In Look at Data: What Makes Bad Figures Bad before digging into the detail Healy introduces the reader to the Good Chart concept. Specifically Healy outlines that "a really good or really graph cannot be boiled down to a list of simple rules", this discussion is similar to that presented in Few's materials - specially that a graph should be examined across two axises - the goodness of the chart, and also its usefulness of the chart to the audience. Any judgement of the "correctness" of the graph cannot be undertaken without giving to consideration to its use and more specifically the audience who is going to be consuming the material (inclusive of their prior knowledge). This moving target for a good chart makes the desire for an algorithmic approach to creating a good chart unobtainable. Instead, all three readings point to a series of design heuristics that the chart designer should take into account in the development of the materials.
After providing a brief introduction to the need for data visualizations, Healy then delves into some of the common shortfalls of charts, namely: bad taste, bad data, and bad perception. In the next section of the article there is some discussion about the human eye and mind’s ability to perceive data correctly in visualizations. In the next section there is an interesting analysis of which types of charts have different error rates in interpretation.
Bergstrom and West, then take a slightly less technical approach to the issue of determining the goodness of a chart. In the initial paper there is an in depth discussion of the need for 0 axis in different chart types, where a particular emphasis is on which analysis types requires views on the absolute vs. relative values. In addition a preliminary discussion of the appropriateness of secondary, non-linear, and broken axis is presented - where the general message is caveat emptor, or put differently the designer needs to have a solid rationale for employing such tools. In the second article a more generalized discussion is undertaken with a particular emphasis on Tufte’s principle of proportional ink. Bergstrom and West present a slightly more approachable version of the principle and provide a solid link to the first article on 0 axis and how this conveys the principle of proportional ink. They then go onto expand the article and apply it to projects such as Gapminder, presenting the argument that not all well received materials follow this principle. In the final sections they then dig deeper into some other areas of “bad” charts, including donut bar charts, 3d charts, changing denominators, and limitations of perception.