Dashboards That Pop: Bar Charts
Bar charts are one of the most widely used charts around. Just because they are widely used, doesn't mean they are used well. Learn some bar chart tips and best practices to keep your ArcGIS Dashboards looking top-notch.
When to Use
Bar charts are used to show categorical or grouped data. They are easy for users to understand, making them one of the most popular chart choices around. Humans are very good at comparing lengths when objects are aligned.
Because we are judging the bars based on their relative size to one another, and not really paying much attention to the value axis labels, it is vital that you use a zero-based value. Truncating your value axis to start at anything other than zero can distort your data. Look at the two charts below. In the charts below the Strategic development category completely dominates Riots/Protests. In the bottom chart, we can see that while Strategic development is quite a bit more than Riots/Protests the difference isn’t nearly as bad as the top chart led us to believe. Both charts use the exact same data, the only thing that is different is where the value axis starts.
To ensure that your chart will always use zero as the baseline, set the minimum value to 0.
Horizontal vs Vertical
For the most part, deciding between a horizontal or vertical chart is a matter of style and space. However, there are certain situations where one orientation is better than the other.
Temporal Data
We are used to thinking about time as moving along a straight line, as we go from left to right, we are moving forward in time. Your temporal bar charts should mirror this – with your date/time field along the horizontal axis.
Negative Values
Similarly, we tend to associate positive values with up, and negative values for down. Which means, you should chart your negative values along the vertical axis.
Tip: To really highlight that the data moves from positive to negative, I’ve added a guideline at 0.
Long Categories
Long category names are easier to read and fit better along the vertical axis than they do on the horizontal.
Stacked Bar Charts
Stacked bar charts show two categorical variables. The primary variable consists of the entire length of the bar, while the secondary variable is represented by stacks within the bar.
When to Use
If you need to talk about both the primary and secondary variables, but your main focus is the primary variable, a stacked bar chart could work. If you are trying to compare the secondary variables across each bar, then you should use another chart.
Looking at the below stacked bar chart, what can you determine about the data? We can easily see how each state compares to one another overall. We can maybe discern that for the most part; category 3 is the largest category across the states. We can easily compare category 1 across states, because that subcategory is the only subcategory that is aligned.
Trying to compare categories 2 and 3 across the states is difficult. Can you really tell which state has the highest number of category 2 between CO, CT, and DC? Does CT or ID have the most category 3’s?
If you really want to compare subcategories to one another, consider a clustered bar chart instead.
Stacked 100% Bar Charts
These charts are stacked bar charts that show percentages rather than absolute values. The same general principles apply to stacked 100% bar charts as they do to stacked bar charts. The one difference is that we now have 2 subcategories that are aligned, making them easy to compare.
In the chart below, category 1 is aligned with the bottom, and category 3 is aligned with the top, making it easy to compare those subcategories across bars.
Bar Chart Best Practices
Categories
Keep the number of categories to about 5 or less. Anything more than that it becomes hard to find colors that are easily distinguishable, and the chart just becomes cluttered. The whole point of data visualization is to help the user quickly and easily understand the data presented. If the users have to spend a lot of time and mental effort to understand a chart, your visualization has failed. If after all that effort the user still doesn’t understand the data, your visualization has failed. If your user doesn’t bother to even try to understand a chart because it’s just too much, your visualization has failed.
Ordering
Be sure to order your bar charts in a meaningful way. Take advantage of any inherent ordering in your data. For example, temporal data should move forward in time from left to right, or age ranges should be ordered from youngest to oldest.
If there is no inherent order available, then you should order your chart based on the value, either from smallest to greatest, or vice versa depending on the story you are telling.
If you think users will be looking for specific categories, say seeing their state’s specific value as opposed to where it falls in the pack, you could consider ordering your data in alphabetical order.
The key to ordering is to think about what makes sense for your data and how it is being used.
Color
I have a whole separate blog just on using color in your dashboards (link at the end), so I won’t really get into color here. I will say that if your chart shows ‘one thing’, you should be using one color. In the chart below, we are showing “one thing’ - the number of colleges in each state. There is no reason to color each state differently.
I do tend to break this rule however, when my bar chart is categorized the same way that my accompanied map is categorized. In those cases, I will color each bar the same color that is used for that category in the map.