Count or rate?

Which of these should be normalized before making a choropleth map, and what would you divide by?

1. Number of COVID-19 cases by county
2. Median age by county
3. Number of grocery stores by census tract

Show me the solution

1. Normalize: divide by population to get cases per 100,000 residents.
2. Already a rate-like value: a median is a summary of each county, so it can be mapped directly.
3. Normalize: divide by population (stores per 10,000 people) or by area (stores per square mile), depending on the question. Or show the stores as points instead.

---

Python and QGIS equivalents

---

Each concept in this episode has a direct tool in both platforms used in this curriculum.

Concept	Python (GeoPandas, Matplotlib)	QGIS
Choropleth	gdf.plot(column="rate", legend=True)	Layer Properties > Symbology > Graduated
Classification method	gdf.plot(column="rate", scheme="NaturalBreaks", k=5) (uses mapclassify)	Graduated > Mode (Equal Count (Quantile), Equal Interval, Natural Breaks (Jenks), Standard Deviation, Pretty Breaks, Logarithmic)
Color scheme	cmap="Blues" or cmap="viridis"	Graduated > Color ramp
Categories	gdf.plot(column="type", categorical=True)	Symbology > Categorized
Proportional symbols	gdf.plot(markersize=gdf["pop"] / 1000)	Graduated > Method: Size
Dot density	Generate random points per polygon	Fill symbol > Random Marker Fill with a data-defined count
Normalization	gdf["rate"] = gdf["count"] / gdf["pop"]	Open the attribute table > Field Calculator > new field "count" / "pop"
Projection	gdf.to_crs(epsg=5070)	Project > Properties > CRS
Interactive map	gdf.explore()	The qgis2web plugin
Layout with map elements	Matplotlib annotations	Print Layout (see Getting Started with QGIS in the Foundations module)

---

Final takeaways

---

• Maps are communication tools: design them deliberately for an audience and a message.
• Use ordered visual variables for ordered data, and match the color scheme to the data.
• Use an equal area projection for thematic maps of rates and densities.
• Map rates, not counts, on choropleth maps.
• Classification changes the story; check a histogram and choose on purpose.

Discussion

• When might an interactive map be worse than a static one?
• Find a choropleth map in the news. Does it map a count or a rate? Which classification does it seem to use?

Key Points

• A map layout needs a title, legend and source, plus a scale bar and north arrow when they help.
• Sequential, diverging and categorical color schemes match ordered, centered and unordered data.
• Every projection distorts something; use equal area projections for thematic maps.
• Choropleth maps need rates; show counts with proportional symbols or dots.
• The classification method changes the pattern a map shows; use the same breaks to compare maps.

Content from Cartography in Python

---

Last updated on 2026-10-20 | Edit this page

Overview

Questions

• How do I turn a correct map into a finished one that others can read and print?
• How should layers, labels and a basemap be ordered and styled?
• When do small multiples need a shared scale?

Objectives

• Style and order layers by their role: subject, supporting features, context and reference
• Label features legibly and by rule, with halos
• Add a basemap responsibly, a locator inset, a true scale bar and a north arrow
• Build small multiples with separate or shared classifications
• Export maps for print and screen

From a correct map to a finished map

---

The Fundamentals of Map Design episode covers what a thematic map shows: map elements, normalization, classification and color. This episode is about finishing: turning a correct choropleth into a map that a reader outside your project understands at a glance and that prints cleanly. The running example is the share of households without a car in Lafayette and West Lafayette, Indiana, with parks and city limits on top.

Visual hierarchy, layer by layer

---

Every layer has a role, and its styling should match it:

Role	Example	Styling
Subject	The choropleth of car-free households	Strong color, full opacity
Supporting features	Parks, city limits	Distinct but secondary: one hue, thin or dashed lines, some transparency
Context	Basemap, neighboring areas	Quiet: gray, low contrast
Reference	Labels, scale bar, north arrow, source	Small, legible, never decorative

Order follows from the roles: polygons of the subject at the bottom, supporting outlines and points above them, labels on top, and the basemap underneath everything. A layer on top of a choropleth needs either no fill, transparency, or a strong outline so it does not hide what is underneath.

Labels

---

Labels are reference information. They should be findable, legible and not in the way.

• Label sparingly. Label what the reader needs to orient themselves or to understand the story: the largest parks, the two city names. Labeling all 68 parks makes none of them readable.
• Use a halo (a thin outline in the background color, called a buffer in QGIS). It keeps text readable over any color and is the single most effective labeling technique.
• Use type to signal kind. Common conventions: italics for water and areas (rivers, parks, neighborhoods), upright for places and points, larger or capitalized type for more important names.
• Place labels inside or next to their feature, never across a boundary. For polygons, a point guaranteed to be inside (a “representative point” or “point on surface”) is safer than the centroid, which can fall outside an L-shaped polygon.
• Choose rules, not exceptions. “Label parks over 30 hectares” is reproducible; hand-picking labels is not.

Basemaps

---

A tiled basemap (OpenStreetMap, Esri, CARTO, Stamen) adds streets and place names for free, at a cost: visual noise, a dependence on an online service, and license conditions. Choose a quiet, gray basemap for thematic maps, keep its attribution text, and do not rely on it for print if you need full control or very high resolution. Basemap tiles are in Web Mercator (EPSG:3857); software reprojects them to your map’s CRS, which can blur their labels slightly.

Locator insets, scale bars and north arrows

---

• A locator inset (a small map of the state or country with the study area highlighted) answers “where is this?” for readers who do not know the area. It does not need a scale bar or legend of its own.
• A scale bar is only true on a map in a projected CRS with little distortion over the map (a UTM zone or a state plane for a county). In degrees or Web Mercator at large extents the scale changes across the map, and a single bar misleads.
• A north arrow is only essential when north is not up, but it is cheap and many readers expect one.

Small multiples and shared scales

---

To compare several variables or years, repeat the same map in a grid of small multiples. The eye compares patterns across panels far better than it decodes several variables in one crowded map.

Decide deliberately whether panels share a classification:

• Same variable or same units (one rate in different years, two rates you want to compare): use the same class breaks and colors in every panel, or the same dark color will mean different values.
• Different variables (income in dollars, children in percent): each panel needs its own classes and legend, and the reader compares patterns, not colors.

Export

---

Destination	Format	Notes
Print, a paper, a poster	PDF (vector)	Sharp at any size; set the page or figure size to the printed size
Slides, a website	PNG at 200 to 300 dpi	PNG keeps sharp edges; JPEG blurs lines and text
Further editing (Illustrator, Inkscape)	SVG	Every element stays editable

Check the exported file, not just the screen: font sizes, line widths and legend spacing often look different at print size.

Callout

The same map in QGIS

The Cartography in QGIS episode makes the same map with rule-based styling and labeling and a print layout with a locator inset. Notebook 01c of this module covers the basics (choropleths, classification, palettes) used here.

The toolkit in code

---

Task	Python
Layer order	zorder= on each plot() call; alpha= for transparency
Zoom	ax.set_xlim(), ax.set_ylim() from a layer’s total_bounds
Labels with halos	ax.text(..., path_effects=[patheffects.withStroke(linewidth=3, foreground="white")]) at representative_point()
Basemap	contextily.add_basemap(ax, crs=layer.crs, source=contextily.providers.Esri.WorldGrayCanvas)
Legend for several layers	Build Patch and Line2D handles and pass them to ax.legend(handles=...)
Locator inset	inset = ax.inset_axes([left, bottom, width, height]), then plot the state and the highlighted county
Scale bar	ax.add_artist(ScaleBar(dx=1, units="m")) from matplotlib-scalebar, for maps in meters
Shared classes	scheme="UserDefined", classification_kwds={"bins": [...]} in every panel
Export	fig.savefig("map.pdf"), fig.savefig("map.png", dpi=300, bbox_inches="tight")

Challenge

Challenge 1: What is wrong with this map?

A map of poverty rates by tract uses a satellite basemap at full opacity, red and green classes, labels for every street, a scale bar on a Web Mercator map of the whole United States, and no source line. List the problems and a fix for each.

Show me the solution

• Satellite basemap: too busy for a thematic map; use a quiet gray basemap or none, and keep the attribution.
• Red and green classes: unreadable for many readers with color vision deficiency, and a diverging scheme for a one-directional rate; use a sequential palette (see the Accessibility episode).
• Labels for every street: clutter; label only what orients the reader.
• Scale bar on a national Web Mercator map: the scale varies greatly with latitude; drop it or use an equal-area projection and a note.
• No source line: add the dataset and year.

Challenge

Challenge 2: Shared scale or not?

You make four small multiples. Which need a shared classification? (a) Households without a car in 2015, 2018, 2021 and 2024. (b) Poverty rate, median income, share of children and population density in 2024.

Show me the solution

1. Shared: the same variable over time, so the same color must mean the same value in every year. Compute breaks from all four years together. (b) Separate: four variables in different units; give each panel its own classes and legend, and let readers compare patterns.

Key Points

• Style each layer by its role; the subject is strong, context is quiet
• Label sparingly, by rule, with halos; italics for areas and water
• Use quiet basemaps and keep their attribution; scale bars need a projected CRS
• A locator inset shows where the map is; small multiples compare, with shared breaks only for comparable values
• Export PDF for print and PNG at 200 to 300 dpi for screens, and check the exported file

Notebooks

---

Notebook	What you will do
2a. Layers, Labels and Basemaps	Stack tracts, parks and city limits, zoom to the cities, label the largest parks with halos, add a basemap and a multi-layer legend
2b. Insets, Small Multiples and Export	Add a locator inset, a scale bar and a north arrow, compare four variables in small multiples, share a scale, and export PDF and PNG

Content from Cartography in QGIS

---

Last updated on 2026-10-25 | Edit this page

Overview

Questions

• How do I turn a correct map into a finished one that others can read and print?
• How should layers, labels and a basemap be ordered and styled?
• When do small multiples need a shared scale?

Objectives

• Style and order layers by their role: subject, supporting features, context and reference
• Label features legibly and by rule, with halos
• Add a basemap responsibly, a locator inset, a true scale bar and a north arrow
• Use rule-based styling and labeling, and build a print layout with a locator inset
• Export maps for print and screen

From a correct map to a finished map

---

The Fundamentals of Map Design episode covers what a thematic map shows: map elements, normalization, classification and color. This episode is about finishing: turning a correct choropleth into a map that a reader outside your project understands at a glance and that prints cleanly. The running example is the share of households without a car in Lafayette and West Lafayette, Indiana, with parks and city limits on top.

Visual hierarchy, layer by layer

---

Every layer has a role, and its styling should match it:

Role	Example	Styling
Subject	The choropleth of car-free households	Strong color, full opacity
Supporting features	Parks, city limits	Distinct but secondary: one hue, thin or dashed lines, some transparency
Context	Basemap, neighboring areas	Quiet: gray, low contrast
Reference	Labels, scale bar, north arrow, source	Small, legible, never decorative

Order follows from the roles: polygons of the subject at the bottom, supporting outlines and points above them, labels on top, and the basemap underneath everything. A layer on top of a choropleth needs either no fill, transparency, or a strong outline so it does not hide what is underneath.

Labels

---

Labels are reference information. They should be findable, legible and not in the way.

• Label sparingly. Label what the reader needs to orient themselves or to understand the story: the largest parks, the two city names. Labeling all 68 parks makes none of them readable.
• Use a halo (a thin outline in the background color, called a buffer in QGIS). It keeps text readable over any color and is the single most effective labeling technique.
• Use type to signal kind. Common conventions: italics for water and areas (rivers, parks, neighborhoods), upright for places and points, larger or capitalized type for more important names.
• Place labels inside or next to their feature, never across a boundary. For polygons, a point guaranteed to be inside (a “representative point” or “point on surface”) is safer than the centroid, which can fall outside an L-shaped polygon.
• Choose rules, not exceptions. “Label parks over 30 hectares” is reproducible; hand-picking labels is not.

Basemaps

---

A tiled basemap (OpenStreetMap, Esri, CARTO, Stamen) adds streets and place names for free, at a cost: visual noise, a dependence on an online service, and license conditions. Choose a quiet, gray basemap for thematic maps, keep its attribution text, and do not rely on it for print if you need full control or very high resolution. Basemap tiles are in Web Mercator (EPSG:3857); software reprojects them to your map’s CRS, which can blur their labels slightly.

Locator insets, scale bars and north arrows

---

• A locator inset (a small map of the state or country with the study area highlighted) answers “where is this?” for readers who do not know the area. It does not need a scale bar or legend of its own.
• A scale bar is only true on a map in a projected CRS with little distortion over the map (a UTM zone or a state plane for a county). In degrees or Web Mercator at large extents the scale changes across the map, and a single bar misleads.
• A north arrow is only essential when north is not up, but it is cheap and many readers expect one.

Small multiples and shared scales

---

To compare several variables or years, repeat the same map in a grid of small multiples. The eye compares patterns across panels far better than it decodes several variables in one crowded map.

Decide deliberately whether panels share a classification:

• Same variable or same units (one rate in different years, two rates you want to compare): use the same class breaks and colors in every panel, or the same dark color will mean different values.
• Different variables (income in dollars, children in percent): each panel needs its own classes and legend, and the reader compares patterns, not colors.

Export

---

Destination	Format	Notes
Print, a paper, a poster	PDF (vector)	Sharp at any size; set the page or figure size to the printed size
Slides, a website	PNG at 200 to 300 dpi	PNG keeps sharp edges; JPEG blurs lines and text
Further editing (Illustrator, Inkscape)	SVG	Every element stays editable

Check the exported file, not just the screen: font sizes, line widths and legend spacing often look different at print size.

---

Hands-on: a finished map of car-free households

---

These steps were checked in QGIS 4.0. They build on the Foundations episodes Getting Started with QGIS (layers, symbology, print layouts) and Tables, Joins and Selections (virtual fields).

Callout

If QGIS asks you to choose a transformation

When layers in latitude and longitude (EPSG:4326) are drawn in a project in UTM (EPSG:26916), QGIS may open a window titled with “Multiple operations are possible for converting coordinates”. For this county any of the options is accurate to a few meters: keep the first (highlighted) one and click OK. QGIS remembers the choice for the project.

Part 1: Load and order the layers

1. Start a new project and set its CRS to EPSG:26916 (bottom-right corner of the window).

2. Load these with Add Vector Layer, Protocol: HTTP(S), Type: HTTP/HTTPS/FTP:

   Rename to	File
   Tracts	tippecanoe_tracts_acs.geojson
   Parks	tippecanoe_parks.geojson
   City limits	study_area.geojson
   Indiana counties	indiana_counties.geojson
   Tippecanoe County	tippecanoe_county.geojson

   Web addresses to copy:

   • Tracts: https://raw.githubusercontent.com/SpatialTurn/Foundations/main/data/tippecanoe_tracts_acs.geojson
   • Parks: https://raw.githubusercontent.com/SpatialTurn/Foundations/main/data/tippecanoe_parks.geojson
   • City limits: https://raw.githubusercontent.com/SpatialTurn/CaseStudies/main/data/study_area.geojson
   • Indiana counties: https://raw.githubusercontent.com/SpatialTurn/Foundations/main/data/indiana_counties.geojson
   • Tippecanoe County: https://raw.githubusercontent.com/SpatialTurn/Foundations/main/data/tippecanoe_county.geojson

3. Drag them into this order, top to bottom: City limits, Parks, Tracts, Tippecanoe County, Indiana counties. Untick the last two; they are for the locator inset.

4. Style City limits with no fill (Simple Fill, Fill style: No Brush) and a dark dashed outline, and Parks with a green fill, no outline and about 80% opacity (the opacity slider under the symbol).

Check: the dashed city outline and green parks sit on top of the tracts.

Part 2: Rule-based styling with a “no data” class

1. Add a virtual field to Tracts with the Field Calculator: pct_no_vehicle, Decimal (double), "hh_no_vehicle" / "households" * 100.
2. Properties > Symbology: Graduated on pct_no_vehicle, Natural Breaks (Jenks), 5 classes, ramp YlOrRd, white outlines. Click Classify, then Apply.
3. Change Graduated to Rule-based at the top. QGIS converts each class into a rule with a filter such as "pct_no_vehicle" > 2.38 AND "pct_no_vehicle" <= 8.59.
4. Click the green +, give the new rule the filter "pct_no_vehicle" IS NULL, the label “No households (dorms)” and a light gray symbol with a hatch (Fill style: BDiagonal).
5. Double-click each class label and round it (“0 to 2%”, “2 to 9%” and so on). Click OK.

Rule-based symbology converted from Graduated, with a rule for no data.

Check: the residence hall tract near campus is drawn in hatched gray instead of disappearing.

Part 3: Labels by rule, with halos

1. Parks > Properties > Labels: choose Rule-based Labeling and click the green +.
2. Filter: $area > 300000 (parks over 30 hectares; $area is in square meters). Value: name.
3. In the rule’s Text tab set the size to about 8 points and a dark green color; in the Buffer tab tick Draw text buffer, size 1 mm, white. That buffer is the halo.
4. Click OK twice.

Rule-based labeling for parks over 30 hectares.

Check: six parks are labeled, each with a white halo. If no labels appear and the numbers in the Preview of $area look far too large, open Project > Properties > General and set Ellipsoid to WGS 84, which is how QGIS measures $area.

Add the two city names as labels in the layout later, or as a small point layer: places are often labeled by hand, in italics.

Part 4: A layout with a locator inset

1. Project > New Print Layout, page size Letter (Layout tab, right panel, after right-clicking the page and choosing Page Properties).
2. Main map: Add Item > Add Map, drawn over most of the page. Zoom it to the cities with the Move Item Content tool and the mouse wheel. In Item Properties > Layers, tick Lock layers so the map keeps today’s layers.
3. Locator: in the main QGIS window, tick only Indiana counties and Tippecanoe County (styled gray and red) and zoom to Indiana. Back in the layout, add a second, small map in a corner; tick Lock layers for it too. Then turn the city layers back on in the main window.
4. In the locator’s Item Properties > Overviews, click + and set Map frame to the main map (Map 1). A rectangle shows where the main map is.
5. Add the title (stating the finding), a legend (untick Auto update, remove the county layers, and in the legend panel hide the “Tracts” group heading if you like), a scale bar in kilometers, a north arrow and a source line, as in the Foundations episode.

The print layout with the main map, the locator inset and the map items.

Part 5: Export

1. Layout > Export as PDF for print (vector, sharp at any size).
2. Layout > Export as Image for screens: PNG, with Export resolution 200 to 300 dpi.
3. Open both files and check sizes, fonts and the legend.

The finished map.

Callout

One layout, many maps: atlas

The layout Atlas (Atlas menu in the layout window) repeats a layout once per feature of a coverage layer, for example one page per tract or per county, and exports them all at once. It is the QGIS way to make dozens of consistent maps.

---

Challenge

Challenge 1: What is wrong with this map?

A map of poverty rates by tract uses a satellite basemap at full opacity, red and green classes, labels for every street, a scale bar on a Web Mercator map of the whole United States, and no source line. List the problems and a fix for each.

Show me the solution

• Satellite basemap: too busy for a thematic map; use a quiet gray basemap or none, and keep the attribution.
• Red and green classes: unreadable for many readers with color vision deficiency, and a diverging scheme for a one-directional rate; use a sequential palette (see the Accessibility episode).
• Labels for every street: clutter; label only what orients the reader.
• Scale bar on a national Web Mercator map: the scale varies greatly with latitude; drop it or use an equal-area projection and a note.
• No source line: add the dataset and year.

Challenge

Challenge 2: Two maps, one scale

Make a second layout page (or a second layout) with two maps side by side: car-free households and the poverty rate ("below_poverty" / "poverty_universe" * 100). Should they share classes? Set them up accordingly.

Show me the solution

Both are percentages and the point is to compare them, so share the classes: in both layers use Graduated with Mode: Fixed Interval or manual breaks (edit the values) at 10, 20, 40 and 60%, with the same ramp. Use Lock layers and Lock styles for layers in each map item so each keeps its own layer. One legend is enough. With shared classes it is obvious that poverty rates reach much higher than car-free rates.

Key Points

• Style each layer by its role; the subject is strong, context is quiet
• Convert Graduated to Rule-based to add a “no data” rule; label by rule with a text buffer (halo)
• A second map item with an Overview makes a locator inset; Lock layers keeps each map’s content
• Scale bars need a projected CRS; small multiples share classes only for comparable values
• Export PDF for print and PNG at 200 to 300 dpi for screens, and check the exported file

Content from Accessibility in Map Design

---

Last updated on 2026-10-17 | Edit this page

Overview

Questions

• Why is accessibility important in cartography?
• How do different types of color vision deficiency affect map reading?
• How can we design maps that are readable for people with color vision deficiency?
• What roles do hue, saturation and lightness play in accessible design?
• How can I choose and test accessible palettes in QGIS and Python?

Objectives

• Explain why accessibility matters in map design
• Describe the main types of color vision deficiency (CVD)
• Apply color-safe cartographic principles, especially varying lightness
• Select accessible color palettes in QGIS and Python
• Test maps with simulators and write useful alt text

Why accessibility matters in maps

---

Maps are communication tools. If part of your audience cannot read your colors, your map has failed for them.

Accessible maps:

• Reach wider audiences
• Are easier to read for everyone, including on poor screens, projectors and greyscale printouts
• Reduce misunderstanding
• Are increasingly required for public and university websites and publications

Callout

Key idea

Accessibility is not an extra step at the end. It is part of good cartographic design.

---

Understanding color vision deficiency

---

Color vision deficiency (CVD), often called colorblindness, changes how some people distinguish colors. Most people with CVD see color, but some pairs of colors look alike to them.

Among people of Northern European ancestry, about 8% of men and 0.5% of women have a red-green color vision deficiency; rates are somewhat lower in many other populations. In a room of 30 people, it is likely at least one person has CVD.

Types of color vision deficiency

---

The eye has three kinds of cone cells, sensitive to long (red), medium (green) and short (blue) wavelengths. CVD occurs when one kind is missing or works differently. When a cone type works differently, the condition is called an anomaly (for example deuteranomaly); when it is missing, an -opia (for example deuteranopia).

Red-green deficiencies (by far the most common)

• Deuteranomaly and deuteranopia (green cones affected): the most common forms. Reds, greens, oranges and browns are easily confused.
• Protanomaly and protanopia (red cones affected): similar confusions, and reds look darker than they do to other people.

Blue-yellow deficiencies (rare)

• Tritanomaly and tritanopia (blue cones affected): blues and greens, and yellows and violets, are confused.

Complete color blindness (very rare)

• Achromatopsia (monochromacy): little or no color perception. Maps look almost entirely grey, so only lightness differences remain.

Challenge

Challenge 1: Why do red-green maps fail?

Why might a choropleth map that runs from green (low) through yellow to red (high) fail for many readers?

Show me the solution

Red-green deficiencies are the most common type of CVD. For these readers, the green and red ends of the palette can look like similar olive or brown tones, especially when they have similar lightness. The lowest and highest classes, the most important ones to tell apart, can become confused. The example in the next section shows exactly this.

---

Seeing the problem

---

The figure below shows the same map of Indiana county unemployment twice: once with a red-green palette and once with the viridis palette. The right column simulates how each looks to a person with deuteranopia.

The same Indiana unemployment map in a red-green palette and in viridis, with simulated deuteranopia

With the red-green palette, the lowest class (dark green) and the highest class (red) both turn olive. With viridis, the colors change but the order from light to dark is preserved, so every class can still be read.

---

Common mapping problems

---

Poor design choices include:

• Red versus green to mean bad versus good, or high versus low
• Classes that differ in hue but have similar lightness
• Many hues with low contrast between them
• Relying on color alone to carry meaning
• Thin lines or small symbols whose color is hard to see

---

The solution: hue, saturation and lightness

---

Every color can be described by three properties.

Hue

The color family: red, blue, green and so on.

Avoid: red with green; blue with purple when tritan readers matter.

Saturation

The intensity or purity of the color, from grey to vivid.

Use: moderate saturation. Very saturated neon colors are tiring and can bleed together.

Lightness (value)

How light or dark the color is. This is the most important property for accessibility.

If classes differ clearly in lightness, they remain distinguishable even when hues are confused, and even in greyscale.

Callout

Best practice

When in doubt, vary lightness more than hue. A good test: convert your map to greyscale. If you can still read the order of the classes, most readers with CVD can too.

---

Designing accessible maps

---

Use more than color

Combine color with a second cue:

• Patterns or hatching for categories
• Labels for key values or places
• Different symbol shapes for point categories (circles, squares, triangles)
• Different line styles (solid, dashed, dotted)

For example, instead of red and green points for two kinds of facilities, use blue circles and orange squares.

---

Recommended colorblind-safe palettes

---

Sequential data

• Single-hue ramps: Blues, Greens, Purples, Greys
• Multi-hue ramps designed for CVD: viridis, cividis, YlGnBu

Diverging data

• Blue to orange (for example ColorBrewer PuOr or BrBG)
• RdBu (red and blue, not red and green)

Categorical data

• ColorBrewer Set2 and Dark2
• Okabe-Ito, a palette of eight colors designed specifically for CVD
• Tableau 10 (check with a simulator)

On ColorBrewer, tick colorblind safe to see only palettes that pass for the common types of CVD.

Callout

Avoid rainbow palettes

Rainbow ramps (such as “jet” or “Spectral” with many classes) have no natural order, create false boundaries where the hue changes quickly, and include red-green pairs. Use them only for genuinely cyclical data, and even then with care.

---

Accessible palettes in QGIS

---

Choosing a ramp

1. Right-click the layer and choose Properties > Symbology.
2. Choose Graduated (for numbers) or Categorized (for categories).
3. Open the Color ramp drop-down and choose a ramp.

Good choices in QGIS: Viridis, Cividis, Magma, Inferno, Plasma, and the ColorBrewer ramps such as Blues, Greens, YlGnBu, PuOr and BrBG. If you do not see a ramp, choose All Color Ramps in the drop-down.

Avoid: Spectral and other rainbow ramps for ordered data, red-green diverging ramps such as RdYlGn, and very saturated custom ramps.

Testing with QGIS preview modes

QGIS can simulate how your map looks with color vision deficiency:

1. In the main QGIS window, go to View > Preview Mode.
2. Choose Simulate Color Blindness (Deuteranope), then try (Protanope). Recent versions also offer Tritanope and Achromatopsia, and all versions offer greyscale (“Photocopy”) previews. The exact list depends on your QGIS version.
3. Check that adjacent classes are still distinguishable.
4. Return to View > Preview Mode > Normal when you are done.

Preview modes also work in the Print Layout window, under View > Preview.

---

Accessible palettes in Python

---

PYTHON

# Sequential, colorblind safe
gdf.plot(column="unemp", scheme="Quantiles", k=5, cmap="viridis", legend=True)

# Single hue
gdf.plot(column="unemp", scheme="Quantiles", k=5, cmap="Blues", legend=True)

# Diverging around a meaningful midpoint (choose a red-blue or purple-orange ramp)
gdf.plot(column="change", cmap="RdBu_r", vmin=-5, vmax=5, legend=True)

Notebook 1c in this module includes a step that simulates color vision deficiency in Python.

---

Testing your map

---

Always test before you publish.

Tools:

• QGIS Preview Mode (above)
• Color Oracle: a free desktop app that filters your whole screen
• Coblis: upload an image to see it with different types of CVD
• Convert the exported image to greyscale in any image editor

Ask: can someone tell the classes or categories apart without relying on hue alone?

---

Accessibility beyond color

---

Color is only part of accessibility. Also consider:

• Readable text: large enough fonts (at least 8 to 9 points on printed maps), plain fonts, good contrast with the background
• Clear legends in the same order as the data
• Alt text for every map shared online (see below)
• Screen reader friendly web maps: a data table or text summary alongside an interactive map
• No meaning conveyed only by hovering, which keyboard and touch users may not be able to do

Writing alt text for maps

Alt text is read aloud by screen readers in place of the image. For a map, describe what it shows and its main pattern, not every detail.

Weak: “Map of Indiana.”

Better: “Map of Indiana counties shaded by 2016 unemployment rate in five classes, from 3.1% to 6.6%. Rates were highest in several northwestern counties near Chicago and in parts of southern Indiana, and lowest in central and northeastern Indiana.”

Every figure in this module includes alt text written this way.

---

Example: good versus bad design

---

	Bad	Good
Palette	Red to green with equal lightness	Viridis or a blue to orange palette with strong lightness steps
Encoding	Color only	Color plus labels or patterns
Legend	Small, unordered	Large, ordered low to high
Online	No alt text	Alt text describing the main pattern

---

Accessibility checklist

---

Before publishing:

• No red-green contrasts
• A colorblind-safe palette
• Clear lightness differences between classes (passes a greyscale test)
• Labels, patterns or shapes in addition to color where categories matter
• Tested with a simulator (QGIS Preview Mode, Color Oracle or Coblis)
• Readable font sizes and a clear legend
• Alt text for maps shared online

Challenge

Challenge 2: Fix a palette in QGIS

Load the Indiana counties layer (https://raw.githubusercontent.com/SpatialTurn/Foundations/main/data/indiana_counties.geojson, as in the Foundations module’s Getting Started with QGIS episode), style unemp_2016 with Graduated symbology and the RdYlGn color ramp. Switch to View > Preview Mode > Simulate Color Blindness (Deuteranope). Then change the ramp to an accessible one and preview again.

Show me the solution

With RdYlGn under the deuteranope preview, the low and high classes look like similar olive tones. After switching to Viridis, Blues or YlGnBu, the classes run clearly from light to dark in the preview, and the map can be read correctly. Remember to switch back to Normal preview mode.

Challenge

Challenge 3: Write alt text

Write alt text for the multivariate map of Indiana in the Fundamentals of Map Design episode (orange fill for unemployment, grey circles for population).

Show me the solution

One possible answer: “Map of Indiana counties shaded in five orange classes by 2016 unemployment rate, with grey circles sized by county population. Marion County (Indianapolis) has by far the largest circle. Several northwestern counties near Chicago combine large populations with high unemployment, while many small rural counties in the south also have high rates.”

Final takeaways

---

Accessible maps are clearer, more inclusive and more professional. Good cartography means designing for all readers.

Discussion

• Have you seen a map that was hard to read because of its colors?
• How could accessibility improve communication in your own research area?

Key Points

• About 1 in 12 men of Northern European ancestry has a red-green color vision deficiency.
• Vary lightness, not just hue, so classes stay distinguishable for everyone.
• Use colorblind-safe palettes such as viridis, cividis, Blues or ColorBrewer’s colorblind safe schemes.
• Add a second cue (labels, patterns, shapes) when categories matter.
• Test with a simulator such as QGIS Preview Mode, and write alt text for maps shared online.