I would expand the defninition beyond strictly quantitative information. Consider the list of names you mentioned. Coloring them according to discrete, non-quantitative categories would be using visual attributes to convey some useful information. Drawing lines between names such as an org chart, genalogy, or other node-link diagram is also a visual display of information that isn’t built on quantitative data (although quantities may be inferred by the viewer).

I would normally be inclined to defend my views on this in detail – especially having clearly caused such ‘A Cauldron of Terminological Confusion’ – but I’m unwilling to spend my vacation time engaging in semantic debates. All I can say is that it works FOR ME – it helps me organise my thinking about this subject – and that’s surely what a personal definition should accomplish. Folks can make their own mind up about the legitimacy of this definition by freely accessing the full opening chapter – it will add more context and flesh out the discussion from my perspective far better than writing an essay in this little comment rectangle https://uk.sagepub.com/en-gb/eur/data-visualisation/book244865#resources.
Thanks.

I wonder to what extent it matters. How much does it matter whether I view data as solely quantitative or if I include qualitative data when I’m visualizing it? Who is really struggling with the definition? Are masses utterly confused because one defines data as quantitative while another includes qualitative in their definition? What is the impact of differing definitions? I suppose I look at data visualization as a way to communicate, so through that lens, I think the distinction may be less of an issue.

I’m with Emily. Why do we need to stick a name on everything? Frankly, I could do with fewer words and more meaning. If something is purposeful, I don’t give a hang what it’s called.

Kim,

I like what you said so much, I’d like to expand on it. “Fewer words and more meaning; fewer meanings and more meaning that is shared.”

Jim,

Good question. You might think that we could drop the word visual because display always means visual, but it actually doesn’t. By “visual display” in the context of data visualization, I mean that the data is primarily presented graphically, as opposed to textually (i.e., words and numbers). Text in always included in a data visualization, but the primary encoding of values is visual in nature. For example, even though I talk a great deal about the value of tables as a means of quantitative display, I don’t consider tables a form of data visualization unless something visual is included to encode meaning.

Given the fact that this blog post is about confusing terminology, I’ll point out that I consider a table a type of chart. Many (perhaps most) people use the terms chart and graph interchangeably. I don’t. I use chart as an umbrella term for several forms of data display, including tables, graphs, diagrams, and maps. Based on this definition, not all charts are data visualizations.

Jorge,

I’m not aware of a power struggle going on here and I’m not advocating a one-size-fits-all definition of data visualization. Your definition of data visualization as “a visual translation of an underlying table” is a bit confusing and is definitely incomplete. You might be right, however, that this definition could work for someone who thinks of data strictly as Excel tables.

Jonathon,

Give me an example of something that you would call a data visualization that doesn’t include quantitative information.

Steve,

One example would be relational charts such as database diagram or organizational chart. Each of these present data in a visual way to enable us to quickly understand the relationships between objects.

Another example would be a workflow diagram. These present data related to a series of events in a visual way as opposed to a strictly textual way.

I’m sure there are other, perhaps even better, examples, but those quickly came to mind.

To me, it’s not about quantitative data vs qualitative data. It is more in how the data is communicated (i.e. visually/graphically or textually).

Jonathon,

Thanks for the examples of visual data displays that do not include quantitative information. Org charts, entity-relationship diagrams, an instructional illustration in a car repair manual, and even the map-based example that Jeff provided, are all examples of of visual data displays that have not traditionally been called data visualizations. It is absolutely true that it would make sense to call them data visualizations were it not for the fact that this term was coined to describe displays that include quantitative information. Similarly, we could argue that the term “information visualization,” which is favored by the academic research community, could be used to describe all visual data displays, but it is not. For instance, displays of things that are physical in nature, such as an MRI scan of someone’s brain, qualifies by definition as a “scientific visualization” rather than an “information visualization.” In all of these cases terms were coined to differentiate some forms of visual display from others because these distinctions were useful. If we could start over and coin the terms anew, I would prefer to use the term data visualization for visual displays of data of all types and to use the term quantitative data visualization for displays that include quantitative information. We cannot start over, however. Changing the meaning of terms once they’re established creates confusion. Much of the confusion in terminology today has been created by people who have become involved in the field without knowledge of its past.

Jeff,

You covered a lot of ground in your comments and I’ll try to address all of your points.

First, my definition of a dashboard. I proposed this definition initially in an article that appeared in Intelligent Enterprise magazine. I did so to cut through an enormous amount of confusion that resulted from the fact that the term dashboard was being used to describe everything imaginable. In the definition I stated that a dashboard displays information on a single screen. Why? As you know, I take quite a bit of time in Information Dashboard Design to explain this. What fundamentally differentiates a dashboard from other forms of display is the fact that it is used for rapid monitoring to maintain awareness of what’s going on. If the primary information that needs to be rapidly monitored is spread across multiple screens, it cannot be rapidly monitored, nor can comparisons and connections be made when all of the elements are not simultaneously visible. Can this goal always be met? No, but to support rapid monitoring, we should always strive for it. My statement that a dashboard should provide the means to additional, supporting information, when it is needed, does not contradict the definition. When the dashboard makes someone aware of something that needs a response, it is often the case that additional information that does not appear on the dashboard itself is needed to determine how to respond. Providing a way to easily navigate from the dashboard to that additional information is part of what makes a dashboard work. That additional information can be provided in various ways, such as pop-up windows and reports that can be separately accessed. A report of details on a separate screen that complements the dashboard is not a dashboard itself. Because it is important to provide easy and instant navigation to additional supporting information, I do not consider a printed information display a dashboard in the full sense, for it does not provided this one key bit of interactivity that a dashboard should provide. Regarding the iPhone-based example of a dashboard that appears on the cover of my book, yes, I consider an iPhone a computer. Don’t you?

When I say that a data visualization is a visual display of quantitative information, I do not mean to suggest that it consists of quantitative information alone. A quantitative display is only meaningful if it also includes categorical information to tell us what the quantities represent. I understand quantitative information just as statisticians do. It consists of measures. A bunch of measures on a screen without labels to say what they represent, however, is meaningless. In Show Me the Numbers, I do not define quantitative information as either quantitative or categorical. I define it as involving both.

In my book Now You See It, as you point out I have indeed contributed to the confusion by using data visualization as an umbrella term that includes both information visualization and scientific visualization. Apparently, at the time, I was trying to differentiate these related terms in a way that differs from my current thinking. It has been many years since I’ve read the particular excerpt that you mentioned. So, I readily admit that my own thinking on these matters is not written in stone and shares in contributing to the confusion.

Regarding the Cleveland Museum of Art’s online collection, I do not consider this a data visualization. It is a collection of photos. The fact that you call this a data visualization nicely illustrates the overly broad use of the term that I’m trying to avoid. At this moment, I am sitting in my office. If I look out the window, I see a collection of visual objects (trees, houses, cars, etc.). Is that a data visualization? It is visual and it provides information. We use definitions to draw the lines in meaningful and useful ways. As a data visualization practitioner, I do not consider photography a necessary part of my skillset. In the separate comment that you posted a moment ago, you asked “What confusion would be generated by broadening the definition?” You have illustrated the confusion that concerns me by citing the museums photos of its exhibits as a data visualization. Terms become useless if they don’t have particular meanings.

Jeffrey:

“And I also disagree that it has to be on one screen. You mention that if the data is on multiple screens then it becomes multiple dashboards, yet on page 36 you write, “Allowing navigation to further detail or to a different set of information that best serves its purpose by standing alone can be a powerful dashboard component.” This is contradictory.”

No, this is not contradictory at all.

The dashboard is a single screen, allowing the user to monitor all of the essential information at a glance.

However, when monitoring high level information, situations arise when you then need to take action and/or further investigate the information, triggered by some change or status discovered on the dashboard.

At that point it is obviously desired to be able to begin that deeper investigation by clicking through from the alert or status that requires investigation, and view additional data.

This does not make the “additional data” a part of the dashboard, and does in any possible way contradict the idea that the dashboard itself is intended to be a single, at a glance, view of the high level essential information.

Also:

“Putting all of your definitions together, it seems you are saying “data visualization is visually representing quantitative information graphically.” And you are including “categorical” data in your definition of “quantitative information”.”

Categorical data is not exclusive from quantitative data. As soon as you count items by category, you have quantitative data (yes, obviously categorical data can be non-quantitative as well. point is that they are not mutually exclusive, and as such, not at odds)

Based on my reading of the chapter, the example used in “Data Visualization for Human Perception” did not include any discernible data that was quantitative, but rather was noted that the example could be further enriched by including quantitative information.

With all sincerity, I simply find your previous explanation (in addition to the alternative I proposed) to be more in line with my own thinking. I find both to be perfectly acceptable working definitions.

As you have clearly stated your position on this topic, I will have to respectfully disagree that including the word quantitative is necessary.

Andrew,

I think your examples outline the difference between defining a term of use in our field and casual speak. I would not CALL a map in my a car data visualization, but I do believe it is a type a data visualization. I also don’t tell my kids that “I’m going to the sport utility vehicle”, I simply say, “I’m going to the car”, even though “car” is not an accurate term in this case.

Steve was talking about the definitions in the context of creating clarity, not what average Joe business user calls them. That person will never use the term data visualization. They will simply say, “send me that…” [report, dashboard, graph, map, org chart], which by my definition, are all data visualizations.

Yes, Emily and Kim stated that view as well. “Why worry about a label on everything?” That’s something to consider in this discussion. Is it your view point then that a definition is not required? It seems the question you and others are really asking is “What is the benefit of defining data visualization?”

@Jeffrey: “Is it your view point then that a definition is not required?”

No, my concern is with over-generalizing the definition. I feel that doing so – in this instance – undermines the point of having a term or definition at all.

“It seems the question you and others are really asking is ‘What is the benefit of defining data visualization?'”

On the contrary, I am saying there is certainly a benefit in defining it. I am, however, doubtful that there is any benefit in opening that definition to allow every conceivable exception people can dream up.

Definitions can be flexible, with a little common sense. For example, when Stephen defines the term “dashboard” with the words “single screen”, he doesn’t say how big of a screen. Would two screens side-by-side not be a dashboard whereas a single screen that is twice as wide is still acceptable? And would it still be a dashboard if printed on paper, as you suggest? If so, how big is the paper allowed to be?

These counterexamples and exception cases simply don’t need to be accounted for in the definition, and adapting the definition to include all of them makes the term less useful.

Some are probably tired reading this thread and others will view the entire exchange as pedantic, but I do think it’s important that these things are discussed. This is actually a common exercise in other fields. In music school, professors ask “what is music?” and in the field of art there have been legendary battles over “what is art?” (ex. the Mapplethorpe exhibit here in Cincinnati 25 years ago – not that I expect a lawsuit over the term data visualization).

I am also not advocating for a wild west definition. This is really about where you draw the line in the sand. I do not think that Steve looking out his office window or a common photograph of something is a data visualization. Nicholas Felton’s new book is “Photoviz – Visualizing Information Through Photography”) so maybe that’s a good term for photography visualizations. That said, I clearly have a broader definition of data visualization than most on this thread. I think the Cleveland Museum of Art is a form of data visualization. Their data is art. The piece of art itself is not data visualization (unless made to be), but the interface allowing me to learn about them, sort them, filter them and drill down for more details makes it a data visualization to me. I can certainly understand people not wanting to draw the line that far away from graphs.

Moving closer in, a map showing flight patterns, by my definition, is most certainly data visualization, even if it’s only showing connections from city to city without any measures.

Personally, I like Steve’s previous definition from Now You See it (page 12). Data visualization as an umbrella term, “covering all types of visual representations”. “Whatever the representation, as long as it’s visual, and whatever it represents, as long as it’s information”.

Unfortunately, if a group of people that actually care about this topic enough to be on Steve’s website can’t come to common agreement on this term then it’s unlikely the broader data visualization community will.

@Jonathan:

For what it’s worth, I don’t think it’s your aim to over-generalize the term. I agree that “quantitative” is limiting. There are types of displays that are primarily or solely qualitative, too many to just consider them exceptions. In that case, I agree the definition could be refined.

@Jeffrey:

Let me put it another way. You say, “I think the Cleveland Museum of Art is a form of data visualization.” Regardless of your reasons for thinking that, *why* does it need to be classified this way? *Why* does the definition of data visualization need to include it?

Furthermore, do you disagree that something can be called a data visualization if the definition doesn’t perfectly account for it? Do you disagree that definitions can allow reasonable exceptions?