R1

Poor Form

Read Healy's introductory chapter from Data Visualization for Social Science:

Look at Data: What Makes Bad Figures Bad

Use the tag “R1” when you post your assessment of the reading and the questions raised.

Data Visualization: A Practical Introduction by Kierann Healy

I think the researches and graphic examples that were provided in the writing aims to identify factors that determine how people interact with visualized data—by investigating the working mechanism of  human perception and its tendency, Kierann Healy discusses how graphic representations could be easily misleading and introduces a wide range of variables that are in play as data is mapped and interpreted on a visual platform.

Healy states the criteria in analyzing a bad figure are matters of poor aesthetic, substantive, and perceptual approaches that were conducted to map the data. A set of inconsistent design choices or over embellished graphic representations could be problematic aesthetic decisions that unnecessarily draw the viewer's attention and make them lose focus. Failing to present a good, unbiased set of data hinders the viewers to draw a substantive understanding from a figure. And visualizing without considering how people will perceive the outcome deeply undermines the importance of communicating the data.

However, I think it's important to note that Healy mentions that even if a figure is considered to have these bad practices, it may be successful in making people interested in looking at it or remember it over its simplified, plain alternative. Diverging from assessing the functionality and purpose of graphic representation as measures that judge the effectiveness of a graph to the value judgement of each individual, a question come to mind: What is really necessary to consider when mapping data, or in other words, what drives the process of data visualization? the data itself or the purpose of presenting that data and its audience?

In a way that Healy puts it, data visualization is encoding information that people needs to decode, understanding and then interpreting it. While I think that it is a safe choice to work within the frameworks of perceptive rules for graphics, I think it's also necessary to keep in mind to not treat them as absolute, divine methodologies. In the writing, Healy mentions the effects of optical illusions and how our "perception is not a simple matter of direct visual inputs producing straightforward mental representations of their content." And I think that this understanding of everything we experience visually are not absolute or objective, but respectfully more subjective and relative applies to the approach we should take when determining something to be a good or bad example of data visualization. How can we decide whether we are designing good or bad visual representations of data unless we know why and to whom that data matters? I'm curious to hear what you guys think that makes a bad graph.

The “Look at data” section was informational in explaining the effectiveness of data visualization. The reading explained the importance of the different styles of charts and how each type of visualization has its strengths and weaknesses. It is important to think about the user’s accessibility and experience when designing visuals for data and the numerous variables that impact these decisions.

A few interesting points that stood out were the sections on the axis and color value. “Our ability to scan the “away” dimension of depth (along the z-axis) is weaker than our ability to scan the x and y axes.” For this reason, it is difficult to incorporate 3D elements into our graphs and the angle of the chart (which is used to show the perspective of 3D) can be misleading in delivering the data. Another interesting point was how people perceive brightness. The same shade of gray is perceived very differently depending on whether it is against a darker background or a lighter one. We are also better at distinguishing darker shades than lighter ones. The interaction also differs based on the scenario: “We will do better at distinguishing very light shades of gray when they are set against a light background. When set against a dark background, differences in the middle-range of the light-to-dark spectrum are easier to distinguish.”

Overall, it is important to think about the conciseness of the visuals when creating a chart. The type of chart must be optimized to the data being delivered and the visuals must be simple enough for the user to easily understand (junk-free), while also delivering honest information.

Healy’s chapter on data visualization provided a thorough overview of the different factors when it comes to designing for data. Factors that range from audience perception, aesthetics, structure, accessibility and more, Healy emphasizes the importance of balance and planning when creating visual datasets. The chapter also discusses the power of data visualization and the influence it holds amongst various audiences. There is the idea that “even tasteful, well-constructed graphics can mislead us” (Healy, 1). I agree and appreciate how he mentions that a design’s visual appeal does not correlate to its accuracy. This section piqued my interest and led me to wonder about ‘nice’ designs that represent false or skewed data. My assumption of misleading visualizations are associated with bold simple bar graphs from extreme left or right wing political news outlets, which I don’t believe to be ‘nice’ or creative solutions.
The chapter also frequently mentions the concept of human cognition and perception when it comes to comprehending data and design. It accentuates the idea that in order to successfully design, one must understand their audience. Data visualization can be quite an anthropological topic, one may take into account their viewers’ environment, culture, background, emotions, etc. In addition to knowing one’s audience, and in order to achieve ‘graphical excellence’ (I enjoyed a lot of his word choice), designers must wrestle with the challenge of providing the viewer with “the greatest number of ideas in the shortest time with the least ink in the smallest space” (Tufte, 1983, p. 51). Quite the task to accomplish! I am curious to hear what my peers believe is an efficient use of data to ink ratio and how their creative process changes when creating while using a given dataset. Healy also provides the example of  Poisson and Matern’s work, and I am interested to hear what my classmates interpreted as the more structured design. The author mentions “if you ask people which of these panels has more structure in it, they will tend to say the Poisson field. We associate randomness with a relatively even distribution across a space” (Healy, fig. 1.20). My initial reaction was to view the Matern piece with more structure. Then again, it goes to show that data visualization heavily relies on the viewer’s perception.

My key takeaways from this reading are that there are general guidelines and protocols, but no hard or fast rules. All of these best practices exist on a spectrum.

For example, chart-junk can be a distraction and pull away from the importance or insights of the data. Sometimes this can be intentional or unintentional. However, there are upsides to a certain amount of embellishment or artistic flair added to a visualization. In some cases, this ‘chart junk’ can improve recall or memorability, maybe even keep a viewer’s attention for a longer period of time.

Things like Gestalt principles can be used for both positive or negative ends, depending on how and where they’re used. Data can be manipulated (or rather, the visualization of the data) to make a graphic more persuasive or highlight key points. However, these manipulations can also tread into the realm of falsifying or misrepresenting information.

While reading this, sometimes I wondered: if given all of these potential pitfalls, why not just represent numbers as they are in their rawest, numerical format? Of course, the answer is easy: because it’s more challenging to read, comprehend, and see the key findings of the numbers in such a format. So even the data itself exists on this spectrum. Too much design can convolute the meaning of the information, and too little design makes the data more difficult to understand to its fullest extent.

I’m inclined to take a somewhat nihilistic approach as a response to this reading: it’s all nonsense, nothing is clear/honest/true. Everyone has an agenda so why interpret any of it because there’s an ulterior motive behind all of it. Of course, that’s not the point of this reading or this course as a whole. It’s better to understand these guidelines and spectrums in order to be a more judicious designer when making these choices for how best to represent information accurately, honestly, and (occasionally) with integrity.