Data Literacy and the Art of the Chart

I do my best to stay away from chart discussions. Charts are one of MANY tools I use within my work (data analysis, if you ask me today. Come back tomorrow, it may be different). There’s lots of discussions about which charts are more accurate, how charts are perceived, and the like. This post is none of that. Instead, I’m borrowing from my favorite topic – linguistics – to apply to my other favorite topic – data visualization – to understand why some users hate some charts.

While it feels like a fundamental skill in today’s world, humans are not natural readers. No, we’ll really look at this point for a few seconds. But first….

4 fun language facts to keep in mind:

  • Humans like language. They will create something so long as another human is around.
  • Languages are transient in nature. They’re spoken or signed – this becomes important, I promise!
  • We commit to languages and phonemes at about 13. After that, puberty throws out everything else and it becomes exponentially harder to learn. Go figure.
  • People with more languages have more tools and more nueroplasticity in learning things. It also becomes SO MUCH EASIER to learn more languages after mastering a second.

Writing

Early prehistoric writings were pictographs. Whole concepts could be embedded in a single picture. This branched out in various types of writings, from hieroglyphics to characters and alphabets. I didn’t study characters much, but they fascinate me. I spent my time in school in part studying alphabets and collect them. It makes a fun party trick.

Writing systems are learned willfully and intentionally. While brains may like languages, they’re not naturally drawn to the written word and it requires a lot of effort for our brains to read. Even Socrates thought writing was rubbish and valued an oral history over a written one. We really do not give Plato enough credit here, kids.

So, here’s your exam:

Memorize these. They’ll come up later.

How writing systems form depend heavily on culture. Arabic is an alphabet from a culture with strong taboo around iconography around living forms. China has numerous spoken languages – sharing a writing system provides a common system of communication. Numerous countries have an official language of instruction and documentation while other spoken languages exist, often without their own writing systems. We do not have to have a writing system to have a language. In fact, roughly over 50% do.

The brain reads various writing systems differently. We know this from studies done on dyslexics in particular, which show that dyslexia differs based on language. Chinese characters rely more on memorization and visual systems – dyslexia gets found in the left middle frontal region of the brain (commonly seen activated on ‘language searching or matching’ tasks on English speakers) while readers of alphabets find their dyslexia in the temporoparietal region (close to where spoken language lives).

For those who can read the characters above, you have these saved somewhere in memory. For most of us who can’t (myself included), we may try memorizing certain features. For me, I rely on my knowledge of sign language linguistics. These words were part of a lesson in other signed languages and how signed languages may borrow from characters or other icons. So they live partially in memory for me.

Visual Communication

Which line is longer?

In cultures with lots of modern architecture, we find people typically assume 3 dimensionality in forms drawn on paper and go for b. When we work with populations that are more hunter-gatherer focused, they don’t get tripped up by illusion this at all (they recognize they’re the same). Our visual intelligence is highly culturally dependent. We make the mistake in thinking that how our brain recognizes spatial patterns is universal. It’s not and highly skewed by the people around us and our environment.

So, back to our example of symbols. I’m adding more for you to remember. Hope you’re keeping a list!

Visual communication exists on a continuum. There’s literal (iconic) and abstract (arbitrary) representations of concepts. While flags are fairly arbitrary in nature, it’s close enough to the Olympics that I’m betting more people might recognize it. Recognition does not make it less arbitrary, it just means you learned it or the TV did a good job reminding you. Land masses can be hard or easy, depending on exposure. They’re iconic in the sense that they match the lines we like to draw on land. The last 2 are more towards the middle. The one is an image of a person digging, which is more iconic. The other is a person with a dollar sign (also abstract).

Do you remember your first exam? By now, you might be realizing these symbols match the spacing on the Chinese characters above. The left is China and the right is my attempt to convey ‘worker’ for both the characters and the icons. Written languages are fairly arbitrary while my images include fairly iconic examples and more abstract (arbitrary) ones as well.

Just like with spoken language, our brains filter out arbitrary symbols that don’t have a learned meaning. This is why so many of us toss the Ikea instructions to the side and just make do. How many of these symbols were meaningful to you?

Cliff notes:

  1. Writing systems are not ‘default human nature’ no matter how much you love your books. And trust me, I love me some books.
  2. Different writing systems use different parts of the brain.
  3. Symbols can be literal (iconic) or abstract (arbitrary). The more arbitrary, the more likely it is NOT to be universal without education or repetition or tied to a physical phenomenon.

Now that you have your masters degree in linguistic theory, what in the world does this have to do with charts?!

Iconic and Arbitrary Chart Choices

Charts are a form of literacy. We don’t debate this. We complain about it, some of us (look the other way) more than others (keep looking that way, thank you).

When we deal with representing numbers, we like to take the above example of reality and translate to this very creative chart:

We know this is hard to understand at a glance. Sure, we have an exact count of books bought or returned. It’s arbitrary, just like our written language. Except, from a chart standpoint, it’s not. Because, I don’t think we get to count this as a chart. Nope, this is a chart translation tool and belongs over with written language, not chart-language. We can consider this the Dante’s inferno of understanding.

In the language of real charts, we could argue my first books graphic is a chart. Except, in April, some books go back to the store. Representing negative numbers in unit charts is hard without some other type of cue. So, next chart please!

If ever there was a chart more loved and used, it’d be a table. The bar chart is everywhere – for good reason. It’s the most easily understood and tangible, particularly when dealing with negative numbers. These are fairly iconic in the world of charts, and not just in the Cher-way of icons. We can almost see those books stacked on this thing.

Besides making a nice mountain look, this area chart helps me follow the flow. It’s more abstract than my bar because it’s telling me something else. These books are connected, they’re going over time. It’s filled in, which helps me know which direction to read it. The chart police are likely to hit their sirens and tell me this shoulda been a line chart. Fine. Tableau wanted to do that in the first place. For purposes of giving you grey hair, it’s dual-encoded – you’re welcome.

Sadly, as a line, this chart is becoming more abstract. I’ve lost my anchor at this point and my focus now is the trend, not the tangible items. Before you roll your eyes, give this a good look.

The bar chart requires very little imagination. The only abstraction we have to do with it is recognize the negative. The line chart asks us to recognize it’s showing a bar chart (hidden items stacked) and that the focus isn’t the item, it’s the movement. This, kids, is worse than learning idioms. Even adding the books to the background disrupts this effect and causes the brain to struggle between the literal count or the flow. The line chart is, surprisingly, fairly abstract.

So what happens when we make scatterplots?

We’ve now hit a very abstract chart. We’re creating relationships between two different concepts – book counts and pages. I could continue to add depth to this, such as types of books, genres, and the like. At this point, my novice chart reader is likely to demand a table.

So, what does this mean to you, Tableau analyst?

  1. Humans like to communicate, usually in a transient fashion. Speech, signs, and apparently SnapChat qualify. Remember this when you try to make change anywhere.
  2. Reading the written word is already hard. Now we know why no one likes email. Reading really is more work than a conversation. For about 13% of us, it’s even harder thanks to dyslexia.
  3. Graphs are an elevation of this complexity and also a further abstraction. They rely on the written word AND math. The knowledge of 0 came about pretty late in mathematics. We get really grumpy about negative numbers, even though we recognized them before 0. Go figure.
  4. The more a chart deviates from a known tangible form we find in the world, the harder it is for non-analysts to understand. In short, we get iconic things intuitively because they match our world and question (and toss) arbitrary things. Figure out where your line is for your users and advance them in ways that work.
  5. Commonality doesn’t change abstraction. Many abstract things are known (E=MC^2) but not necessarily tangible.
  6. Things can be iconic and have a low level of precision (say the experts). This is why people argue about pie charts. I’d include one here, but I got hungry. Maybe next time.
  7. Charts combine tangible items (counts of things) with alterations in focus (look at this trend, pattern, or relationship).

And now you know why I don’t write about charts.