Storytelling is a term that’s becoming popular in data visualization and data science in general, but there's no agreement on what it actually means: it can have both positive and negative connotations. This mini-brain-dump has a few perspectives to be aware of when we use the term, especially if the context is not clear.
I’m aware of the following meanings of storytelling:
- As a way of communicating a message: explaining an outcome
- As a way of communicating a structured narrative: understanding the journey
- As a logical fallacy: generalizing a rationale from insufficient information
- As deception: making the story different from reality
With these very different meanings, it’s impossible to make general statements about the value of storytelling. I usually avoid the term storytelling because of the possible negative connotations, but there’s nothing inherently wrong with it.
I often say data visualization is another medium of communication, and I see storytelling (except for the fallacy definition) as a technique for communicating in any medium. Storytelling is not the only technique and not even a prominent one for data visualization. More often, a data visualization is part of some story.
Random tidbits from around the industry
Within the data world, I first heard about storytelling from Tableau researcher Robert Kosara at a conference in 2012. He was positioning stories as the next big thing for dataviz, and Tableau made it part of their product and their marketing, even sponsoring a conference about it called Tapestries. So to some (lessening) degree, data storytelling has been associated with Tableau.
Cole Knaflic has a great book called Storytelling with Data and has built a training brand around that phrase. I think she uses the term in the messaging way, which is that storytelling is a way of getting a point across. Many people already know adding a story is a good technique for communicating, so transferring those techniques to data is part of the appeal of her brand.
I notice Microsoft PowerBI has a Data Stories Gallery.
There’s a nice dataviz podcast called Data Stories.
In his Truthful Art book, Alberto Cairo lists storytelling as one of his three mindbugs. There he means it in the logical fallacy sense, where we are inclined to invent explanations to match what we see. More recently, he tweeted a call for a ban of the word storytelling.
An argument based on evidence and careful reasoning is not a "story".
More often than not, a #dataviz or infographic isn't a "story".
"Storytelling" puts you in a frame of mind whereby you feel compelled to fit information into a predetermined mold.
Ban "storytelling".
Presumably he’s reacting to the overuse of the term. And partly he’s just trying to start a conversation with an extreme position. But there is some truth to the harm of using the term out of context.
This week, there was a Buzzfeed article about shoddy data handling by scientists, which included this use of the bad sense of the word:
Based on the emails, Nosek said, “this is not science, it is storytelling.”
This is at least evidence that in some fields, storytelling still has strong negative connotations.
JMP speaker and dataviz statistician Kaiser Fung has long used the phrase “story time” (2010) for the fallacy sense, naively manufacturing an explanation from data. As Columbia statistician Andrew Gelman described it, “that pivot from the quantitative finding to the speculative explanation.” Kaiser has also been doing JMP Explorers events around the notion of storytelling in the communication sense.
Last year, dataviz educator Jon Schwabish wrote a series of blog posts on the meaning of “story” with regards to data visualization. He argues that our use of “story” should be restricted to the structured narrative sense as used by authors and screenwriters, requiring emotion and a climax to the narrative. Data and visuals can be part of a story, but a visual or even a sequence of visuals in not a story in that sense.
Dataviz author Stephen Few has used story has a communication metaphor, but last year expressed concern over the term's overuse and "regret" over own his past use, and added:
I’m not diminishing the usefulness of data storytelling, which can be incredibly powerful when appropriate and done well. I’m merely pointing out that data storytelling is not some new endeavor or skillset that dominates data visualization. It is a minor—but nonetheless important and useful—aspect of data visualization.
Closing Thoughts
This post is more about the word storytelling rather than any of the specific meanings of the word. We don't need to ban storytelling at JMP, but we should be aware of the different meanings, not use the term out of context or as a panacea like some people use big data or machine learning. Consider using a more specific term if possible.
You all in the field know much more about communicating than I do, and I hope to learn more from your perspectives if you have time to chime in.
