dashboard / Other / Tableau

The role of visual perception in dataviz

2017-09-20

Tags: , ,

If you know me you know I am passionate about psychology, cognitive science, visual perception  and how to work our brain. Knowledge of cognitive processes and memory is needed to make efficient and fast decision-making reports, but it is worth knowing during our everyday life, how and by what the human mind can be influenced, because the world in which we live the deception plays a significant role.

Cole Nussbaumer Knaflic is one of the bests in this field whose works have taught me. Showing us how to communicate effectively with data focused on simplicity and ease of interpretation. I highly recommend her book – Storytelling with Data – to anyone, of any skill level, who needs to communicate something to someone using data.

Last week I just bumped into one of her posts on Twitter.

That was a dataviz challenge.

"The Gestalt Principles of Visual Perception help us to identify clutter and leverage how our audience sees."

She found this graph on the Economist, headlined, “Hurricanes in America have become less frequent.” She wanted to show the data and ask for another suitable headline.

You can read her original post here.

Let’s take a look at this stacked bar chart. It shows the hurricane’s frequency from 1951-2016. The headline says: Atlantic hurricanes making landfall in the United States, by category.

not considering visual perception makes charts confusing

What are the 5 categories?

The Saffir-Simpson Hurricane Wind Scale is a 1 to 5 rating based on a hurricane’s sustained wind speed. This scale estimates potential property damage. Hurricanes reaching Category 3 and higher are considered major hurricanes because of their potential for significant loss of life and damage. Category 1 and 2 storms are still dangerous, however, and require preventative measures. In the western North Pacific, the term “super typhoon” is used for tropical cyclones with sustained winds exceeding 150 mph.

If you are interested in getting more information about the hurricanes check this site: http://www.nhc.noaa.gov/aboutsshws.php

Bar charts are one of the most common visualization types. Then what is the problem with stacked bars? I think stacked bars are useful for the total and if you would like to highlight the most interesting category at the bottom of the bars to make it easier to see the trend. But what about the categories at thetop? Can you tell me preciously according to the above example which blue category is longer or shorter? Viewers can’t make accurate visual comparisons of lengths.

 

”Stacked vertical bar charts are meant to allow you to compare totals across categories and also see the subcomponent pieces within a given category. This can quickly become visually overwhelming, however,…it is hard to compare the components across the various categories once you get beyond the bottom series because you no longer have a consistent baseline to use to compare.”

                                                                                                                                                              ~ Cole Nussbaumer Knaflic

In a classic graphical/visual perception paper, Cleveland & McGill studied how different bar chart designs impact the accuracy with which viewers can complete simple perceptual tasks.

You can find here the whole study: http://www.math.pku.edu.cn/teachers/xirb/Courses/biostatistics/Biostatistics2016/GraphicalPerception_Jasa1984.pdf

Study on visual perception

Stacked bar charts can represent more than one piece of information. Cleveland and McGill termed it as a divided bar chart. A divided bar chart is an easy way to represent more than a piece of information on a bar chart, but there are consequences. Stacked bar charts pose a risk of misjudgment in the ordering of categories when evaluated with Cleveland’s 10 elementary perceptual tasks [Cleveland and Mcgill 1985].

If we return to our example: it is difficult to order the categories within a bar and difficult to compare the same categories of a hurricane  (1-5) on different bars from 1951-2016.

A useful alternative to bar charts is dot plots.

Dot plots were invented by William S. Cleveland to provide improved visualization of the kinds of data displayed using bar graphs. The dot plot may be thought of as an adaptation of the scatter plot for use when the vertical axis variable is categorical.

With the bar chart you can compare lengths as well as position, but if you are a fan of Edward Tufte (or you have a boss who is a fan of Bertin) then the dot plot has the best data-ink ratio representation.

Here is my makeover chart:

 

We see instantly and effortlessly the exact number of hurricanes by categories. And yes, the trend for all categories declined but this is not a good news for us because the number of most powerful 3 -, 4 – and 5- category – hurricanes are increased.

Cole also published an extended data file, that included monthes so I could just quickly see when were most hurricanes and I found this result.

Just with all data visualizations, practice cautiously to ensure you create a visual that is easy and quickly determine an accurate understanding.

@IvettAlexa

 

 

 

4 likes

Author

Ivett Kovács

I love taking datasets into beautiful and informative visualizations. Besides data, I am also very passionate about traveling, surfing, cooking and exploring the outdoors with my dog.

Your email address will not be published.