-
-
Marketplace
-
Channel Resources
Articles from this Site
-
Lost in Translation
Not long ago, I sat in the office of the Director of the Census Bureau of a country in South America. (As I fiercely protect the identity of all my clients, I have made a few adjustments to this true story to mask the identity of this country.) I had been working on issues in the understanding of data across international boundaries, and at this particular meeting I had a small delegation from the U.S. State Department along with me, as they were interested in meeting the Director and staff of this Census Bureau. The Director had a graduate degree from a university in the United States, so we had no problem conversing in English. As our conversation moved into the use of data visualization, the Director opened a PowerPoint presentation he recently made at a conference on South American demographics and econometrics, which had been held in Brazil.A Common Language -- with Words
Since the language of Brazil is Portuguese and the language of all attendees coming from other countries in South America was Spanish, I learned that the conference organizers had decided to have a common language for the text on all presentation slides. They all used English. For me, this was a wonderful decision as my English is much better than my Spanish, or Portuguese. I thus had no problem understanding the text captions and legends I was being shown and I could completely focus on the visualizations. Now, if only we had 'a common language' for visualizations.
An Un-Common Language -- with Visualizations
The Director was very proud of one slide in particular. Figure 1 is a facsimile of a part of that image.
Figure 1
I looked at the image for a few moments and asked tentatively "What are you trying to show here?" The Director eagerly described this as presenting the percent of the GDP (Gross Domestic Product) generated by four major business sectors for each of those countries. Furthermore, the size of the donut hole is proportional to the Per Capita GDP.
I was surprised at the choice of visual metaphor. Yet, especially in the presence of the delegation from the U.S. State Department, I certainly did not want to create an international incident by seeming to challenge the choice. I thus said delicately, "There are other ways you could have presented this data. Why did you select this format?" The answer was short and clear, and is a profound commentary on the way data visualizations are prepared and presented around the world today.
"I had discovered this format in Excel just a few days earlier, and I liked it. It seemed to be a good format to use to present the data in a way that would hold the audience's attention. It did hold their attention. I noticed them looking at it with great intensity."
The Wrong Common Denominator in this Visualization Language
Just because a visual metaphor resides in the 'Chart Wizard' of a software product, it is not necessarily appropriate to use in any presentation. Today we live with artifacts left behind by a war ... a business war of the late 1980s and early 1990s. In those heady days, three major spreadsheet products were fighting for market share and, as each tradeshow or trade magazine had a 'Spreadsheet Shootout,' each product wanted to have a longer 'feature list.' Many of the visual metaphor options we see today appeared in that context -- not because they were well-researched, or were useful to effectively communicate in a common language to all who viewed them, but because they were another line item on a 'feature list.' Today, the common denominator of many of these visual metaphors is that they appear on desktops across the world as an option in a Chart Wizard. There certainly is no common denominator of understanding.
Is the fact that the presenter 'likes it,' the best selection criteria? Though this individual had a graduate degree, he had never been able to create this format before discovering it in the Chart Wizard. Very likely there would be many in the audience who had rarely, if ever, seen it either. As a writer, I occasionally use an on-line thesaurus where I occasionally discover new words. Yet, would I select a new word because I liked it? It is much more likely that I would carefully consider how well this word would fit in to 'a common language' of textual communication between me and my reader. There should be the same kind of concern for building 'a common language' of visual communication between presenters and their audience. A first step in responding to this concern is establishing a common denominator of understanding among the visual metaphors used.
One can understand why the audience looked at this image with "great intensity." They could not understand the data, and they were trying desperately to see the relationships that the presenter was attempting to describe. In the process, they may very well have missed much of the presenter's words.
A More-Common Language -- with Visualizations
The concepts behind the acronym NOIR offer a foundation for building a common language of visual communication. I have created this acronym, and applied these concepts to data visualization, because they help us to better understand the scales of measure we use for every dimension of data we attempt to visualize. I presented these concepts in my BI Review column of December 2006, http://www.bireview.com/article.cfm?articleId=277 The Big Picture -- A CEO Allegory, and I elaborated on them in March 2007, http://www.bireview.com/article.cfm?articleId=310 A NOIR walk across Manhattan. This is the foundation of the discipline known as 'graphicacy.'
Figure 1 presented us with five dimensions: two nominal dimensions (Country and Business Sector), and three ratio dimensions (GDP, Per Capita GDP and Percent of GDP). Figure 2 displays another way to present these five dimensions. By putting the Country dimension on the vertical y-axis, we can write the nominal names horizontally (perpendicular to the axis) for easy reading of the text. Each of the three different ratio dimensions can then be placed on the horizontal x-axis, to the left and right of a central y-axis. The second nominal dimension, Business Sector, is represented by a color code and a text legend.
In Figure 2, each of the three ratio dimensions is presented with a visual metaphor that preserves the ratio nature of the dimension; while in Figure 1, two of these dimensions had lost the richness of their ratio scale. Per Capita GDP, in Figure 1, was presented with the size of the donut hole. The viewer has no idea whether Per Capita GDP is proportional to the area or diameter of the hole. (In fact, it is neither.) That visual metaphor reduced Per Capita GDP to an ordinal dimension so we can only tell which country had a larger Per Capita GDP than another, but we can not perceive any true proportion. In Figure 1, GDP is only presented in text, with no visual metaphor at all. Both of these problems are corrected in Figure 2.
Each horizontal line in Figure 2 represents the data for a particular country. The countries are sorted the same as in Figure 1, by the value of their GDP, but now this is visually evident from the bar chart on the left side of the figure. Figure 2 now presents a much more common language of visual communication. Furthermore, it presents more visual information, yet uses less screen real-estate. Relationships between the dimensions are more readily apparent ... in a way that was nearly impossible to discern in Figure 1.
Of the countries with the top three GDPs, notice how the Agriculture sector (green) is a shrinking fraction of their GDP. All of them have major seaports, and much of that high-efficiency agricultural production is shipped overseas. Uruguay makes up for its smaller land area by having its Industry & Commerce sector (salmon) and Services sector (yellow) being the largest percent GDP among these three countries.
The countries with the two smallest GDPs, Bolivia and Paraguay, are both landlocked. With geography as a constraint on developing their Agricultural sector and its export potential, their agricultural production is focused on small local markets. Yet, Agriculture is a major fraction of the GDP, and the Services sector is a concomitantly minor fraction. Looking at Figure 2, we visually see that for countries with a relatively small GDP (at left), the balance between the Agriculture and Services sectors may have a close relationship with having a very low Per Capita GDP (as seen at the right). There is much to international econometric modeling, but this image clearly presents interesting interactions among five key factors.
Visualizations: Lost, or Found, in Translation?
For a word processor or spreadsheet to type English, Spanish, Portuguese, or any other language, we consciously set the correct spell checker and keyboard. Once the words are typed, there is a clear process to translate this text in to another language. Yet, in each different language version of our spreadsheet, we find the same 'Chart Wizard.' We are lulled in to thinking this is an international language.
Since we assume this is a universal language, we rarely stop to check or make sure everyone understands the same thing in the images being presented. Thus, in any organization, there may very well be a visual language barrier in the same conference room where all participants converse in, and read, the same linguistic language. We need to improve our graphicacy skills in the use of data visualization so that we do not have to communicate across this visual language barrier.
Howard A. Spielman, M.B.A., Ph.D., President of Management Semiotics International Inc., can be reached at HASpielman@ManagementSemiotics.com.
For more information on related topics, visit the following channels:
