I am participating in January's Storytelling With Data challenge of creating a small multiples chart. I work with education data a lot and one struggle I've encountered is how to visualize academic data that shows an achievement gap between races/ethnicities.
The Challenge of Achievement Gap Visualizations
The difficulty is that there are often many variables that need to be included (i.e. race/ethnicity, school/grade, subject) and each of those variables have multiple categories (e.g., Black, White, Hispanic, Asian).
I've tried several ways to visualize achievement gaps in the past: dot plots, clustered bar charts, and bar charts with "target" lines. However, I hadn't tried small multiples.
The Data
I started with real 3rd Grade Reading Proficiency data from the Ohio Department of Education, but removed the school names and changed the data points slightly to anonymize the data.
Below is what the data table looks like. At a glance, you can see that the scores in the White column are usually higher than the others. However, it's nearly impossible to take away a clear understanding of how the other groups relate to each other and how big the gaps are from this table. That's where data visualization comes in.
The Visualization
Here is what a set of small multiples bar charts look like for a district with six elementary schools. I used this trick to make my bar charts identical for each school.
The Critique
I still feel the visualization may be too visually overwhelming. I tried to make it easier on the eyes by consistently color coding each racial/ethnic group across the schools but with so many schools and so many racial/ethnic groups, it still feels like too much to take in at once.
In addition, I think the display would benefit from more white space between the schools, but I ran into issues with the standard page width and trying to avoid the race/ethnicity labels auto-rotating.
The Wrap-Up
I think this is the best visualization option that I've tried for the achievement gap, but I don't think it's perfect. I'm curious if anyone else has encountered this issue and how they resolved it.
Do you think this is a good way to visualize achievement gap data? How could the charts be improved?
The Revision
Based on feedback from other data experts at Storytelling with Data, I revised one school's data visualization. I rotated the chart to make the bars horizontal instead of vertical so the labels were not as smashed together, reordered the bars so that the group with the highest proficiency is first, and added outlines to indicate the difference, or gap, between the largest bar and the others.
To add the "empty" boxes showing the gap, I added a second column of data representing the difference between 100% and each group's percentage. Then, I changed the chart type to a stacked bar chart. I removed the fill color for the new segment of the bar, but left the outline, which I made the same color as the bar.
I think this change does a good job of emphasizing the gap, if that is your main focus with this data. What do you think?
