Potential Information Visualization Research Projects

Last month I spent a great deal of time thinking and writing about information visualization research, mostly bemoaning ways in which it usually misses the mark. A few days ago, Steven Franconeri of Northwestern University welcomed my invitation to talk about infovis research projects that would address real problems. When he asked if I already had a list of potential projects, I admitted that I haven’t written one, but offered to do so. I really should write down ideas for potential research projects when they occur to me, but I haven’t had a convenient place to record them. I’ll fix this soon. A few days ago I took a few minutes to scour my memory for ideas from the past, and quickly made a list of 18 potential projects. Before I publish a list, however, I’d like to collect ideas from you as well.

Infovis researchers would benefit from hearing about the problems that practitioners currently struggle to solve. It can be difficult to know what’s actually needed when you spend most of your time at a university, whether you’re a student or a professor. Infovis is still a young field with much to learn and much to develop. We who use data visualization in our work can help ourselves by helping the research community to understand what’s needed.

Infovis research projects fall into a few different categories. In general, they 1) study phenomena related to data visualization that we don’t fully understand through observation and experiments, 2) develop potential solutions to problems and test them to see if and how well they work, or 3) develop new conceptual structures (a.k.a., taxonomies) for understanding data visualization. Here’s a potential example of each:

  1. Determine the effects on a scatterplot’s aspect ratio for interpreting the existence and nature of correlations.
  2. Develop and test a means to trigger blind sight (seeing things without conscious awareness) as a way of drawing someone’s attention to a particular area of a visual display, such as a particular piece of information in a dashboard that needs attention.
  3. Develop a clear way for people to think about the differences in which data visualizations should be designed to support data sensemaking (i.e., data exploration and analysis) versus data communication (i.e., presentation).

Please give this some thought. If you think of potentially useful infovis research projects, respond with a description. If it has already been done, I’ll let you know, assuming that I’ve come across it. After a few days of collecting your ideas, I’ll compile a full list of potential research projects and publish it on this website. I will also keep it updated with new ideas and with information about research projects that are undertaken to address them.

Take care,

Signature

6 Comments on “Potential Information Visualization Research Projects”


By Shannon Burch. January 6th, 2016 at 9:14 am

Similar to your #1, I’d like to see the effect on all charts caused by varying aspect ratios, to scientifically determine the ideal. I have heard 1 to 1.3 is the ideal ratio, or to try to size them so lines have a 45 degree angle (this one I question as forcing lines to 45 degrees may make trends clear, but it can also exaggerate or flatten them) but I’d love to have some sort of evidence rather than saying that 1 to 1.3 is aesthetically pleasant or that’s just what you do.

In addition, I’d love to see a comparison of some of the alternates to bar and line charts tested versus a simple bar/line to see the difference in comprehension in a measurable manner. When the client wants a doughnut chart because it is “cool”, I go through a demo showing how it’s harder to compare sizes for pie slices versus a bar, and how a doughnut chart makes it even harder because you cannot see the angles. When they want to use the areas of circles to show differences in magnitude, I show them two circles and ask them to estimate the difference in area, then point out that they were off by 200%, and that they certainly want to know the difference between something 6 times bigger versus 12 times bigger. When they want to use an area chart for multiple trends over time, I point out that it is very easy to misread the movement of the top line of the area section as being a change, when it’s actually just the effect of the areas stacked below “pushing up” the other value. You get the picture. Part of what I do is education and I am happy to talk about data visualization options, but it would be even more persuasive to be able to say something like, “According to studies, 45% of viewers of doughnut charts were unable to accurately rank a set of 5 values, while all people who viewed the bar chart could.”

By Stephen Few. January 6th, 2016 at 10:14 am

Shannon,

Some research has been done to determine optimal aspect ratios for line graphs, beginning with the work of William Cleveland many years ago, who recommended “banking to 45 degrees,” which was refined by Jeff Heer and Maneesh Agrawala 10 years ago in the paper “Multi-Scale Banking to 45 Degrees.” I’m not aware offhand of similar research for anything other than line graphs, however.

I agree that more research should be done to more firmly establish the perceptual problems of area graphs. Pie charts have been the object of several studies, most of which were done many years ago. I summarize the findings of these studies in my article “Save the Pies for Dessert,” which many people like you use when they’re trying to make the case against pie charts. I believe that it’s time for some new studies to put these issues to bed.

By Shannon Burch. January 6th, 2016 at 11:31 am

Thanks for the pointer towards Cleveland and Heer/Agrawala’s work; I knew I had read some of that work but had forgotten where. And of course, I have read your article on pies! But I’d love to see that sort of research expanded.

I was thinking about how the study would be constructed; obviously, we want to avoid the things like “how long do people spend looking at the chart” or “do people remember the chart later” as an measure of the usefulness of the chart. We’d want to ask questions that are representative of real-life investigations you might be doing with charts: “Is Product A trending up or down? More or less than product B? Which had higher sales in 2010?” etc. Then, I would like to measure not only whether people get the correct answers, but how long it takes. A chart where people CAN get correct information, but it take a bit longer, is different from a chart where you literally cannot get the correct information at all. Maybe we could even determine if certain charts are more likely to be completely misleading—that they are more likely to have people determine the exact opposite of the truth!

I’d also like to have a monetary reward for the correct answers, to make it more likely to map to real world usage. When people are looking at charts, it’s typically because that information is vital to them–and not just 5 or 10, but tens of thousands, even millions of dollars on the line. I think having a monetary reward would both encourage serious thought and prevent a superficial judgement of the charts based on newness or aesthetics. They may like shiny 3D area charts at first (though I think people who USE the charts are always less thrilled by those sorts of things than the people who make the charts may think) …but NOT when they lost $5 because they couldn’t accurately tell trend movement!

By Enrico Bertini. January 11th, 2016 at 4:06 pm

Stephen, fantastic initiative! I see this as one of the best concrete ideas stemming from the whole debate on visualization research happened last month. Thanks a lot for starting this. I am looking forward to seeing this list. It would be great if other practitioners would like to do the same.

Shannon, a few papers exist that focus on area versus position/length and such. Here are a couple off the top of my head:

Javed, Waqas, Bryan McDonnel, and Niklas Elmqvist. “Graphical perception of multiple time series.” Visualization and Computer Graphics, IEEE Transactions on 16.6 (2010): 927-934.

Kong, Nicholas, Jeffrey Heer, and Maneesh Agrawala. “Perceptual guidelines for creating rectangular treemaps.” Visualization and Computer Graphics, IEEE Transactions on 16.6 (2010): 990-998.

By David Foster. January 13th, 2016 at 5:04 am

A relatively simple one that I have always wanted to know about is how many rows of tabular information can a person interpret quickly and how quickly they can interpret then same information in charts.

In businesses I regularly come across spreadsheets with 100’s (if not thousands) of rows of information that the business relies on.

It would be helpful to point to an academic study that shows how the speed of interpretation is effected by the number of rows and how much faster interpretation was with charts.

I was thinking about identifying trends and outliers in sales vs margin data as a simple example.

Basically I want a nice response to the ‘just show me the data’ behaviour pattern.

By Andy Cotgreave. January 14th, 2016 at 1:58 am

Hi Steve
Something I’m interested in is to what extent preconceptions influence interpretation of charts. I’ve been playing with Population Pyramids recently: they’re familiar and therefore hard to dislodge as a chart for displaying population. However, they’re really weak for several key purposes, such as comparing gender at any given age group.

I tweeted an example to different ways of showing them, and the replies from Adrian Lee were really interesting, as he described the preconception problem: despite known issues he prefers the orginal style: https://twitter.com/acotgreave/status/686591596343767040

This also relates to that thing about pie charts I contacted you about. Yes, pies are bad, but do the benefits of familiarity sometimes mean we’d be better using them for some audiences? If someone has a strong preference for or understanding of a chart type do they get better comprehension even if it’s not the ideal display type?

Leave a Reply