About a year ago, I was asked to comment on using a version of this chart for a client dashboard.

The person requesting the review called the chart a rose diagram based on the name in their software library, but I don’t believe that they used an area encoding, so I think it is actually just a stacked pie chart stretched around a circle, and not an actual rose diagram. If you look at the chart closely, you will see that it contains exactly the same data as the (much more readable) stacked bar chart, but the requester was feeling pressured by a client to add the rose diagram to the dashboard because it needed to be more “fun.”

If you’ve read my previous article, you already know how I feel about fun in data vis. If adding some visual interest doesn’t conflict with the primary purpose of the chart, then I’m all for it. Visual flair is fine in its place, as long as it doesn’t interfere with the job of a chart in this context: to support a user task. In an analytical setting such as a business informatics dashboard, I take a pretty hard line on this distinction.

In my experience, the request to make things “more exciting” usually has less to do with meeting an end user’s need than it does with enforcing compliance — it’s often middle management’s first answer to the question, “how do I get people to actually look at this thing that we spent so much time and money creating?” As often happens, examining the underlying need rather than the request often reveals a much more interesting question of company culture, combined with the perceived relevance and usefulness of the information presented, and is almost always better addressed in a different way.

As a designer, it is your job to care more about your user’s comfort and ability to successfully complete their task than you care about your own aesthetic impulse. To me, wavering on that commitment is self-indulgent, especially in the type of task-focused work where you will find most data vis. There’s always a balance to strike here, but when you choose to put your own creative urge before your user’s needs, then you are making art, not design. There is nothing wrong with art: it is a beautiful and valid pursuit that illuminates and enriches the human condition. We could all use more art in our lives — at the appropriate time, and in the right context. 

There is real value in using aesthetics to create an experience or inspire a comparison that someone might not otherwise get from a chart: work that pushes the edges of what counts as “data” and whose perspective matters can provoke important conversations about bias, representation, and the dominant power structures that undergird our definitions of truth. I personally delight in indulging and encouraging the eccentric perspective; it just depends on the situation.

Of course, aesthetics and function absolutely can (and absolutely should) work together toward the same goal. Aesthetic appeal performs a critical function in making design that delights, which is something that we should all strive for, even in a business dashboard. Analytically-focused solutions that dismiss design as unnecessary “window-dressing” often overlook its functional contributions. Disregarding the human experience of your users will always diminish the quality and success of your work. 

Great design means finding that elusive area of overlap where function and form align to create something that supports a task and considers the needs and desires of the human using it, in the context for which it is intended. Function that disregards form is simply not as good (or as useful) as a design that takes the user into account. In cases where the aesthetic impulse overrides the primary purpose of the chart, you’re more likely to create frustration than delight.

To me, the primary function of a dashboard is to be able to read the data, not to be excited by its visual form. The user can and should be excited by the content, because it gives them valuable information in a digestible form that they might not get another way. But the dashboard itself does not need to be exciting, and certainly not — I say this with a shudder — “sexy.” It just needs to help a user complete their task, as seamlessly and pleasantly as possible, and preferably without any inconvenient obstacles, frustrations, or other unnecessary arousal. Making a chart harder to read actually works against that aim.

I do not see the role of a data vis designer as that of a gatekeeper, separating good charts from bad, and telling my clients or users what they should want. My job is not to tell people to eat their broccoli and like it: it is to ensure that the visualizations I build do the best possible job of helping my users succeed, and to deploy every skill in my toolbox to support their needs.

Let’s take a look at some of the pitfalls and stumbling blocks that this particular chart puts in the user’s way, as soon as they try to engage with the chart and actually read the data that it contains.

Uncertainty about scaling. Should I read values as area, or as radius? How can I tell? Florence Nightingale’s original rose diagram used area to scale the chart elements, but people are terrible at estimating areas, particularly for distorted or irregular shapes. As I mentioned before, this chart actually appears to be a radial stacked bar, which would use a length encoding that scales radially out from the center. Without getting into the code for scaling the data, it’s hard to tell what it’s doing. Most people will try to use radius as a proxy for area when they estimate values anyway, so a radial version would probably be the better choice, but the bottom line is that the user simply can’t tell without more information. A good legend and clear chart labels are critical here.

Comparing proportions. It is incredibly hard to compare irregular areas that have very different sizes or shapes. Consider the difference in how you perceive the relationships between the blue and green areas when you view them as radial stacked segments compared to slices in a pie chart (a visualization that is itself notoriously bad for comparing proportions). 

Radial distortion. The hardest thing about radial/polar graphs is that the circumference of a circle gets bigger as you go out, and it gets bigger faster than the radius. No matter how you encode the information (length or area), people will tend to misjudge the importance of some arcs — it’s just a question of which direction the distortion goes. Length encodings will unduly emphasize the outer rings, where the area grows faster than the data because the arc width is larger. Area encodings will emphasize the inner rings, where the radial distance increases quickly because the arc width is so small. Both scaling methods create distortions, and you can see how much this affects your perception of the chart by comparing the same data side by side (the gridline values are meaningless here, and only serve to guide your eye). I don’t believe that we do our users any favors when we grab their attention with a flashy chart that also makes it harder for them to do their job. As a user of your dashboard, it is my job to read and digest a lot of information as quickly as possible, and to use it to make correct and informed decisions. In this context, I care most about speed, accuracy, and precision. A visualization that causes me to trip over visual artifacts or easily misread the numbers will quickly transform my emotion of delight at the novelty of the chart into one that all designers should work to avoid: frustration. 

In a more casual environment where there is less emphasis on accuracy and the ability to ingest data quickly,¹ the area-scaled rose diagram might actually be more successful. If the goal is a quick general overview that minimizes small differences (supported by the exaggerated sizes of the area mapping), grabs the user’s attention, and provides an interesting visual display rather than ensuring legibility of the data, then the rose diagram could be a reasonable choice. The original rose diagram is famous for having done just that, and it’s not an exaggeration to say that it helped to change medical history — for the better. There is also a clear rhetorical bent to this visualization choice, and someone who wants an undistorted view of the data is well advised to beware.

Let’s say that you are determined to go ahead with the rose diagram. For your context, the chart’s novelty is worth the price that you pay for a less accurate visual form, and you (or your client) really, really believe that this exciting new chart does exactly what your user needs and wants. Are there things that we can do to make this chart better? Absolutely.

Minimize distortions where you can. Adding a white center circle helps to reduce radial distortions by avoiding those tricky center wedges where the difference in scaling is largest. The distortions are still there, but at least they are smaller. Note that the empty center pushes me more toward interpreting the chart as a radial stacked bar, though, which may or may not be correct.

Provide cues to tell the user how to read the chart. The biggest challenge with a novel visualization is that people simply don’t know how to read it. For some people, new things are exciting. For others, they are disconcerting, overwhelming, or even scary. Depending on the analytical skills of your user group, an unfamiliar visualization might actually scare people off rather than pull them in. Show these users how to read your chart. Even something as simple as adding an axis can help.

Reinforce the visual encoding with values whenever possible. Most of the problems with this chart come from the user trying to estimate values. If you remove the need to “eyeball it,” you improve the accessibility of data in the chart. The user’s overall impression of the data might still be somewhat distorted by the chart, but the numbers are right there to fix it and support understanding. 

Include interactions or annotations that clarify the visual form and reduce common confusion. Let’s face it: you know that this isn’t the best chart for its analytical purpose. Once you’ve acknowledged that and understood its limitations, you can actively counteract them with thoughtful design. Create affordances (such as interactions) that provide additional support for user tasks whenever possible, and make sure to include clear signifiers to indicate that an action is available as well. The designer who requested my initial review did an excellent job of adding labels and interactions to support interpretation of an otherwise difficult chart, which made the design much more usable than if they had just relied on the chart alone.

Use styles to clarify the chart. Going with a less accurate visualization(voluntarily or otherwise) doesn’t mean that you simply throw in the towel and walk away. As designers, this just means that you have more work to do. There are several things you can do to make a weak visualization more readable. Even small decisions like adding lines between the chart wedges can strengthen groupings and help to clarify how the chart should be read. Dust off your UI skills and play around a bit: there’s probably something you can do to make this chart work better.

Explore alternate visual forms. Once you’ve decided to prioritize novelty, next comes the due diligence. Are there alternate encodings that would keep the novelty, but improve the visualization and reduce the conflicts between function and form? Here’s where you get to be creative: put on your design/data vis hat and get to work riffing off the basic structure of the chart to find other solutions that might fit the bill. 

To me, these last two options are where the real fun and creative opportunities lie for data visualization designers. Shaking things up is not about simply choosing a chart that no one has seen before, but in thoughtfully optimizing that chart for the task it is meant to serve.

Incidentally: the team working on this dashboard did decide to use the requested chart in their initial implementation, but during usability testing, it emerged that users were confused and could not read the data (even with detailed labels to support the task). We’ll take a look at some of the alternatives in more detail a future installment, to illustrate how different chart options support users in different ways, and how we can choose between different chart types by focusing on the user task. In the meantime, you might also want to look at this article on balancing client expectations by Benjamin Xiao. 

[1] Note also that this context means that a user is less likely to pause and check the values listed in the chart, and so any distortions added by the graphic are more likely to escape notice and may go uncorrected.

The post Why Can’t We Have More Fun? appeared first on Nightingale.

Learning to go from data to story is hard. You often have to wear many hats — statistician, analyst, domain expert, graphic designer, and more. One practice used consistently by data visualization experts is “sketching.” Most folks who practice data visualization have notebooks full of ideas for symbols, encodings, and narrative flows. Previous discussion in the DVS Slack channel has addressed this. But what makes sketching such an effective method? What can people getting started with data visualization learn from this practice?

Sketching helps us integrate different kinds of knowledge

Why do professional dataviz experts sketch? This is an important question to ask if you want to help learners develop this practice. Luckily, academics have been digging into this for a while — we’ve tried to translate some of what they’ve found into understandable motivations.

For one, the creation of external representations aids the learning process in many different ways. There have been studies on the role of externalization in learning, and others on the application of drawing to reason and learn within science education. These studies find that:

• drawing an idea makes explicit the metaphors and shorthands you’re using to think about the idea;
• a shared representation leads to more productive conversation and ideation in group contexts;
• physical representations allow you to see characteristics and relationships of concepts more easily.

We sketch for lots of reasons, but these studies highlight some of the key motivations that might inform your own approach to sketching for data visualization. Think about why you want to sketch: what do you hope to learn from the process that you didn’t already know?

The idea of integrating visual, verbal, and mental modes is further explored in Joanna Kedra’s thoughts on visual literacy. Kedra argues for the importance of visual literacy skills in an increasingly “multimodal” environment, and lays out visual literacy skills as being on par with reading literacy and numeracy. We evaluate and interpret images, “translating” between visual and verbal modes of communication, so sketching helps us try out images that communicate our ideas. Makes sense! Of course data literacy and visual literacy are not synonymous — data does not necessarily take a visual form — but, when we’re learning to sketch data visualizations, they are definitely interrelated.

These academic research projects establish why sketching is so important. Our brains are constantly translating the visual and the verbal, so externalizing this process helps us communicate and process more effectively.

Sketching to overcome “visualizers’ block”

People new to visualization often fall back on using data representations they are familiar with — bar charts, pie charts, etc .— without considering alternatives that could more effectively, or more appropriately, show the data. When you’re sketching you’re exploring a broader design space of your data visualization, having a visual conversion with yourself (and others) about how to represent the data effectively for your context and goals. Grammel’s work on how novices create visualizations shows us how this default to known charts particularly happens when learners have difficulty figuring out how to answer questions they have about statistical relationships with which they are unfamiliar. Similarly, Walny’s work on how learners sketch shows us that learners default to numerical representations over more abstract or symbolic ones.

Our takeaway? Drawing can be daunting, but sketches are useful even when the output doesn’t look “professional.” We learn more when we experiment and try new things, and you can surprise yourself with what you see when you try to put your thoughts and ideas down on paper. If you’re stuck on something, even if you feel unimpressed by your drawing skills, try sketching your way out!

When trying to design a data visualization, you can get stuck in lots of ways. Bressa’s work with data visualization novices, workers at a food bank, reveals some more insights from which to learn. This group struggled to move from text to visual descriptions. They also had difficulty managing the spatial aspects of their visualizations. Sketching helps with both of these challenges! Targeted approaches to visual brainstorming can address these barriers more directly, for instance if you get stuck on how to visually represent a concept or word you can try an exercise like creating a visual “word web.”

Building and bolstering data literacy

Why do we care? Our Data Culture Project is a free, lightweight, self-service curriculum of playful activities designed to help people build “data cultures” in their organizations. One activity in this toolkit, Sketch a Story, focuses on helping all kinds of people learn how to sketch — challenging them to quickly draw visual “data stories” based on word frequency data in political speeches and song lyrics. This is an exercise in creating a narrative from data: whatever factoids or relationships stick out to them in the frequency data are transformed into a visual representation, with participants working together in small groups with big paper, markers, and crayons (or when circumstances require, using online tools like Miro).

We’ve run this activity dozens of times, leaving us with a corpus of hundreds of sketches created by diverse groups of data literacy learners — higher-ed students, staff at non-profits big and small, government employees, journalists in newsrooms, and more. Feedback from this activity suggest it has a positive impact: more than half of participants in a testing workshop said it made them feel more comfortable with analyzing text, and a majority also found it both useful and easy.

A preliminary review of these sketches has shown some interesting patterns. For example, color (as opposed to other variables like size, shape, font, or position) seems to be the most common method used to differentiate different elements and categories in a sketch. And the majority of stories being told are comparisons: showing the difference between two parts of the dataset, or identifying an outlier within an overall trend. As we move forward with our exploration, we hope to discover more patterns, and add to what we know about the process of learning to sketch and work with data to tell stories.

Remember: stick figures are OK! The point of sketching isn’t to produce a beautiful work of art, but to get your thoughts and ideas into a format that leads you to new insights and ways of thinking. Importantly, sketching allows you to flex your creative muscles without any constraint: with your paper and pencil, you can do whatever you want. While more constrained tools like software and programs for creating data visualization undoubtedly have their place in the process, they don’t allow for the kind of unfettered exploration you get from just sitting down and sketching. Hopefully, this post helps you understand a little more about what research has to say on how and why we sketch to help us visualize. We look forward to learning even more in the course of our own research, and sharing these findings about the nature of sketching to help you with your own learning process.

This work was also accepted to the 2021 IEEE Visualization Conference.

The post How and Why We Sketch When Visualizing Data appeared first on Nightingale.

Looking at the wide range of contemporary board games presented there, I couldn’t help noticing how much board games have in common with data visualizations. In fact, at their core, all board games are data visualizations. Data and information are visualized as pieces of different colors and shapes (called meeples) placed on boards specifying coordinate systems. The rules of the game determine how the current situation of the data can be transformed into a more desirable state.

Board game players are willing to pay 30–50 € for standard games, and well over 100 € for elaborate expert games. Players spend hours and hours poring over these visual representations of data. That is a degree of user engagement that would be great to also achieve for data visualizations.

The good news is that many of the elements that make board games so engaging, fun, and accessible, are equally applicable to data visualizations. In the following, I will discuss a few such points. Board games use easily readable data encodings, use overarching plots and metaphors, have graphic design that fits the topic, and represent the data in physical form.

Board games tend to use easily readable encodings of data. Categorical data is usually encoded via color hue and shape. This goes, for example, for the different kinds of meeples controlled by each player. Numerical data is usually encoded via location among common axis, number of elements, and size of elements. Board games seldom include more difficult to discern encodings like shades of a color hue (light to dark) or orientation. Using them would quickly result in misreadings and confusion.

The table shows the encodings used in traditional and contemporary board games. The game of Go uses the simplest encoding with black and white stones (interpreted as categorical color hues here although factually color shades) placed on a grid (position). Modern games very seldom use further encodings beyond those already used in the game of Monopoly, first patented in 1904.

In data visualization, if the intention is to get information clearly across, easily readable encodings should likewise be used. The experimental encodings of data art play a very important role in extending the boundaries of the genre. But for many, such elaborate encodings pose a barrier to understanding. I personally have to admit to often skipping elaborate data art if it is too tiring to decode.

Board games make use of overarching plots and metaphors to integrate masses of complicated information. Typical settings of board games include medieval trade, fantasy adventure, armed conflict, and science fiction exploration. The setting provides the information encoded on the board with an easy to understand and memorize mental model. Entirely abstract board games are much rarer. Chess is the most popular abstract strategic board game in the western world. In 1924, Bauhaus designer Josef Hartwig created suitable abstract pieces for the game. The forms reflected the movements of the pieces. These did not catch on. Today, chess still uses the metaphor of two armies with knights and bishops maneuvering against each other to kill the other’s king. The human brain craves tangible plots and metaphors.

Over the last few years in data visualization design, there has been a strong trend to move from presenting rational arguments towards telling emotionally involving stories. This was especially initiated by Cole Nussbaumer Knaflic’s 2015 book “Storytelling with Data.” Narratives integrate lots of individual data visualizations into a whole to make a clear point. The narrative also makes individual facts much more memorable. A good story usually consists of a three-part structure with introduction, conflict, and resolution of conflict.

Board game publishers go long ways to make the graphic design fit the topic. Often the general mechanism and layout of board games are designed by one person (the board game designer/author), and the final illustrations done by a professional illustrator, who sometimes remains unnamed. Illustrations, color palettes, and fonts are chosen to reflect the content.

A wide range of illustration styles are used from rational flat infographics to realistic and very artistic styles. Photographs are rarely used as image material in board games. One reason could be that the use of somewhat abstract illustrations and icons makes it easier to remain in a mental state of imagining and abstract reasoning. In Germany, there is even an award solely for the visual design of board games, the Graf Ludo. If something is beautifully designed we are much more willing to invest time understanding and engaging with it.

Data visualizations can equally be made more enjoyable by using a graphic design language that fits the topic. Header fonts can be chosen to go along with the topic. Color schemes can set the general mood of a visualization. Integrated illustrations and icons can serve decorative purposes. Many good examples of this can be found in the Tableau Ironviz qualifier Dashboards (not in the quickly prepared finals).

Part of the fun of playing board games is to have tangible objects before you. The quality of the game material plays a big role in the enjoyment of a game. Usually, cardboard, wood, and plastic are used. It is nice to touch and literally walk around visual representations of information.

Most data visualizations are pure digital products for the screen. But for workshops, showrooms, and conferences it can be worthwhile to bring a visualization into the physical world. A low-level method is to print a (static) data visualization out as a large poster. Today, there are many possibilities of turning digital graphics into physical objects by 3D printing plastic, laser cutting plywood, or laser engraving on plastic, metal, or glass. If one is willing to put in some manual work, the possibilities are endless.

In this article, I have demonstrated how many elements that make board games so engaging can also be applied to data visualizations. The points discussed in this article are the use of easily readable data encodings, the use of overarching plots and metaphors, the use of a fitting graphic design, and the physicalization of the visualization. These are the more static aspects of board games. Further discussions would be warranted for the social aspects, the interactivity (think interface design), and the gratification and rewards integrated in board games (think gamification).

After a long day, I left the board game trade fair with lots of new ideas and inspirations. To me, it is clear that data visualization designers can learn quite a few tricks from board game designers and illustrators. But the inverse also holds true. I’ve seen quite a few board games that could have been improved with basic data visualization know-how. And it makes me think: “How would a board game look that was designed from the ground up by a data visualization designer?”

The post What Board Games Teach Us About Data Visualization appeared first on Nightingale.

A bright, energetic consultant stares intently at Microsoft PowerPoint. This hero, inexperienced but devoted, hears a door open nearby. A boss approaches.

The boss sits down and asks to see our hero’s latest wireframe for a potential dashboard to sell to a middle-management client. Upon looking at the PowerPoint slide for about five seconds, the boss tells our hero that he doesn’t know how to explain what he asked for any better, but whatever he’s looking at, this isn’t it.

The boss turns to a workstation right behind the hero and pulls up a news website. Right there, beyond the boss’s pointed finger, the hero sees an article published by a major news journal that includes a series of advanced charts. Animated bar charts race against each other in one frame. In another, balls fly across the screen into little bins to represent potential outcomes. The charts all require a high degree of experience to generate.

The boss stands up. “Make it look like this.” He leaves.

The hero only stares. The article uses charts and algorithms that our hero has never seen before. The hero’s preferred software or language won’t even make some of them. Some charts look touched-up in an expensive image-editing application. Our hero doesn’t have access to that.

Freeze frame and focus in tight on our hero. Yep, that’s me. You’re probably wondering how I got myself in this situation. Well.

I Don’t Even Like Balsamic Vinegar, and Work Won’t Let Me Install Balsamiq on My Laptop

Look, Silicon Valley, you win. Really. You’re always going to pump out new technologies that we have to know, or else we’ll fall behind. New certifications, new dataviz platforms, new algorithmic techniques. Ok. Cool. I’ll spend my life tracking down the latest thing and hoping I’m good enough compared to my peers, but that’s life. Right?

Well, sometimes. That’s life… when you’ve done this for a few years. When you’re first starting out? You can’t tell a colorblind palette from a grayscale one. Your resume still includes things like “Captain of the JV Volleyball Team, 2011–2015”. Murder. I liken this to an old meme from several years ago.

I can think of no better metaphor for learning anything at all in the modern data science environment. Transformational data products sit across an ever-growing chasm of technical, unexplained difficulty from the simple charts made by a novice. You want instructions for an animated, clickable pie chart in plain language? Tough.

Donut Charts > Pie Charts, just like Pie > Cake

You, young budding data scientist or consultant who got forwarded this article, you have it pretty bad. Maybe not as bad as I had it. I graduated from college during the Great Recession! I did 3 internships before I got a job in Washington D.C. earning $34,000 a year! You know how far $34,000 a year takes you in Washington? About one student loan payment and, maybe, a taco.

Yet, perhaps you also have it bad in your own way. I lived underemployed for years but at least I found work. If I was ever a failure (personal note: Was? Don’t get too cocky now), at least I knew I was probably in the wrong line of work because a global financial crisis forced its way into my young adulthood.

You have none of that. Instead, you have ten times the things to learn in one-tenth the time. Draw an owl? Please. Draw ten. Use a different language or dataviz app for each. Roll in all the latest dataviz techniques in the newest books. And can you have it on my desk before lunch? Byeeeee.

I Don’t Know Excel and You Can’t Prove I’m Lying

Because you know. You just know, beyond doubt and reason, in that deep pit of your stomach where you keep all those thoughts you have about leaving everything behind and jumping onto a train with vagrants. Deep down there in your soul of souls. You know the place?

Down in your heart of hearts, you fear what we all do — getting exposed. You fear the boss, the one who thinks they’re cool and “gets data” and “knows what Python is”. You’re terrified that they’ll come in and ask you to create this cool thing they just saw someone post on Twitter — an entry in the Tableau IronViz competition, no less — and now they want to see you make that cool thing too.

Down this pit of despair, you descend as you read up on polygonal mapping workflows and “Radials Made Easy” articles. Oh, but they’re actually just links to 30-minute YouTube videos with data that looks NOTHING LIKE YOURS BECAUSE THEIRS ISN’T PIVOTED, *takes a breath,* WHY WOULD YOU SPEND NO TIME IN YOUR ARTICLE DISCUSSING WHAT THE SHAPE OF YOUR DATA IS BEFORE MAKING YOUR CHART YOU ODIOUS CATFISH. You’ve reached rock bottom. You can’t do it. You can’t make that chart.

Or at least, not right now, you cry to your boss. Maybe with more experience? It’s not fair in the least, but thanks for playing. Your boss will now make a face that makes it pretty clear in their cool-boss passive-aggressive way. A face that says you’re not their dataviz person. Because you can’t do the thing they saw that one time.

“I thought you were the expert at visualization?” Nightmare. Update your resume.

The Only Way Out Is Through

Dear friends, the community hears your cries.

And in truth, there’s no time like the present to jump right on in. The Tableau community on Twitter, double-edged sword that it is, can expose your boss to visualizations that you won’t know how to make for a few years, but it can also introduce you to some really, really cool people.

When Tableau Zen Master Yvan Fornes posted a fitness tracker last January that included a radial chart and made the whole workbook free to download, I knew this was a great chance to learn the techniques for radials in Tableau. But I had absolutely no idea where to start, so, I did this.

And then a really cool thing happened.

Within a few hours, I was able to make this dummy chart…

Now if a client ever needs that kind of product, I can say yes to them.

That’s a great way to learn Tableau, by the way. Watch Twitter for workbooks that do things you don’t know how to do, then download them, rip into their calculated fields, and create your own recipe for it. Save, rinse, repeat.

What Do You Mean You Only Know JavaScript? Isn’t That the Same Thing as Java?

That’s also the trick to beating imposter syndrome: just doing something hard once, no matter how bad. Every data scientist in the world, be they analysts, engineers, visualizers, and anything in between, is a force to be reckoned with from the second time they do something onward. The point is, yeah, you’re going to come across new situations and fail them. It’s part of life and we all go through that. As my son’s cartoons say, “Sucking at something is the first step at being sorta ok at it.” Find helpful people and just be honest. You don’t have to compete in IronViz. You don’t have to switch on the fly between Tableau and PowerBI. You don’t even need to know algebra—well, no, you definitely need algebra to do this line of work. Calculus? Depends.

Maybe all this is still more than you need. Maybe you’re waiting for an answer on Stack Overflow. Or for a matplotlib blogger to explain how to put value labels on bar charts without a terrifying-looking For loop (Python devs, go check out Altair!). Or how to get a map to zoom in with D3 without digging through four textbooks.

That’s all, all of it, fine. It’s part of your job and you’re coming onto the scene at an exciting time. All these great, awesome, wonderful tools and all these coding libraries let you do so many amazing things! And yet, only a handful of people can explain how it all works from start to finish in regular words.

None of that is your fault, as long as you keep looking and learning. It is, I submit, your duty to the data science community to search out the answers and share them with the rest of us. A data scientist’s brain is like a shark: it has to always be thinking, moving, learning, analyzing, studying, creating, logging, cleaning, proofing, plotting. So I’m going to say this as clearly as I can: failure isn’t being under-experienced, it’s stopping. To stop is to admit defeat, and that will never do when your cause is all the knowledge in the world.

Hey, maybe you don’t have questions. Maybe you’re just looking through the plethora of new technologies and wondering “well… where exactly do I start?” The answer can only ever be: at the beginning. Ask yourself what you want to do, and why. Then seek out a way to do that thing. Then do that thing.

Or maybe instead of learning a new, exciting plotting library, you’ve been studying for an AWS exam for a month and you just failed it earlier today, so you went home and wrote an article so you could remind yourself that everything’s fine, and you’re fine. You’re fine. You’re fine… but you failed by two questions, and… and how could you let that happen?! What a sham, what an imposter, everyone at your job already knows that you’re a failure and a fraud and you just gave them proof, you absolute idiot. How could you ever get this far being so bad?! Donald Trump is already drafting a tweet calling you a loser, you dreg. Adulthood is filled with people telling you every time you do things wrong, but when you do something right there’s nothing but a similarly-sized void where you’d expect affirmation to be. But look at what you learned, Phil. You didn’t pass the exam, but you know a lot more about what you wanted to know about.

Maybe all of our inexperience, immaturity, failure, is just what we get for trying and shouldn’t be thought of so poorly. Maybe failure is wholesome. Maybe the real Solutions Architect exam, or interactive sunburst chart, or web-published study, or good feedback round with the client isn’t the real prize: the failures getting there are.

Go make a good mistake. Close this article and draw the rest of your chart. You’ll do fine.

Philip Hawkins wants to write science fiction when he grows up. He works as a consultant for the US government and is unsure what else to put here, if anything.

The post Draw The Rest Of The Chart: Imposter Syndrome and Leveling Up in Data Visualization appeared first on Nightingale.

A while back, I discovered that information design was its own field, and I wanted to know the fundamental principles and best practices. I also found myself lost. Whether it was picking out a learning resource, charting tool, or styling option, it wasn’t always clear why to choose one over the other. Like so many just starting out in the field, it was frustrating to hear the same answer to many questions: “It kind of depends.”

But are there certain sorts of problems that data visualization can help with or needs it’s especially good at meeting? In a recent conversation, we asked members of the Data Visualization Society (DVS) to give their take. Welcome to a rough guide on knowing where data visualization can help you.

1. To challenge assumptions and expand the mind

When a Swedish professor was taken aback by how much misconception existed in the world, he went on to make it a mission to “fight devastating ignorance with a fact-based worldview that everyone can understand”. His name was Hans Rosling, co-founder of Gapminder Foundation, and his common tool of choice? Charts on global trends, combined with great storytelling.

Data visualization is particularly good at helping us to think beyond our own personal experiences, especially if it’s on issues that are little known or not often covered in the media. When an event or issue easily springs to mind, we tend to exaggerate the importance or frequency of it. Likewise, we tend to downplay things that we are not as familiar with.

By making the data more memorable and “sticky” in our mind, vizzes can help overcome availability bias, which is our tendency to give preference to information or events that are more recent or observed personally.

The effect is amplified when we insert the audience in the data visualization. For example, in The New York Times’ interactive article ‘How Much Hotter Is Your Hometown Than When You Were Born?’ piece on climate change, the audience is asked to enter their hometown and year of birth. The article then goes on to compare the number of hot days one can expect today relative to when they were born, as well as what they can expect as they age. Climate change becomes something more personal and relatable.

That said, sometimes it takes more to convince your audience. In one dataviz practitioner’s experience, when the charts don’t match expectations, the audience’s first instinct may be to find fault with the data. Cultivating trust is necessary, particularly when audience members see themselves as domain experts on the topic.

2. To reason about data

It’s a myth that designing visualizations is only for the end of the data analysis process or when you are ready to communicate some insights. Coming up with quick and dirty prototype charts or even sketches has its benefits at the early stages.

The more obvious upside is that it’s easier to pick out patterns and outliers from a chart than chunks of raw data or rows of summary statistics. But besides using data visualization as a way to understand, we can also use it as a way to think. By translating our internal thinking process into objects in the external world, we clarify our ideas and make them more actionable.

In my case, thinking or prototyping through vizzes gets me in an experimental mode. It opens up more possibilities and curbs any pesky perfectionist tendency to get everything right the first time. For others, it can serve as a complementary approach to writing, which fosters a more sequential frame of thinking.

3. To unleash beauty into the world

Some pleasures cut across age, gender, and ethnicity: rainbows, bubbles, and ice cream to mention a few. When a data visualization looks and feels a certain way, it can give rise to similar feelings of delight.

Maybe we can’t always put a finger on the particular aesthetics that made the data sing, but we can usually agree that the visualization was beautiful. In this sense, the desire to produce a data visualization (or even data art) is linked to a human impulse to create beauty.

Some DVS members even see data visualization as a way of reclaiming space — to breathe new life into the serious spaces that typically neglect or reject beauty. Beauty embedded in a data visualization also has a functional role. To quote design guru Paul Rand:

Ideally, beauty and utility are mutually generative. In the past, rarely was beauty an end in itself… The function of the exterior decoration of the great Gothic cathedrals was to invite entry; the rose windows inside provided the spiritual mood.

In the data visualization context, beautiful design serves to guide users to key elements and aids in their understanding. Yet, despite the net positive results, people often resist the pursuit of beauty or dismiss it as a frivolous act altogether, particularly in the business world. When handling dashboard designs, dataviz practitioner Jason Forrest previously found his colleagues rejecting suggestions of beautiful designs in favor of something more basic. His workaround was to make the design prototype anyway, which got people more interested in the suggestions.

4. To connect in an attention-starved world

Data visualizations, particularly beautiful ones, have the benefit of acting like eye candy. They can get your audience to stop, look, and hopefully, engage with the data. Information designer and LinkedIn instructor Bill Shander tells us not to underestimate the “value of eye candy to simply generate interest”.

The Pudding is a great example of tapping into data visualization’s eye candy power with its rich visual essays on topics like how women’s pockets are inferior and the laughter climax of Ali Wong’s stand-up comedy. As data journalism becomes a global field, we will see more outlets drawing on data visualizations to create compelling content.

Why does it work? As a visual metaphor for data points, data visualization has the ability to make ideas more easily digestible and captivating at the same time. In this way, ideas embedded in visualizations are more likely to persist and spread.

It’s not only dataviz or data journalism folks who recognize this, either. Marketers know that eye-catching data visualizations combined with a powerful narrative can be very shareable and persuasive, as exemplified in the “Data Visualization + Data Storytelling Is Marketing Gold” article making the rounds on the internet.

5. To define and shape data culture

If the hallmarks of a culture include the community’s shared languages, food, music, and social habits, then in much the same way, data visualization practices are a hallmark of an organization’s data culture. Ultimately, the kind of data visualizations that can be produced is the result of how data is organized and integrated from various sources in the organization.

As DVS founding member Elijah Meeks puts it, “Do you boil everything down to a few KPIs? Are you comfortable with sophisticated representation? Do you think about uncertainty? All that is expressed in the data visualization that you use in your presentations, your email reports, and your public dashboards.”

And despite the growing imperative to be data-first, data-driven, or [insert new buzzword] in today’s economy, it’s not possible to build competency in everything at once. Data visualization can be a gateway for getting people to be more comfortable with unfamiliar data practices. For dataviz practitioner Wendy Small, the use of more simple data visualizations like line charts has been a healthy and effective way to encourage new approaches to reading data as part of a data literacy initiative.

The clarity that data visualization provides also encourages people to work better. When dataviz practitioners Keisha and Evelyn Münster handled projects on process-related data, they found that visualizing the process details proved more illuminating than relying on some aggregated numbers. It also sparked better conversations about what was going on.

6. To experience hobbies (or anything else) through a different lens

Data visualization is not all serious business. It lets us geek out to our heart’s content on our interests. Exploring the visual patterns that emerge from a data set on a hobby is another way to enjoy the hobby.

If you enjoy beer, check out Nathan Yau’s chart of beer styles, which looks like a beer-colored mosaic, complete with details on flavor as well as how high each beer style tends to be in alcohol content and bitterness. It’s a great way to explore multiple types of beer, from the American Lager to Fruit Lambic, without the risk of a hangover. Or maybe you are a fan of meta, and you may enjoy Christian Swinehart’s data storytelling of the storytelling structure in Choose Your Own Adventure Books. There seems to be something out there for everyone.

What about the “quantified self” movement, where people mine vast collections of personal data and visualize them?

When journalist Lam Thuy Vo’s marriage dissolved a number of years back, she created a blog Quantified Breakup to organize her responses in data visualizations. One post showed her apartment-related weight loss as she got rid of furniture. In another post, she tracked text messages exchanged with people she met online after the divorce and visualized those messages as sparks that flew off the screen.

It’s like taking pictures to remember a beach vacation in Bali, except we’re not just logging one-off moments. We’re tracking the same thing over time and compressing the result into a series of ebbs and flows or whatever patterns it shows. In some way, the process of visualizing this data functions as a form of self-discovery or, if we’re hurting, self-healing.

A Final Note

There you have it, six reasons we create data visualization. This list is not exhaustive by any means, but it shows the value of data visualization as it sits at the intersection of beauty, influence, and expression.

Thanks to the Data Visualization Society members for contributing to the discussion, including: Alexwein, Bill Shander, Bridget, Cameron Yick, Elijah Meeks, Evelyn Münster, Erica Gunn, Jack Merlin Bruce, Jason Forrest, Keisha, Matthew Montesano, Nicole Edmunds, Stephen Singer, Wendy Small.

The post Why We Find Joy and Value in Creating Data Visualizations appeared first on Nightingale.

Every couple of weeks, the Data Visualization Society (DVS) Slack hosts an in-depth discussion on data visualization issues. An inviting space for collective amens and groans, members have ruminated on important topics like the biggest data visualization issues, how to get started in data visualization, experiences learning data visualization, data literacy, and creating positive data visualization cultures within organizations. Receiving validation that one’s experiences aren’t exclusive can make these discussions feel like a warm hug while learning via our divergent experiences can be illuminating as well.

While many of the challenges that arise in these discussions are applicable to data visualization for target audiences in low- and middle-income countries (LMICs). There are also unique challenges when visualizing for LMIC audiences, which have not been extensively documented. Individuals who create data visualization products for LMIC audiences recently shared their experiences via the DVS Slack and a Google Form. Summarized below are highlights from the discussion which offer tidbits to anyone interested in visualizing data for LMICs and ideas on how as a community we can support establishing the data visualization field in LMICs.

What kind of LMIC data exists?

LMIC data is plentiful for some sectors (health, economy, demography) and less plentiful in other sectors (culture, environment). Collecting nationally representative data is expensive, and data availability is a reflection of local and donor priorities, as well as funding availability.

For all countries, including LMICs, data that is not accessible or does not exist can make issues or people invisible. Lack of data access is a common issue in health and the sciences and is a significant barrier towards “open data.” Missing data has been discussed by the European Journalism Center related to marginalized communities in Afghanistan, Pakistan, Kenya, Kyrgyzstan, and the Balkans, and the absence of the African content in “big data.”

Health and demography LMIC data are particularly well-funded, and several data sources are publicly accessible for those interested in using this data to create visualizations. Large nationally representative household surveys are conducted in most LMICs every three to five years by the Demographic and Health Survey Program and UNICEF Multiple Indicator Cluster Surveys Program. Data sets are available for public download. There is also substantial investment in routine health management information systems (HMIS) in LMICs, which offer more up-to-date data collected at the community and facility levels. District Health Information System (DHIS-2), the most widely used HMIS platform, is used by 67 LMICs. DHIS-2 data is generally not publicly available, however, some LMIC countries integrate DHIS-2 data in publicly accessible tools.

Examples of other LMIC data sources:

• AidData
• IHME Data
• National statistical bureaus
• Our World in Data
• UN Data
• Wellcome Global Monitor
• World Bank

Is the LMIC data that exists of acceptable quality to visualize?

Data quality and confidence are nagging concerns for all LMIC data — regardless of sector—and can be quite political. For example, the Government of Tanzania passed an amendment to the National Statistical Act in 2018 criminalizing questioning of any government statistics. The World Bank, in response, threatened to put a $50 million grant on hold. Which raises a familiar existential crisis that plagues anyone who works with data: What is the truth? As DVS member Duncan Geere commented:

“We have a responsibility as a community to examine and question what’s coming into our particular phase of the process, rather than accepting it blindly as truth. Too often I see people making beautiful graphics built on datasets full of lies.”

One contributor to poor data quality is inconsistent data management and cleaning practices, which can be a barrier to getting to visualize data.

“…the potential impact of the data visualization is reduced by the time & resources necessary for making mitigating decisions around the data quality, creating additional reports & visualizations to highlight data quality issues, and even create new processes.”*

Improving data quality is an important step to visualizing data. As with visualizing any data, visualizing LMIC data should involve a reflection of data source and triangulation of multiple data sources if available.

What are tips for navigating unique experiences and challenges when visualizing for LMIC audiences?

There are common challenges experienced by anyone working in data visualization, which have been discussed on the DVS Slack: preferences for certain types of graphs, traffic light colors (psst…there are better approaches!…perhaps an idea for a future Medium post), and variable data literacy. Highlighted below are some more unique experiences when visualizing for LMIC audiences.

1. Prioritize data visualization approaches that are easily understood and actionable by the audience

Sometimes prioritizing a snappy data visualization approach isn’t the most appropriate. Since a lot of the data visualization for LMIC audiences is tied to the development sector, a common target audience for data visualization products is high-level government officials and other decision-makers that are bombarded with data from individuals advocating for different priorities. With limited time to digest data, it becomes extremely important to use data visualization as an effective, memorable communication tool.

“One guiding principle I have is that the graphic is only as useful as the audience finds it. This principle has two implications: 1) that feedback and revision is essential in the process of making an effective graphic and 2) that audience understanding/acceptance should have greater weight in making decisions than design preferences.”*

If you are unable to justify the added value of a complicated graph over a more simple approach, you may want to re-evaluate if it’s the best choice.

“…if there are not compelling benefits to why a complex graphic is THE best way to display the data and has benefits that cannot be replicated by a series of more simple figures, then we need to re-evaluate the utility of the visualization. There have been many instances when simple (albeit boring!) graphs have been the most effective way to communicate data. There is merit in simplicity that is justified by the perceived utility by the intended audience.”*

But does this mean that you should only create bar and line graphs? Not necessarily. One contributor described how in their experience, deviating from bar and line graphs can cause confusion and interpretation challenges among colleagues in Sub-Saharan Africa. To overcome this, they encourage stronger facilitation and interpretation support, and annotated graphs:

“Adding key interpretation messages or symbols inside the chart area helps to focus the participants’ attention towards the key take home messages, and aids in understanding how to interpret less common types of visualizations.”*

2. Account for variable technology access

One of my pet peeves is being sent an interactive dashboard that has been made for a LMIC audience, which loads painfully slow even with my relatively reliable internet connection in Baltimore, Maryland. I become an eye roll emoji: ?. My patience is admittedly not super; my bigger concern is whether the actual target audience will experience the same trouble loading the dashboard.As DVS members and survey participants commented: not everyone has reliable internet access. This should be taken into account, particularly if designing interactive data visualization platforms that may require high bandwith. Internet access is unavailable and/or unaffordable to the majority of the world’s population — predominately those in LMICs. Many individuals in LMICs rely on phones for primary internet access, and smartphone usage is quite high. The popularity of social media platforms like WhatsApp and Facebook, and particularly the role these platforms play in sharing information in LMICs is an opportunity to encourage mobile-friendly data visualization to accurately communicate data.

Projectors and printing services in LMICs can also be variable in quality, which can affect visibility of graphics that are highly dependent on color. In general, it can be beneficial to avoid nuanced color gradients and light colors, which may not show up with the intended visual acuity.

3. Make sure approaches are culturally-sensitive

While metaphors can be powerful in data visualization, cross-culturally they can be problematic. Icons used also must be appropriate to the target audience. For example, I have seen baby stroller icons used to represent babies for audiences in sub-Saharan Africa. Strollers are less commonly used in the region, and thus the icon is less appropriate for the audience.

What resources exist to help grow the field of data visualization in LMICs?

DVS members shared the following resources that may be of interest to individuals in LMICs:

• Tableau offers donated licenses to small non-profits, students, and educators.
• RStudio offers scholarships to those in under-represented countries to attend their annual conference and supports regional R groups like LatinR and Africa R Users.
• R Ladies chapters exist in several LMICs, and participation is not limited to ladies!
• Regional training hubs like Eastern Africa Statistical Training Centre and non-government organizations that support data visualization like eHealth Africa.
• Global Health eLearning Center’s free course on data visualization

“My observation is that data visualization is not a very known field. And even though many of the students intensively work with data, there is not enough occasion that stimulates their curiosity for exploring other visual forms than the very basic ones to represent their data. I think the (word) field ‘data visualization’ is not sufficiently taught or discussed in the place where we were, and that’s probably the case for many students in Sub-Saharan countries. I personally know more EU or US-based than Africa-based vis practitioners. The accessibility of the tools and materials were also discussed: Not enough books on the topic, Not all students have continuously reliable access to Internet, the language in which most well rated and known tutorials are written.”*

DVS members and survey participants commented that it is more common to see data visualizations for LMIC audiences prepared by non-LMIC individuals. A major reason is that recognition of the data visualization field may be lagging in LMICs compared to that in non-LMIC settings. There also can be fewer educational and technological resources available to those in LMICs that want to learn about data visualization.

As a field, how can we help encourage the growth of data visualization in LMICs? Suggestions discussed include the following:

• Creating accessible training resources in different languages
• Providing accessible support and mentorship
• Hosting conferences and meetups that are affordable and accessible to those living in LMICs
• Encouraging collaboration between LMIC and non-LMIC individuals and ensuring that those in non-LMICs engage with local communities when developing data visualizations.

DVS will be revisiting this topic in September for discussion. In the meantime, please feel free to continue to share your experiences designing data visualization products for LMIC audiences via the Google Form survey.

*Submitted anonymously through Google Form survey

Special thanks to the following DVS members that participated in the discussion: Dolly Andriatsiferana, Ahmad Barclay, Amy Cesal, Will Chase, Jason Forrest, Duncan Geere, Erik Kemp, Amanda Makulec, Elijah Meeks, Noëlle Rakotondravony, David Mora, Stephen Singer, and Wendy Small

Thanks to Stephen Singer and Noëlle Rakotondravony. 

The post Visualizing Data for Audiences in Low & Middle-Income Countries appeared first on Nightingale.

Despite how well-intentioned, the distance that forms between a data team and its audience can often be attributed to a lack of empathy. This can lead to a reactive relationship with your audience; simply responding to requests for information without any real idea why. Its likely that your colleagues think you just ‘don’t need to know’ why they need the data (or visualization) in the form that they expect. Maybe they think you don’t care why they need it. Regardless, if you’re trying to find new methods to explore data and visualize it in meaningful ways, you need to reposition your relationship with your audience so that they trust you enough to collaborate.

This article itself is a collaboration between myself and Elijah Meeks and it explores some methods to reposition your relationships in order to build more trust. His article outlines some concepts for exploring new types of data visualizations in collaboration with your audience once you have that trust foundation. It’s been fun to collaborate on this, so please let us know what you think in the comments!

You have to want to earn it

There are many ways to build trust with your audience, but the most important is that you want to earn it. It sounds minor, but the moment you decide that you want to earn your audience’s trust, you shift away from a reactive mindset and towards a collaborative one. Reactive relationships revolve around a request and a reaction to that request. Trust is also important in these relationships (and we all have them) but the requestor is likely coming to you because they know they will get what they want — it’s a transaction.

Collaboration is a bit more complex because it means that both parties work together to solve whatever issue is in focus. While collaboration may seem more difficult, the benefit from collaborating with your audience is usually greater than the individual contributors. Ideas are built in an inherently iterative manner, as everyone involved has to contribute ideas. There’s a dynamic flow of communication in collaboration and the moment anyone’s trust erodes, the collaboration usually ends.

Trust becomes the currency of your collaboration, so how do you earn it?

1. Empathy is curiosity

In order to collaborate, you need to know enough about what your audience needs so that you can tailor your approach to help them do it. It all starts with empathy, meaning, do you actually care about what they want and why they need it? Empathy is universal, it’s non-hierarchical, it’s free — and best of all — empathy creates more empathy.

One of the best ways to breed empathy is simply by talking to people. By becoming curious about what your audience is doing and why they need data for it, you begin to position yourself as a person that cares about your work and how you can be a good collaborator with them. You’ll be surprised at how infrequently your audience has been asked basic questions about what they do and who they are. Everyone likes to talk about themselves, so when you ask a question it signals their importance in your relationship. Just by showing curiosity, your audience will start to wonder why you care so much, and this usually forms the basis for more conversations. Before you know it — empathy creates more empathy.

A case study from Elijah Meeks:

“From a practical perspective, empathy has to be demonstrated in your work as well as your meetings. In most cases, when I’m asked to create an analytical product at Netflix, it already exists in some other form (a Tableau dashboard, an rShiny app, or a notebook). The first thing I do is re-create the data visualization in the form they expect (a bar chart, a cumulative density function, a table), and only then do I introduce other visual forms as context or prototypes for new approaches. With data visualization, audiences have strong associations with certain data taking certain forms — not just the charts but often the whole “dashboard”. Demonstrating empathy in these situations is accomplished by rendering the form they expect and then gently transitioning that form into something more effective. Hence the emphasis on semantic similarity in exploratory design.”

2. Talking to people helps everyone learn their roles

When you care enough to ask your audience how you can engage with them, you will likely learn how you can work together to get it done. This effectively activates your role in the relationship; and in doing so, helps to demystify your work and allows you the additional insight you’ve been looking for to create more impact for your audience.

Remember that data visualization is itself a form of communication. When building data visualizations we have to avoid the tendency to look at requirements as just a set of checkboxes to tick off. This isn’t going to be accomplished by waiting for your audience to suggest a different technique — they probably don’t know other data visualization techniques, or worse, they’ve been burned by a beautiful but useless chart before. Instead, by talking to them about their needs, you learn what your audience might want to see if they were more aware of advanced techniques.

Talking to people doesn’t just mean gathering their requirements and understanding their goals — you continue to collaborate with your audience through a dialogue built with charts. Your goal ultimately is to communicate the salient details of the data, but on the way, the charts you develop to help your audience understand the possibilities, as well as the visualizations they’ve already relied on, becomes a conversation of its own.

This kind of relationship changes the traditional, reactive relationship. “Show me the numbers” is therefore not a transaction any more but becomes a story told in concert with your audience. The data comes to life through your conversation and the collaboration is strengthened. Trust has been earned.

3. Create a shared understanding

But talking to people isn’t the whole story. In order to earn trust, you also have to ensure that everyone understands (and agrees) on the definitions of the data, what the terminology being used means, and where they can easily learn more if they do not. This is a 2-way street, as the terminology that your audience uses may likely differ from your own, so mutually understood terms become key to having a meaningful conversation.

While this seems absolutely commonplace, I’d argue that most teams do not share a common understanding of the terminology or even of the data itself, what the data is called or what it represents. Just take a quick poll of your colleagues and you’ll see.

A byproduct of crafting explicit definitions is that it exposes inconsistencies. Exposing the definition of the underlying assumptions often reveals when your audience thought they were using a shared definition but were not. Large-scale Business Intelligence applications (or even file-sharing platforms) can actively be leveraged to foster alignment to create these understood definitions and metrics. Since multiple colleagues/teams are all using the same platform, it allows for these common terms to radiate through an organization and call out discrepancies. Central, integrated systems serve a single source of truth that seeks to consolidate institutional domain knowledge and spread consensus.

What’s more, taking the steps to find, discuss, and agree upon those definitions requires the empathy to care about the importance of a shared language, and the time and focus it takes to explore this… builds more trust!

4. Data sources and quality

Even when you’re talking and agree on the terms, the actual data source and quality of the data is important to expose. If your audience doesn’t have confidence in that data source, it will undermine the validity of whatever it is you’re trying to show them, thus reducing their confidence in you and eroding the trust you’ve been working so hard to build.

If the data quality is bad, just point that out and expose the uncertainty. Exposing the source and expressing your confidence in the data will cause you to explain why it’s robust or crappy and will be just one more journey you’ll take with your audience. Everyone will walk away understanding the system and their roles a bit more. Definitions will be reinforced, empathy expanded and trust cemented.

How to cope with uncertainty is a different question, as many of the data visualization books just don’t give enough clear guidance. Maybe this is because our industry hasn’t established the go-to methods for visualizing these facets of the data. So you’ll need to discuss this uncertainty and how to visualize it with your audience. When visual methods don’t work, fall back to using callouts, titles, and other text to explain in short, easily digested sentences (preferably with links to detailed definitions).

The Trust Infrastructure

All this is just to prove that you want to collaborate with your audience. In doing so, it’s highly likely that your relationship with your audience (even if it’s hierarchal) will change. “You don’t need to know” will likely turn into a discussion about what happens next after the data is understood. As a result of all the trust currency, your methods of presenting data with an expanded data visualization literacy will likely change from simple reporting to a collaborative conversation on the insights inside of the data.

It’s at this point that we as data professionals become valuable thought partners. When everyone in a decision ecosystem understands their role, it allows for a more inclusive exchange of ideas. This fluid communication turns from earning trust into building respect and allows a team/organization to move from a siloed, reactive mindset towards an integrated, iteratively minded team focused on solving problems as they come in a fluid manner.

The trust infrastructure you build is an important strategic necessity to get the impact you seek in your work, and if you do it right, it will make your work more enjoyable too.

The post Build A Trust Infrastructure Between Your Team and Your Audience appeared first on Nightingale.