Some consider Nigel Holmes, whose work tends to be more illustrative, the opposite of Edward Tufte, who preaches the data ink ratio. Column Five Media asked Holmes about how he works and what got him interested in the genre.
As a young child in England, I loved the weekly comics “The Beano” and “The Dandy.” They were not like American comic books; they were never called “books,” for a start. These English comics from the late 1940s and early ’50s had recurring one-page (usually funny) stories featuring a cast of regular characters. They had names like Biffo the Bear, Lord Snooty, and Desperate Dan. The comics were printed on poor-quality newsprint, which seemed to go yellow as you were reading it, but there was something very attractive about them.
I like the small dig on Tufte around the middle, while citing the paper that happens to find that Holmes’ graphics were more memorable than basic charts.
My own work at first was a little too illustrative, and Edward Tufte made a big fuss about what he thought was the trivialization of data. Recent academic studies have proved many of his theses wrong.
It seems the arguments haven’t changed much over the decades.
“Recent academic studies have proved many of his theses wrong.”
I don’t think this is true…?
A study found that charts with a bunch of crap on them were more memorable.
Not that they presented data better, or that the data presented in the charts were more memorable, if I recall correctly.
Not to mention that the ‘basic’ charts used in the study as a comparison were very poor examples that do not follow much of generally accepted good practice.
Interesting read nonetheless.
Not quite, it wouldn’t have been published if the measure was that meaningless (I hope!).
It tested comprehension of the message and content of the chart, for which junk-y charts were substantially and significantly better after a 2 week delay. No difference after a 5 minute delay. It was measured in free recall (points deducted if a prompt was required) and answers judged for accuracy and completeness by an interviewer. From the methodology section: “A complete answer scoring four points might be ‘The chart shows that campaign expenditures by the house increased by about 50 million dollars every two years, starting in 1972 and ending in 1982’…”
That’s not to say it’s a great study. It’s not: sample size of just 20 students, almost all of whom occasionally work with charts (both creating and interpreting them) as part of their studies, making them an atypical sample as well as a small sample.
But in its defence, it included another interesting element often dropped by people trying to characterise it as “Tufte vs Holmes” – it used eye-tracking to see how much time people were focussed on data vs noise parts of the chart. Quick summary: Most (67%) of the time looking at embellished charts was focussed on elements containing data, which was lower than for plain charts (78%) – and “this did not result in a longer time to describe the charts”. Something useful in there. 13% looking at Holmes embellishments, and for both, about 20% looking at other elements (titles and labels were treated as data, so it’s not clear what these were).
As far as I know, the idea of this study has never been replicated or investigated further, which is a shame. I feel like, so many years on, the debate should have moved on from the same old tired tribal war of Tufte vs Holmes, of people quoting opinions written in VDQI vs people alluding to this one small study.
@J – The findings of the study seemed to prove that the recall of data trends and value judgements were improved with the more illustrative charts. What do you consider to be the indicators of a ‘better’ data presentation? I would argue these are (in order of logical interaction, not importance):
1. Appeal – encouraging the viewer to engage with the content
2. Comprehension – providing the viewer with a sound understanding of the content
3. Retention – aiding in the viewer’s recall of the content
While this is only a single study and doesn’t necessarily invalidate Tufte’s approach, the study’s findings suggest that these illustrated charts outperform in both appeal and retention. The fact that data trends were recalled also suggests that viewers also understood the data being presented, at least at a high level. That said, Tufte’s priorities seem lean heavily toward comprehension of precise detail as a primary objective. We must recognize that the objectives of data visualization vary with their application – and priorities shift accordingly.
That is certainly the downside of this type of study…while there may be validity to some of Holmes’ views, a large number of people use this type of study as an excuse to ignore good design practices, and use it as an ‘I told you so’.
Clearly, how to approach data visualization depends a great deal on the intent and the audience.
I would not agree at all with your list of criteria.
I would strongly assert that comprehension is the key criteria of data visualization.
I would argue that providing that comprehension quickly and easily is very key.
I would say that appeal carries far different weights depending on the intent and the audience, and more importantly that appeal is far too broad a term to really be on the list.
It should not be ugly, like the ‘basic’ charts used in the study.
But it also does not require illustration to be attractive and inviting.
I am not going to rail against Holmes, I just don’t see enough merit in this study to support his claim that “studies have proved” Tuft wrong.
Holmes answer to a question in the interview seems to indicate he strikes a good balance between appeal (illustration) and comprehension (data ink ratio).
Q: You’ve mentioned in previous interviews that you believe that the clean presentation of data should be the ultimate goal for infographics. How do you view the relationship between illustration and information? …
A (Holmes): People who do this kind of thing have to have a kind of split brain: analyzing numbers, then portraying them. When this balance gets out of whack, the result is not good. Too much illustration gets in the way of the info; too much reliance on abstract data can leave the reader floundering in a sea of lines and numbers.