Read the two posted papers: Cleveland and McGill, Heer and Bostock. Write an essay that builds on the above two articles and explores the questions:
Your essay should include at least 2 additional studies on perception and visualization beyond the assignmed reading.
Cite all sources, no plagurism, and feel free to use images, or your own graphics to make support your thesis.
The dictionary defines perception as a way of regarding, understanding, or interpreting something; a mental impression. it can also be defined as an interaction between the human visual system and statistical graphics. Here in this paper, we discuss the role of perception and its importance based on studies done by various scientists.
Cleveland and McGill have described graphical perception as a process of visually decoding any encoding on graphs. They hypothesized a set of elementary perceptual tasks, these tasks are carried out when people are trying to extract quantitative information from visualisations. Cleveland and McGill performed experiments with elementary perceptual tasks, and attempted to show that if a graphical form involves elementary perceptual tasks, it leads to more accurate judgments and hence increases the chances of correctly percieving patterns. This theory was then utilised in analysis as well as redesign of various graph forms. For example, based on Cleveland and Mc Gill’s study, it is evident that the use dot charts and bar charts instead of divided bar charts and pie charts, use framed-rectangle charts instaed of statistical maps. These better charts lead to better judgments.
After Cleveland and Mc Gill many other researchers performed experiments to test different chart types. These studies attempted to measure how every individual encoding variable affects the accuracy as well as response time of estimating values of the data that is presented.
Heer and Bostock’s research focused on experimenting on crowdsourcing of graphical perception. Since crowdsourcing is low cost and is scalable, cit was seen as an attractive option to evaluate visualizations. Heer and Bostock experimented to find out how feasibile it is to use Amazons Mechanical Turk to evaluate visualizations and to use the methods to obtain better insights into visualization. The study shows that if we use qualification tasks and verifiable questions this helps to ensure high quality responses.
A study by Simkin and Hastie predicted how graph types and judgmental types interact and determine the speed as well as accuracy of quantitative information extraction by a human. Their research also points out that we not only need to consider the visual, but we also need to consider the human beings for whom the visuals are intended. The analysis consistently pointed to anchoring as one of the key processes for proportion and comparison judgments. The more accurate anchoring was possible with position and angle codes that accounted for their superiority over the length code. Overall, processing angles was deemed more difficult than processing linear aspects. While making comparison judgment, the position code was better than other two codes. Length has less accurate anchoring, angles provided least accuracy.
Study by I. Spence and S. Lewandowsky shows proportions and percentages. It summarizes that the comparisons among multiple segments take longer and seem to have lower accuracy. Pie charts were the worst except when multiple segments were compared. Tables were worse to everything except when we need to communicate absolute values.
Data visualization is a way to present both quantitative and qualitative information for better perception of the consumer of he data. Visualization is also very important to understand the results from data analysis. A study was done by Zubiaga and Namee, this used a crowdsourced platform for experimentation. Eventually, it was concluded that histograms were better perceived than density trees.
All these studies show how important graphical perception and how it is related to data visualization. These studies also show how it is important to have some guidelines related to what kind of visualisation to use for a certain type of data so that the consumer of that information is able to percieve the data with accuracy.
Following are the important points that we need to consider while thinking of graphical perception and its relevance:
Graphical representation includes slots for the conceptual message of the graph. Elementary code and judgment interact to determine performance. Elementary processes of anchoring, scanning, projection, and detection operators explains the interactions.
Scientists through the years have done variety of work on graphic perception and visualization. There are various factors such as cognitive abilities etc. that impact a human beings perception on a visualization.
[1] W. S. Cleveland and R. McGill. Graphical perception: Theory, experimentation, and application to the development of graphical methods. Journal of the American Statistical Association, 79(387):531-554, September 1984.
[2] J. Heer and M. Bostock. Crowdsourcing graphical perception: Using mechanical turk to assess visualization design. In ACM Human Factors in Computing Systems (CHI), pages 203-212.
[3] D. Simkin and R. Hastie. An information-processing analysis of graph perception. Journal of American Statistical Association, 82(398):454-465, 1987.
[4] L. Tremmel. The visual separability of plotting symbols in scatterplots. Journal of Computational and Graphical Statistics, 4(2):101-112, 1995.
[5] Lewandowksy S, Spence I (1989) The perception of statistical graphs. Sociol Meth Res 18(2&3):200-242.
[6] Zubiaga, Arkaitz and MacNamee, Brian. (2016) Graphical perception of value distributions : an evaluation of non-expert viewers??? data literacy. Journal of Community Informatics, 12 (3). ISSN 1712-4441