Two statements from a controversy on data visualisation: statisticians vs. visualisation specialists, statistical graphics vs. Information visualization (a.k.a infovis). A controversy? Not really!
The visualisation expert: ‘And yet, visualization is much, much more than what it appears to be at first glance. The real power of visualization goes beyond visual representation and basic perception. Real visualization means interaction, analysis, and a human in the loop who gains insight. Real visualization is a dynamic process, not a static image. Real visualization does not puzzle, it informs.’
Robert Kosara, UNC Charlotte, http: // eagereyes. org/
The statistician: ‘ … differences between statistical graphics and infovis. In statistical graphics we aim for transparency, to display the data points (or derived quantities such as parameter estimates and standard errors) as directly as possible without decoration or embellishment. ‘In a modern computing environment, a display such as Nightingale’s [infovis] could link to a more direct graphical presentation …., which in turn could link to a spreadsheet with the data. The statistical graphic serves as an intermediate step, allowing readers to visualize the patterns in the data.’
Andrew Gelman, Dep. of Statistics and Department of Political Science Columbia University, New York Antony Unwin, Department of Mathematics University of Augsburg
Read the two articles published in the joint newsletter of the Statistical Computing & Statistical Graphics Sections of the American Statistical Association, Volume 22.
‘This volume features two articles both looking at the aspects of “graphical displays of quantitative data”. In the first paper “Visualization: It’s More than Pictures!” by Robert Kosara, Robert sheds a light from the point of view of an InfoVis person, i.e. someone who primarily learned how to design tools and techniques for data visualization. With the second article “Visualization, Graphics, and Statistics” by Andrew Gelman and Antony Unwin, we get a similar view, but now from someone whose primary training is in math and/or statistics.’
In the introduction Jürgen Symanzik gives an excellent crash course in data visualization and its power:
‘It appears as if statistical graphics have helped to detect the unknown and unexpected — again! Most of us know the classical examples from the last 150 years where statistical graphics have helped to discover the previously unknown. This includes John Snow’s discovery that the 1854 cholera epidemic in London most likely was caused by a single water pump on Broad Street, a fact he observed after he had displayed the deaths arising from cholera on a map of London. A second, well–known example is Florence Nightingale’s polar area charts from 1857, the so–called Nightingale’s Rose (sometimes incorrectly called coxcombs), that demonstrated that the number of deaths from preventable diseases by far exceeded the number of deaths from wounds during the Crimean War. These figures convinced Queen Victoria to improve sanitary conditions in military hospitals. Many additional important scientific discoveries based on the proper visualization of statistical data could be mentioned, but the most important fact is: New discoveries based on the visualization of data can happen here and now!
This is a message we should carry to our collaborators, students, supervisors, etc.: Statistical graphics (or visual data mining, visual analytics, or any other name you like) typically do not provide a final answer. But, statistical graphics often help to detect the unexpected, formulate new hypotheses, or develop new models. Later on, additional experiments or ongoing data collection as well as more formal methods (and p–values if you really want) may be used to verify some of the original graphical findings.’
Jürgen Symanzik Utah State University