Accessibility in Academic Research: Using Color

So my fiancé is a PhD student. That fact is only relevant in that it’s my main source of information about processes at various academic institutions. While he is a student and works for only one university, he is on projects with other universities, and regularly reviews papers for journals.

There are times he shows me or tells me about websites, surveys, graphs, or practices that clearly violate current accessibility standards. At times he has mentioned this to the people in charge of these items and he frequently gets the response along the lines of “I don’t think people with visual impairments will use this.”

As someone who strives to ensure accessibility practices are followed, this response makes me frustrated. But, rather than having this be a series with me venting about these transgressions, I’d like to take this time to give a brief overview on how academic faculty and staff (or other researchers) can build accessibility practices into their workflow and why they should.

This series is going to focus on a few main problems with accessibility in academic research, but by no means are they they only ones. If you’re curious about other accessibility considerations, take a look at Anne Gibson’s Alphabet of Accessibility on the Pastry Box Project; it’s fantastic.

Color usage in research graphs/charts

Most people experience normal color vision. Most people tend to create images that are visually distinct to them. If they can tell the difference between two colors on a graph or chart, they believe their audience will be able to as well.

However, color blindness affects about 1 in 12 men and 1 in 200 females. There are an estimated 300 million people worldwide that experience some type of color blindness. This means that there’s a high likelihood that someone you have met or know has some form of color blindness.

The most common type of color blindness is red or green color blindness. However, there are other forms of color blindness, including blue deficiency and total color agnosia (everything appears as shades of grey).

All color blindness can vary in its severity, and isn’t limited only to the deficient colors. Colors with red, blue, or green in them can be difficult to discern as well. For instance, purple is comprised of varying shades of blue and red. When you take away the red from your vision, blue and purple can look very similar.

Lots of scholarly research includes graphs and charts to represent data and results. However, depending on your color scheme, and how you’ve set up your chart, it may not be readable by everyone.

Example of a Bad Graph

A bar graph with seven different colors to represent seven different items. There is a color legend on the right of the graph. This graph looks distinct for users with normal color vision.
This is a graph that appeared in a reputable online academic publication. To those with normal color vision, this graph does a great job of separating out the items by color.
A bar graph with seven different colors to represent seven different items. There is a color legend on the right of the graph. This graph looks distinct for users with normal color vision.This displayes the earlier full color graph as it would appear to a user with protanopia (red deficiency)
Using the slider you can see on the left what the graph looks like to users with normal color vision, and on the right, what it looks like to users with red deficiency (protanopia). In the second graph, the blue and purple colors used to indicate blueberries and grapes look very similar, as do the green and orange colors used to indicate apples and tangerines.
A bar graph with seven different colors to represent seven different items. There is a color legend on the right of the graph. This graph looks distinct for users with normal color vision.This is the same graph from earlier, but the color is gone. It appears like it would to a user with monochromacy.
Using the slider you can see on the left what the graph looks like to users with normal color vision, and on the right, what it looks like to users with no color vision (monochromacy). In the second graph, the blue and orange colors used to indicate blueberries and tangerines are difficult to distinguish.

How to Fix the Problem

I will mention that the legend in the above graph does list the items in the order they appear, but that’s not always possible in graphs, nor is it always how it is done. So, what can we do to adjust these graphs to make them more color blind friendly?

  1. Vary the hues/saturation of the graph colors so they appear different to users with monochromacy. For instance, in the full-color graph above, you can see that the color for peaches is much lighter than the color for tangerines. In the monochromacy graph below it, you can see that the two bars on the right of the graph have different levels of white and black. The peach displays as a much lighter grey than the tangerine, making it visually distinct.

2. Use patterns in data visualization to convey the different data sets. The below image shows an example of a graph that uses patterns to convey the differences, so that color isn’t the distinguishing factor. Regardless of the colors a user can see, they will all be able to pick up on the different patterns and will be able to easily differentiate them.

This image shows a bar graph with three different bars. The left-most bar is filled solid with a turquoise color, the middle bar has a turquoise mesh pattern on top of the white background, and the last bar has a diagonal striped pattern on the white background.
This graph shows one way to distinguish different areas of the graph using different patterns to differentiate the data.

3. Don’t rely on color/legends alone to identify the data; use labels. Many graphs use a color legend to help viewers identify what the different parts of the graph represent. However, for users with color deficiencies it can be difficult to see and identify the colors in the legend, and on the graph. Adding in labels to the different data sections avoids that problem neatly. It clearly labels what the viewer is looking at (if they don’t need alt text-that’s a separate post). The graph below demonstrates using labels to identify the different data.

This is an example of a graph that doesn't use color as the only indicator to differentiate the data, they also include labels on the different bars of the graph.
This graph uses color to distinguish the bars on the graph, but it also uses labels on the bars to differentiate them as well.

Resources

So, how can you easily incorporate this into your workflow? Chrome has several plugins that mimic what different types of color blindness look like, so that you can inspect your work as you go. I recommend the Chromacy Chrome plugin to check your work online.

If you want something that will show you what it looks like on work you’re doing on any software you have on your computer, you can download Color Oracle (free) to use on your computer. It is compatible with both Mac and Windows computers.

If you’re creating work in JupyterLab, my fiancé developed a plugin to help you check for color blindness problems in the software. Neither he nor I gets any kind of compensation from you using this. You can see how to use it in this GIF below.

For help figuring out which colors to use (if you aren’t using patterns or labels), you can use ColorBrewer to find compatible colors. While this website helps generate colors for maps, you can use the same colors on something else once they’ve generated it; just please credit them for it. At the top, just select the number of data classes you have, and it will generate a palate for you (max is 12).

These are by no means the only tools you can use, just some of them. I hope you’re able to use them in your work to make your research more accessible!

The Impact of Research Accessibility on Future Research, Part 2: The Effect of Open Access Articles on Research

This is a continuation from last month’s post Part 1: Researchers’ Information Seeking Behavior, so if you haven’t read that yet, I would suggest you read it first.

In the post, I explored the question, how much of future research is impacted by whether someone can quickly and cheaply gain access to other research? Last month’s post focused on how researchers’ information seeking behavior has an impact on the research they find, and therefore, future research. This month’s post explores how open access research has made an impact on emerging and future research.

Continue reading The Impact of Research Accessibility on Future Research, Part 2: The Effect of Open Access Articles on Research

The Impact of Research Accessibility on Future Research, Part 1: Researchers’ Information Seeking Behavior

I was recently visiting a friend who has recently started a Master’s and PhD program at a large U.S. university; while I was visiting him, he had to work on his research, as many graduate students have to do in their “spare” time. After doing some searching he determined that the nearest book available on the topic he wanted was in Ireland, and since he needed the information in the next couple of days, he decided not to request the book through interlibrary loan (ILL). This got me thinking, so much of what people do happens so quickly, that there isn’t much time to wait to get access to research. This made me wonder, how much of future research is impacted by whether someone can quickly and cheaply gain access to other research?

Taking the book from Ireland, for example, how much does the fact that the nearest copy of a publication is in another continent affect the future of someone’s research? If I’m working in a field that’s emerging, it stands to reason that there’s not an abundance of publications available on these topics. And if I can’t access a certain piece of information I want quickly or cheaply enough, I’m likely to forgo reading it, and rely on other sources that comply with my time and money constraints. If research builds upon other research, does the inaccessible research get lost?

It’s nearly impossible to design and carry out a study that could measure this definitively, because there are so many variables like the subscription resources offered from institution to institution, the availability of research from different disciplines, and the variability of how quickly a researcher needs the research they seek.

As I started looking into this question, I realized there are many components to this, so I focused on two areas, and I’m publishing this post in two parts. The first part, is about researchers’ information seeking behavior, and how their behavior towards research that is not quickly accessible may affect future research.

The next post will be about how open access and subscription articles affect the future of research.

Continue reading The Impact of Research Accessibility on Future Research, Part 1: Researchers’ Information Seeking Behavior