Data Visualization Design, Part 3: Chart Junk in Data Science


As data scientists one of the tasks we have is communicating what we’ve found during our data analysis to various types of audiences. Data visualization is how those findings are shared. Today I want to continue the topic of data visualization by focusing on the role of chart junk in data science. In particular, I want to explore a paper by Scott Batesman et al. called ‘Useful Junk? The Effects of Visual Embellishment on Comprehension and Memorability of Charts’.

The main argument Batesman makes is interesting since it tries to quantify the connection between ‘chart junk’ and interpretation accuracy and long-term recall. The authors argue that using strong images in charts like those used by Nigel Holmes is sometimes appropriate for graphical visualizations depending on the medium and context. They argue the extra encoding does not negatively affect whether people can interpret and remember the information they saw. However, it’s important to note that Batesman et al. don’t advocate extra ink as a general principle and they state their research is a starting point for ‘this phenomenon should be better understood by researchers and chart designers.’ It’s important to not read this one paper out of context as drawing a conclusion that one design principle is better than the other.

The research methods in the Batesman paper can’t accurately represent the general population’s reactions to ‘chart junk’ versus plain charts. The viewers in this research were 20 university students so there is probably an age biased not accounted for in the results. Additionally, it would be interesting if there were differences in other parts of the population (age, gender, income, etc). This is one snapshot in time from a relatively small number of people.

The Holmes graphics were also presented without medium context. For example, seeing ‘chart junk’ on a social media website might be entirely more appropriate than seeing it in an academic journal since the audiences and formats are different. Social media content has to be more engaging to capture our ever shorter attention spans. It’s also important to note that ‘chart junk’ is by nature adding an artistic element to the information whereas the plain charts Tufte’s principles describe do not have this artistic element. Before using embellishments in my data visualizations, I would proceed with a lot of caution and keep in mind Tufte’s minimalist design principles.

Next week I’ll give a few examples of removing chart junk from visualizations.


One thought on “Data Visualization Design, Part 3: Chart Junk in Data Science

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s