Barcharts could be worse, but we can do better.

What is a barchart?

Barcharts are a way to compare categories of a categorical variable. What are they used for? Much more than that.

In Principles of Biology at my university, student are required to create a barchart showing the data

Why is this bad?

The amount of information conveyed by this vizualization technique is scant compared to the amount of space it takes up. When a barchart is used to show the mean or median plus and the ‘error bars’, the data is compressed such that the reader has to trust the authors interpretation of the data.

In many experiments in cell and molecular biology, the number of samples per group is less than 5, sometimes as low as three. Generating a bar

Psychological implications of barchart design

When we use bar charts to represent are data, a few probelms arise. First, the choice of standard deviation (SD) vs standard error (SE) arises. The majority of papers inappropriately use SE because the value is smaller, suggesting a greater difference between groups. In the modern era, is it really acceptable for us to not show our data in this way?

The second concern of a bar chart is lay interpretation. Given a bar chart, we shade in the values below the number, but not above. if the bar chart is being used as a summary statistic, then we are not saying values are more likely to be above or below the top of the barchart, but graphs of this style are often interpreted thusly, according to psychological research [CITE].

The state of the field.

I believe two primary factors (1) awareness and (2) ease of implementation have brought us to where we are now. The solutions to this aren’t easy, but I blieve we can.