Data Storytelling 5: Bar Charts
I know you think you already know bar charts and don't need to read this, you're exactly the type of person who *should* read this
Picking up where we left off on the last storytelling post.
0 - Area Graphs
If you’ve never heard of them, here’s a quick visual:
They suck, never use them. Next graph.
1 - Vertical Bar Chart
A lot of new data professionals like to avoid bar chart because they think bar charts are common. They need to distinguish themselves by picking a different type of graph. This ends up being a blunder, instead of avoiding bar charts because they are common, you should leverage them instead because since they are common, it means your audience is already quite familiar with them.
An advantage of bar charts is that they are easy for our eyes to read (quickly). Our eyes can see which line is the tallest, and which line is the shortest, and thus arrive at whatever specific conclusion between a few categories you wanted your audience to have. Since a bar chart doesn’t take much thought process for your audience, it means they will be very likely to *actually* read the graph, instead of just skimming over it.
You should use this to your advantage.
1.1 Y-Axis Labels
When you are working with bar graphs, you need to make your y-axis start with 0. Sometimes (usually news outlets) try to deceive their audience by having the y-axis start at a higher number.
You can see an example by Fox News below:
The graph on the left is what was presented to their audience for a few seconds. The graph on the right is what it should have looked like, with proper scaling. Needless to say, when the BowTied Jungle basically makes fun of news outlets, this is why.
You might be tempted to try to do a presentation where the y-axis doesn’t start from 0, but rather a different number (like above), to try to sell a conclusion to your audience. Here’s the problem with that, if even a single C-Suite calls you out on it… you’re done.
No chance to defend yourself, no second chances, you’re just done…. All the work you’ve done immediately gets questioned, and your trust goes to 0
2 - Stacked Vertical Bar Chart
A stacked vertical chart can be a huge hit or miss. In theory it’s meant to allow you to compare totals across different categories, and see many different sub-components within each of the different categories.
Here’s the problem though: This can very quickly become visually overwhelming to your audience. And, like I said in my previous posts, your audience doesn’t want to do any thinking
when they are looking at your visuals, the less thinking they have to do, the more likely they’ll actually read it.
The biggest problem with Stacked vertical bar charts is that once you get beyond the bottom series because you no longer have a consistent baseline to use to compare, it makes comparisons harder for our eyes to make. Here’s an example:
After the light blue comparison, doing any sort of comparison between them becomes harder and harder.
3 - Waterfall Chart
The difficult thing about the stacked bar chart is that comparing each individual sub-category. So, a waterfall chart aims to solve that problem, by pulling apart the pieces of a stacked bar chart. If you’ve never seen a waterfall chart before, here’s what it looks like:
This graph is basically showing how we started off with 4000 sales at the start of 2013, and how we eventually ended up with 7375 sales at the end of the year.
Now, you can produce many waterfall charts for multiple products. Then you can visually see if there’s any sort of pattern between the different products vs the different months.
4 - Horizontal Bar Charts
A horizontal bar chart basically takes a vertical bar chart and rotates everything by 90 degrees. Which means when someone is reading the graph, they are reading it left to right, then down left to down right (Z shape). This means that whereas a vertical bar chart is easy to read, a horizontal bar chart is EXTREMELY easy to read. And, remember what I said about your audience not wanting to use their brains?
Here’s a side by side comparison of a vertical vs horizontal bar graph:
Your audience will tell you they have no preference, but I’ll be direct with you. They prefer the horizontal format because it’s easier to read. They are literally reading the category first, and then seeing it’s value, and then going downwards (Z Shape).
5 - Stacked Horizontal Bar Charts
Now, if you just take a stacked vertical graph, and then rotate it, it’s actually not going to help you much here (other than being slightly easier to read). Example below:
What you’ll want to do instead is to convert all the values to percentages. This way, the entire graph can be scaled to 100%. This means you can now compare each sub-categories much better. Take a look at this graph below, see how it’s much easier to read than the one above:
The goal of the stacked bar graphs is to compare the sub-categories within each main category. This is why the conversion to percentages help. This way, they have 2 easy base comparisons they can do. The one on the left (blue - Marine Sprite), and the one on the right (purple - Reborn Kid)