Perception Essay Assignment
ANLY 512 - Data Visualization
Himanshu Soni
5/12/2018
Analysis
“A picture is worth a thousand words”, theories of effective data visualizations aim to analyze and explain the relation between data and their perceptually motivated visualizations. We know the main purpose of data visualizations is to assist in making good decisions by drawing insights from data. There are three key points to be kept in mind when discussing visual perception; first, it is selective in the sense that the brain pays attention to only certain things and avoids getting overwhelmed; second, users are drawn to similar or recognizable patterns; and lastly, the brain can only register so much when observing a visual. The effectiveness of the visualizations, hence, depends on several factors such as preferences, cultural setting, and structural correspondence with the data it represents. Structural correspondence is given higher importance and is considered a necessary condition over other parameters. The process of data visualizations should therefore focus on and be measured in terms of the easiness and comprehensiveness.
Jeffrey and Bostock [1] present interesting findings on the viability of crowdsourcing graphical perception experiments using on Mechanical Turk. They essentially perform three experiments on proportional and rectangular judgements, gridline and contrast, and lastly on spacing and chart size. In their first experiment, participants tended to make more errors while observing circular area over rectangular area. The second experiment showed subjects selected lighter alphas on greater display color resolution. The last study focused on spacing, it showed accuracy plateaued as chart height increases.
Mehdi Dastani [2] emphasizes on analyzing the relationship between data and their effective visualizations, wherein a process model for effective data visualization is introduced. This process starts with an input data which consists of data elements. The second step is to determine elements representing elements in a way that structure of the visual elements perceptually represents the data structure. The data structure is however abstract at this point, the last step in data visualization is the layout process which transforms the abstract perceptual structure to a visualization that provides insights.
Process Model