From Bad to Better: Bar Charts
The average admin assistant has little control over the way in which scientists present their graphical data. I like to make sure that anyone who wants help can get it. The usual way is to steer them toward some design books. Tufte is great when you’re thinking about abstract stuff, but the best way to reach scientists is to move from before to after.
For example
It all started with a slightly worn bar-chart printout from my boss. He informed me that the creator left, and we need to get the chart ready for inclusion into a paper.
As you can see, I got a printout of a very basic Excel bar graph. In order to save time (and better describe the situation), let’s just say I got some Excel garbage. Time to roll up the sleeves and see what we can do with it.
The first thing to do is to effectively draw over the chart in my image editor of choice: Fireworks. By doing this, I can preserve the exact measurements of the bars and take care not to unintentionally alter the data. After that, the X-axis lines are lightened so they interfere less with the bars.
Additional issues
Those of you who work with scientists know that you can get some really awful looking graphs with the basic Excel settings. Take note of the following problems:
- The colors have no meaning whatsoever.
- The bar labels are completely disconnected from the chart itself. Meanwhile, the Y axis is completely unlabeled.
- The bar labeled “Teff” is a baseline of sorts, which all the data is meant to contrast with. There is no way to know this.
- What is the title of the chart?
Final alteration
To finally take the chart from bad to better, here’s what we have:
The colors are simplified, the Y-axis contains data and is organized by size. By making a few strategic changes, we have a much nicer looking graph.
Final note: Ensure that you make all source data (in this case, the tabular data) available to any who request it.


