The Week in Vis 06 (Mon, Oct 8 – Fri, Oct 12): Evaluation

by Mike Gleicher on October 4, 2018

Class Meetings
  • Mon, Oct 8 – Lecture:Evaluation
  • Wed, Oct 10 – ICE: Arrival Distributions and DC2
  • Fri, Oct 12 – No Class
Week Deadlines

So last week, we took a detour to talk about how to make visualizations. This week, we’ll try to figure out how to determine if the visualizations we make are any good. The readings get at this from different sides.

I realize there’s a design challenge due at the end of the week. But these are particularly important readings. The concepts in #1 (Munzner) are really important (and can extend beyond visualization). #2 Cairo gets in a differently useful, and thought provoking way (and you’ll recognize his example). #3 Tufte points at some things that can go horribly wrong with a visualization. Chris North’s article (#4) is so short that he doesn’t actually tell you about his cool experiments – but really gets at why its so hard to assess whether tools are good.

Which leads to #5 which takes on a very different meaning given the recent news. I did not plan for the timing. I will change this paper from “required” to “strongly recommended.” I still really recommend reading it – it’s a great example of trying to get an experiment right (even if the original paper it is trying to replicate may not have).

You may want to look at this week’s learning goals Learning Goals 6: Week 6 – Evaluation.

Readings (due Mon, Oct 8 – preferably before class)

Evaluation is such a big and hard question. This will get at the key concepts.

    1. Analysis (Chapter 4 from Munzner’s Visualization Analysis & Design) (Munzner-04-Validation.pdf 0.5mb)
    2. The five qualities of great visualizations (Chapter 2 of The Truthful Art) (theTruthfulArtCh2.pdf 10.0mb)
    3. Graphical Integrity (Chapter 2 of Tufte’s The Visual Display of Quantitative Information) (1-VDQI-2-GraphicalIntegrity.pdf 62.2mb)
    4. Chris North, “Visualization Viewpoints: Toward Measuring Visualization Insight”, IEEE Computer Graphics & Applications, 26(3): 6-9, May/June 2006. pdf (doi; 4 pages)This is a good introduction to the challenges of visualization evaluation. And it’s short.

Optional

This one was supposed to be required, but I decided to reduce the amount of reading. It’s still strongly recommended.

  • Dragicevic, P., & Jansen, Y. (2018). “Blinded with Science or Informed by Charts? A Replication Study.” IEEE Transactions on Visualization and Computer Graphics, 24(1 (Proceedings InfoVis 2017)), 1–1. DOI PDFI want you to read an empirical paper. I pick this one because it takes quite a simple question and tries to be painstakingly thorough with it. Moreover, it is mainly trying to replicate an experiment that got a lot of press. While the authors didn’t set out to contradict the prior paper, it seems they got a different answer to the same question.

The “Chartjunk” paper would be required reading – except that we’ve already learned about it from Cairo, The Functional Art Chapter 3 (theFunctionalArtCh3.pdf 11.4mb). It’s worth looking at if you’re really interested in the topic. And the Few blog posting may be more valuable than the article itself

  • Bateman, S., Mandryk, R.L., Gutwin, C., Genest, A.M., McDine, D., Brooks, C. 2010. Useful Junk? The Effects of Visual Embellishment on Comprehension and Memorability of Charts. In ACM Conference on Human Factors in Computing Systems (CHI 2010), Atlanta, GA, USA. 2573-2582. Best paper award. project page w/pdf (doi). (10 pages)This is a pretty provacative paper. You can pick apart the details (and many have), but I think the main ideas are important. There is a ton written about this paper (those of the Tufte religon view this as blasphemy). Stephen Few has a very coherent discussion of it here. In some sense, I’d say it’s as useful than the original paper – but I would really suggest you look at the original first. While more level-headed than most, Few still has an Tufte-ist agenda. Reading the Few article is highly recommended – in some ways, its more interesting than the original.

Chapter 4 of Munzner is based on an earlier paper that was quite influential (at least to my thinking). It is somewhat redundant with what is in the chapter, but for completeness, you might want to see the original:

  • Munzner, T. (2009). A Nested Model for Visualization Design and Validation. IEEE Transactions on Visualization and Computer Graphics, 15(6), 921–928. (pdf) (doi)

In case you cannot get enough of Tufte, you can get his ideas on what is good (Ch5) and bad (Ch6).

If you’re wondering whether the deceptions Tufte mentions actually fool people, here’s an empirical study of it:

  • Pandey, A. V., Rall, K., Satterthwaite, M. L., Nov, O., & Bertini, E. (2015). How Deceptive are Deceptive Visualizations?: An Empirical Analysis of Common Distortion Techniques. In Proceedings of the 33rd Annual ACM Conference on Human Factors in Computing Systems – CHI ’15 (pp. 1469–1478). New York, New York, USA: ACM Press. (doi)

Some other stuff on evaluation:

  • Lam, H., Bertini, E., Isenberg, P., Plaisant, C., & Carpendale, S. (2011). Empirical Studies in Information Visualization: Seven Scenarios. IEEE Transactions on Visualization and Computer Graphics, 18(9), 1520–1536. http://doi.org/10.1109/TVCG.2011.279
  • Correll, M., Alexander, E., Albers Szafir, D., Sarikaya, A., Gleicher, M. (2014). Navigating Reductionism and Holism in Evaluation. In Proceedings of the Fifth Workshop on Beyond Time and Errors Novel Evaluation Methods for Visualization – BELIV ’14 (pp. 23–26). New York, New York, USA: ACM Press. (http://graphics.cs.wisc.edu/Papers/2014/CAASG14)What happens when I let my students rant.
  • Gleicher, M. (2012). Why ask why? In Proceedings of the 2012 BELIV Workshop on Beyond Time and Errors – Novel Evaluation Methods for Visualization – BELIV ’12 (pp. 1–3). New York, New York, USA: ACM Press. (link)Me ranting about how evaluation shouldn’t be an end unto itself. The workshop talk was much better than what I wrote.
  • You should read at least one of the papers by Michelle Borkin and colleagues on the memorability of visualization. These papers are very provocative, and provoked some people to be downright mean in attacking it. You don’t need to worry about the details – just try to get the essence. The project website has lots of good information.Michelle Borkin et. al. What Makes a Visualization Memorable? pdf InfoVis 2013 (10 pages).
    This is another radical thought of “maybe Tufte-ism isn’t all there is – and we can measure it.” Again, we can quibble with the details, but they really re getting at something real here.

    Michelle Borkin et. al. Beyond Memorability: Visualization Recognition and Recall. InfoVis 2015. (pdf); 10 pages

Previous post:

Next post: