Visuals for the Methods Section

You can tell when someone is getting their data viz eye well-honed because they start asking where ELSE can I include visuals that will better explain my work? That’s Sue. She had grown comfortable with high-impact, story-telling charts in the Results section of her reporting but was staring down a solid wall of text when it came to her Methods section. She pitched the question to me in an Academy Office Hours session: What can I do with the Methods section?

In Sue’s case, her methodology included a phone survey and she had a thick paragraph describing who they included in their sample, who was removed from the sample, and how many people they ultimately included. This is a grand opportunity to include a waterfall chart. Waterfalls show the increases and decreases, usually over time.

You might also be able to tell a story about sample size with nested boxes.

If your methods section includes a demographic description of your sample, try adding even simple bar charts to that section of your report.

And if your demographic description includes a comparison between your sample and the general population, pop a proportion plot in there.

To be honest, my preference is that we move that Methodology section to the end, after you discuss your Results (which is what people are most interested in reading). No matter where you choose to trot out your methods, that content is sure to get more attention when you add some data visualization.

In the Academy, we have tutorials on how to make each of these charts in Excel, Tableau, and R. Heck, in the tutorial on Demographics, we share a dozen different ideas for ways to make that data more graphic. That tutorial, just like this blog post, came about because an Academy student asked me a great question, “How do I…?” In addition to tutorials on just about any chart type you can think of, we are also on stand by to dream with you and, together, build some awesome solutions.

Add Labels ON Your Bars

Meg, an Academy student, sent me a great question for our monthly Office Hours call: How do I replicate this?

Pretty cool chart, right? It comes from Global Web Index, who put out tons of relevant data in interesting graphs.

Meg said she keeps an eye on their work and wants to emulate their style a little more, especially how they put the category labels inside the graph, sitting on top of their bars, rather than out to the left like we would traditionally see. Is there an easy way to do this or is Meg setting herself up for inserting a lot of textboxes?

It is even easier than you might think! I’ll show you the directions on how to do this in Excel but the strategy will be very similar if you are in other programs.

In addition to the data you want to graph set up in a summary table, add an extra column for those labels and, to give yourself plenty of space for your words, the data you add under Labels should match the highest percentage in your dataset. Then insert a basic bar chart.

Right-click on one of the Label bars and select Format Data Series. Change the fill color to No Fill. Then right-click on one of those bars again and select Add Data Labels. This will add 35% to the end of every Label bar you just turned No Fill.

Right-click on those labels and select Format Data Labels. In the menu that opens, change the label position to Inside Base and, in Label Contains, check Category Name and uncheck Value.

Haha! We are so close. Most of the labels are bunching onto two lines. To fix that, look in the same Format Data Labels menu and click in the third icon you see across the top (we have been in the fourth icon). UNcheck Wrap text in shape. Boom!

Then you just format. Delete the vertical and horizontal axes. Delete all gridlines. Add data labels to the other bars. Remove Labels from your legend. Remove the chart border. Add a storytelling title. Make rock n roll horns with your hand and bang your head a little cause you are a rock star.

Thanks to GWI for the inspiration and thanks to Meg for the great question.

One of the best skills to grow is how to be on the lookout for inspiration. It is really easy to criticize other people’s work. It is much harder – but far more valuable – to identify what really worked, what you can learn from, what can be integrated into your own skill set (with credit).

That’s why “say two things awesome about this visual” is a repeating theme in our Office Hours sessions inside the Academy. We hone an appreciative eye and figure out how to love and adapt instead of just critique. You can come bloom with us in the Academy, too.

Graphic Designers Need To Learn Data Viz

I am a data nerd. I grew up, professionally speaking, designing studies, collecting data, and trying to get people to make use of the analysis and results. After I married a graphic designer, I realized that no one was going to pay attention to my amazing, glorious, wonderful data if my reports and presentations looked like Microsoft defaults. From that point forward, I became a design nerd, too.

When data and design successfully bridge, we can change the world.

Data nerds need designers. Design is how we create engagement and spark action.

But designers typically avoid a deeper understanding of data and how to best present it. Why?

Traditional chart software, like Microsoft Excel, can feel intimidating (yo, it can feel intimidating to data nerds, too). But the graphing tools in Adobe Illustrator and InDesign are mediocre at best, only capable of the simplest of chart types.

In addition to unsupportive software, my design clients often tell me that they see themselves as “art people” not “numbers people.” Data visualization isn’t included in a designer’s academic coursework. 

My buddy Peter Brakeman of Brakeman Design, said that sometimes the trouble is compounded by the initial design request. Clients will ask designers to “make this look pretty.” So, Brakeman said, “a bar graph showing the amount of corn shipped to China for each of the last five years has ears of corn stacked up to the desired total for each year.” That design would satisfy the client’s request of making something that looks pretty, even if it is the wrong graph type and doesn’t communicate the data effectively. It is not that a corn graph is technically bad design, it just has the potential to be so much more.

Designers have traditionally felt that data nerdiness is not in their wheelhouse. However, today’s organizations are data-driven. I’m not even talking about companies like Facebook that are collecting big data. I’m talking about the smallest nonprofit in your neighborhood, who is doing its best to use data to improve programs and services and make a bigger impact on the world. Everyone is looking to be more strategic by using data to inform decisions.

This is an area designers can no longer afford to ignore.

So where are you supposed to learn how to do data viz justice?

Tackling data might seem as painful as all the questions from your family about when you’ll start having kids.

But I’ll let you in on a secret: Most of the hard work is simply stripping out the bad design baked into our default graphs and using familiar design principles like unity and proximity to present the information in a clear way — much like designing a clear and user-friendly interface, for instance. As a designer, you’ll feel at home once you clear the initial intimidating hurdle of just getting started.

Sherri Days is a graphic designer who enrolled in my Graph Guides program. Graph Guides is an intensive experience — you learn to make 50 charts in a year under the guidance of a dataviz expert.

After just the first few lessons, Sherri was making knock-it-out-of-the-park visuals. Figuratively and literally: This one is about baseball.

Sherri combined a spectrum display (a great chart for qualitative data) with a set of dot plots. Stunning work. I love it so much.

For another lesson, Sherri pulled in salary data from Michigan State University.


Part of what makes Sherri’s work so impressive is her graphic design background (though, if you don’t feel strong there, we’ll teach you that, too). You are going to be good at this.

Her design eye factored into this beautiful line + area chart on Aretha Franklin’s career. She said “I wanted to show the point when her recording career really took off. The color scheme is based on that record.”

We do not teach designers to create the type of data-driven visualizations that look so complex and intricate that they could be trendy art for your living room wall (though there is a place for that in the world). And you don’t necessarily need to learn a programming language in order to make great graphs. Sherri is using Excel. Yes, Excel.

Come join the Graph Guide program. We aim for a sweet spot, where designers readily understand data and where create compelling visualizations inside broader stories that engage, inform, and inspire.

Parts of this story originally appeared on Medium

How To Teach People Data Visualization

Have you ever accidentally taken one step too far down the internet rabbit hole? This has happened to me more than once when I’m trying to google the solution to why my chart isn’t working. I click into a Q & A forum that teases the answer – but the solution is some code that makes me run away screaming.

When we are trying to teach people data visualization, we don’t want them running away, screaming. And it happens FAST. One innocent step too far and BAM, folks are outta there.

The trick to getting people to stay around is something I learned back in my teaching days: Scaffolding.

Scaffolding means you meet people where they are and take them to the very next step, without overwhelming them with too much, too fast.

I think educators borrowed this idea from the construction industry, where they use scaffolding to support workers while they build from the ground up. You don’t start at the floor and build the penthouse, ya know?

So when it comes to teaching data viz, we don’t start with a Rockstar Level 10 Sankey Diagram. We meet people where they are.

Here’s how this often plays out in our Graph Guide program.

Students know they are going to blast through 50 graph tutorials in a year. After that, you are expert. Or, as one of our students said:

Many of our students use their year in the program to springboard from a software they know and love to something new and fresh.

Marianna is one of those students. She was super comfortable cracking out basic charts in Excel. But she wanted to get even better at it AND she wanted to broaden her skill set into Tableau.

My organization was signaling an interest in Tableau and I wanted to be prepared in the event that they did go through with it. I am glad I started learning to use it because they did in fact roll it out and I was given a Tableau license. I would have been completely lost without Graph Guides. Tableau is an amazing tool but after using Excel for so many years I found the switch tricky at first. Now I am hooked and feel really confident using it.

– Marianna

She’s right! Tableau IS tricky at first! So when I was creating Marianna’s custom curriculum, I didn’t drop her into the deep end. We spent the first few months maxxing out Marianna’s Excel skills. She learned new graph types to make there, rocked out some excellent visuals, and even learned a wee bit of code, with me coaching her along the way.

Then, after she already knew how to make a dot plot in Excel, we scaffolded. She watched our Orientation to Tableau video (just this lowers your blood pressure significantly), tried out a bar chart, and then created that dot plot that she already knew inside Tableau.

When we scaffold learning, we build off of existing knowledge so that the new thing we are teaching is way less intimidating. It becomes accomplishable.

Marianna eventually swam her way to the deep end, just with structured support that makes her ultimately confident and independent.

Near the end of her year in the Graph Guide program, Marianna used her increasingly impressive Tableau skills to knock out one more tutorial and show off her progress. She made this lollipop gantt chart to show how she blasted through her customized learning path. She began last May, so you can actually see her start out making quick gains in Excel (and some Word/Powerpoint related lessons). Then she jumps into Tableau and, at first, each new chart takes a long time to finish. Her pace quickens as she gains skills. Then it slows down again as she moves into more complex Tableau tutorials.

And now Marianna is bilingual. She switches to whichever program makes a certain chart easier to produce and her design skills are now so strong that no one can even tell which program she relied on. That’s amazing.

Of course, any of our Graph Guide students could have continued to rely on google searches to help them make progress with their graphing skills, playing the hit or miss game forever. It’s just that when you commit to a program with scaffolded support from people who know both education and data visualization, you go further, faster.

Skyrocket Dataviz with Graph Guides!

Graph Guides is a custom-built, year-long sprint through 50 Academy tutorials.

When you enroll, we’ll assess your current data viz skill set, build you a customized learning path, and hold your hand as you blaze your way to new talents.

Enrollment re-opens Spring 2022

Friends Don’t Let Friends Use Stoplight Color Schemes

A google image search of “data scorecard” turns up these results:

Full, just FULL, of stoplight color schemes. Friends don’t let friends use stoplight color schemes. This is the tiny hill I am willing to die on. So Friend, let’s talk about why this has got to stop. There are three solid reasons, each of which, on its own, is more than enough to get the red-yellow-green scheme uprooted.

1: It is not color-blind friendly.

The most common form of color-blindness is red-green. Which means that when we use the stoplight colors to communicate that red means “bad” and green means “good,” people who are red-green color-blind literally can’t tell if you are talking about the worst stuff or the best stuff.

That first scorecard that popped up in my google image search:

looks like this to someone who is red-green colorblind:

That’s a hot mess, Friends. One that will make you non-compliant with 508 accessibility standards and put you at risk for a lawsuit. Even Beyonce got sued because her website was not 508 compliant. And you don’t think you are better than Beyonce, DO YOU?

If the threat of a lawsuit isn’t even to compel everyone to let go of the culture of red-yellow-green, do it just because its the right thing to do for people with this impairment. Have a heart. Geez.

2. It doesn’t work in black-and-white.

In this day and age, do we still need to worry about people printing in black-and-white? We sure do. Funding proposals are still distributed to committees on paper. Boards of directors still review printed board books. In fact, the more top secret the contents, the more likely it is to be distributed on paper. And black-and-white printing is cheap. Those of us old people who still own printers at home are printing (in black-and-white) to get away from the screen in these always-wired days.

Here’s how that scorecard looks in black-and-white:

Can you tell what’s what? Friends, this is another hot mess. The upshot is that when we fix this color issue for colorblind folks, it tends to solve the black-and-white problem, too. We feed two birds with the same seed.

3. It doesn’t emphasize anything.

When color emphasis is everywhere, nothing stands out. It just results in a visual that feels noisy and busy. A better option would be to restrict your color-coding and select only reds or only greens. Anything that isn’t in red, you’ll know is good to go. Or, conversely, anything that isn’t in green needs attention. Pick just one color to get your audience to focus.

In this scaled-back version, I only used red dots to indicate where the target is not being met for the most current month and the year to date. The tiny trendlines will tell you if we were previously higher or lower than the current month. So much more streamlined, way less noisy, works for the colorblind and in black-and-white.

The red/yellow/green color scheme is so deeply entrenched in some company cultures that it can seem like it’ll be around forever. And that suggesting a change is like screaming into a hurricane. But I’m here by your side as your fight for a more accessible change. I’m up on this tiny hill and, Friend, there’s room for you to join me.

The Answer Lies Within

The most annoying Magic 8 ball answer would be “The answer lies within.” Because, it’s like, why would I even be consulting you, Magic 8 ball, if I already knew the answer?

Yet in reality I can’t even believe how often this is the case. People will actually tell me the answer without even realizing it.

Here’s what I mean.

Despite all of my prompting, nudging, cajoling, and demanding, some people still struggle to figure out what to say at the top of their slide. There’s some DEEP unlearning we have to do if we want to effectively tell stories with data.

So here’s my trick to sneak the story out of someone. I ask them:

What’s your goal for this slide?

Amanda Peden even v-o-l-u-n-t-e-e-r-e-d this information for me, when she sent me tricky viz to address in an Ask Me Anything session that I held with the folks at Oregon Health Authority.

Amanda emailed me, “I’m really getting especially stuck on the diagram on slide 4… The goal of slide 4 is to give an overview of how telehealth work is spread across the agency and highlight the type of work that each division does. I also thought I’d highlight where the leadership sits.”

Ok, here’s where Amanda started with her slide:

If you are coming in from the outside, like I am, sure yeah this is busy and confusing. I don’t know what half this stuff means. The quote marks around “Division” are throwing me off.

But if I look at Amanda’s goal for the slide, I don’t even need to know what half of this stuff means. I just need to see that “telehealth work is spread across the agency” and that leadership sits in two of these columns.

This is so easy to fix!

I just replaced Amanda’s old title with the goal she said herself (and I added some highlights to leadership):

Regardless of whether you are an insider or an outsider, you get the point. Because it’s at the top of the slide.

If you are struggling to figure out how to tell a story with your data, reframe your thinking and ask yourself, what’s the goal of the slide? I promise, the answer is already in you.

COVID Data Viz

Star Wars or Star Trek?

I totally wasn’t expecting this MD to pitch me that question on live TV. Good thing I have a teenage son who has spent years preparing me for the right answer.

The docs on the Medical News Network brought me on to talk about which kind of nerd I am and about high-quality COVID data viz.

In particular, I highlighted the work of two clients.

The New York City Department of Health has been producing important briefs on how the experience of the pandemic has been uneven for people of different races. We talked about this hella awesome dot plot, among other visuals that I taught them to create in our workshops over the years.

Their whole report is full of clear charts that make systemic racism evident.

I also showed off the dumbbell dots I made for Novavax. These babies made it all the way to THE Dr. Anthony Fauci and every major news network.

But more than just lightly bragging on air, we talked about why these particular charts work and what stories they can effectively tell.

Take a few minutes and watch – you’ll learn something. The least of which will be: Star Trek or Star Wars?

Will Amazon Lift Their Ban on PowerPoint?

When Jeff Bezos said in his 2018 annual letter to stakeholders, “We don’t do PowerPoint,” you could almost hear the stock price of Microsoft sliding downward.

Presentation designers around the world started looking for second jobs. CEOs in other organizations started to follow suit, because whatever Bezos does is gold. But nearly everyone who has ever sat through a corporate meeting was likely sighing with relief.

Let’s face it: Most PowerPoint presentations are painful to sit through. They tend to cause confusion and waste everyone’s time.

However, when a PowerPoint is done right, a good presentation can actually help organizations make bottom-line business decisions and get to market faster, unearthing good ideas and halting bad ones. So what Amazon got strategically wrong was that they blamed the tool — PowerPoint — when the issue was how they were using it.

In his letter, Bezos explained that instead of a PowerPoint presentation, everyone would silently read a six-page written memo at the start of the meeting.

But Bezos conceded that while some memos Amazon employees wrote have “the clarity of angels singing,” others “come in at the other end of the spectrum.”

Just like PowerPoints. The tool isn’t the issue.

Writing a clear memo isn’t necessarily easy, either, Bezos said. In fact, it may take a week or more, he said. And the same goes for a good PowerPoint presentation.

I’m hoping that the new CEO, Andy Jassey, can recognize that:

Well-designed slides might take more time but they ultimately lead to clearer business decisions and more money in the bank.

(Not that Amazon needs more money.)

So what makes a better PowerPoint presentation? It isn’t slapping some tables on a slide, which is what I usually see when consulting with corporations about their data communication culture.

This example about Amazon came from an article published in The Street using information from MWPVL International:

Tables are difficult to mentally process. It takes an incredible amount of cognitive energy to compare the numbers from different cells and paint a picture in one’s mind’s eye about the differences and magnitude of change.

This is how people end up spending half the meeting trying to sort out what they are looking at, which can delay critical decisions and squash healthy conversation.

Instead, to be better understood, the data in those tables need to be visualized as a graph. But graphing in an Excel-default style almost guarantees extra clutter and confusion.

Here’s another graphic The Street used that not only violates accessibility standards around red-green colorblindness, but it uses more labelling than necessary and doesn’t have a clear point.

The lack of clarity is probably the largest contributor to unhelpful PowerPoint presentations and is the biggest reason as to why most people are glad to see them go.

People are on the lookout for the big idea. They are attending the presentation to learn the speaker’s insights, which need to be readily apparent. People are so tuned into trying to find the big idea, they’ll create one where one may not have been intended.

Your audience will look at a slide for about 3 seconds, asking “So What?” The answer needs to jump right out, before we waste precious meeting minutes stumbling to guess at the point. Here’s a clearer and more visually pleasing way to remake this slide:

As I explain in Effective Data Visualization, high-impact data visualizations are the result of small but important tweaks.

Here are my four rules for making better PowerPoints.

  1. Say Your Point
    I gave the whole visual a more insightful title – one that will focus the conversation in the meeting and get decisions made.
  2. Pick the Right Chart Type
    An area graph is a more appropriate chart type – one that better communicates a trend over time – than a column graph.
  1. Use Smart Colors
    I used Amazon’s orange, rather than red and green and removed the silly beveling on the columns.
  1. Reduce Clutter & Add Emphasis
    In the previous graph, I assumed there was a point to the color change for 2015, so I indicated it in a better way with a marker and a label. I removed the rest of the data labels from the graph and I took out every other year in the x-axis. I even changed to font to one a better, non-default one. 

The original visual was probably made in less than 30 seconds by clicking on default buttons in PowerPoint.

Sure, it took me a few minutes longer to generate this improved example. But as Bezos said, high-quality presentations take time. The payoff for the investment of your time is decision-making that is clear and efficient, high-quality and fast. And that’s what Amazon is all about.

Don’t give up on PowerPoint. Just use it properly.

The Missing Part of Style Guides

Half of the people in my data visualization workshops don’t even realize there’s a style guide lurking around somewhere in their organization, but chances are the Communications department has been begging people to use the style guide all this time.

One big reason more folks don’t adhere to the company style guide – or even know that it exists – is because it doesn’t have all the guidance they need.

Style guides are a gift – the people who put them together have already made some difficult decisions that guide your graph-making. This is the document where you’ll find the exact colors you should be using in your charts. If you’re lucky, you’ll also find details about font sizes. Awesome! This takes out the guess work and makes it such that everyone’s viz looks the same. And THAT brings you cohesion and consistency.

BUT – most style guides don’t adequately address data visualization.

The data visualization component of a style guide should include the chart types you should use for the common datasets you work with and how those chart types should be formatted for brand consistency.

Here’s a visual of what I mean:

The more you can be reallllllly specific about how the charts should look, the more that everyone’s chart will look the same. Brand consistency!

Even better is if you can construct the example charts so they are templates, essentially finished, where a team member can just edit the data and be done.

I’ve ushered in data visualization style guides with many organizations and let me tell you: It takes team work.

The best results come when the data people and the communications people work together.

Because I often see conflicting needs that the other group doesn’t know about – until they talk and align.

For example, the Comms team usually dictates a primary font, usually something custom, and a backup font, like Arial. But the data people who follow my advice know that our graphs are best when they use a condensed font. When data people don’t see a condensed font in the Comms style guide, they go rogue and pick whatever they like. And then there isn’t brand consistency and y’all don’t look like you have it together. Because you don’t.

Instead, the data people gotta sit with the communications people and ask for a condensed font that cooperates with the other fonts in the general style guide.

Comms people don’t typically work with graphs so they won’t know how the decisions around fonts, font sizes, and colors impact dataviz unless the data folks speak up.

It helps if you can see some models to follow. I look to Amy Cesal for guidance here. She’s developed many data viz style guides that are publicly accessible. Read about her work here and check out the collection of data viz style guides here.

The Number One Thing Holding Back Great Data Visualization

Once a season I’ll get an itch to overhaul some part of my house. It’s an old house, so it usually happens that in the process of fixing one problem, we identify another that will send me back to my local hardware store and double the length of the project.

As we hauled the paint cans into the bedroom to start the repaint, I spotted pits and bumps in the wall that my perfectionism couldn’t ignore. We needed to patch and sand before we could prime. The project that should have taken a day ended up taking three.

Now, I coulda easily just slapped some paint on that uneven wall and passed the real fix on to the next homeowner. But when you are invested in the quality of the outcome, you don’t cut corners. Instead, you start anticipating that every project will take twice as long as expected.

If you ask your boss if she relates to my story, she’ll probably say yes – so many of us have been to the hardware store multiple times in one weekend.

What many bosses don’t see is that great data visualization is a lot like a small home reno project.

*IF* you care about a quality outcome, you know you gotta attend to whatever is akin to those dents and bumps that need patched and sanded. You’ll wrestle with the cleanliness of the data, the data structure, the incompatibility with your data viz tools, an error in the code, and the list goes on. Am I right?

Most data visualizers have been around the block enough to know that Hiccups Happen (trademarking this). You are wise enough to recognize that the project will take twice as long as expected. It’s just that your boss doesn’t get it yet.

In fact, a survey of 1,766 data visualizers who attended the Outlier Conference held by The Data Visualization Society said exactly this. They were asked “What do you think other people in your organization just don’t get about the data visualization work that you do?” The most common answer, by far, among those who responded to this question, focused on TIME.

Out of the 745 people total respondents, over a third talked about time constraints as the biggest thing they wish other people understood. This was an open-ended question, so the responses were full of gems that articulate this issue so well.

The effort required to create something really well-designed is often minimized. I get requests that folks think will take me a few hours, when in reality it’s 10 hours of work to clean the data, re-design the visual, and correct errors.

Wise data visualizers read that quote and nod their heads in agreement. Trust us, you want those errors corrected.

Survey respondents are obviously dealing with totally different scopes of work, but I saw off-the-cuff time estimates range from the above-mentioned 10 hours to *weeks* of work, either of which will probably blow your coworker’s mind.

Simple visuzlizations [sic] might reflect weeks of work obtaining, cleaning, and analysing underlying data and developing data pipelines

The good news is that the overwhelming discussion of TIME means you are not alone in feeling like you are set up with some unrealistic expectation.

The tougher news is that it sounds like we have to start training our colleagues to anticipate a lengthier lead time.

I love how this survey respondent pointed out that our colleagues might be misled because they think tech makes us more efficient.

I think the time and effort involved in creating a viz is often underestimated because stakeholders assume that viz technologies and tools automate much of the process; as a result, the time spent cleaning data; iterating and selecting a design format; and determining color palettes, fonts and annotations often isn’t acknowledged.

But the tech will not know the data story and the best design format. Great data viz still requires skill, art, and human touch.

Perhaps strength in numbers can help us provide a gentle reality check to those we work with. Time is the number one thing holding back great data viz. You want an awesome dashboard? Don’t ask the night before.

PS. My faaaaaaaaavorite response in the survey was from someone who, like me, is clearly on a first name basis with the folks at their local hardware store.

It’s like painting a room, you need to spend much more time doing the prep work (i.e. data prep, analysis, verification, iterations, etc.) than you actually do painting (i.e. final viz) to get a great result.
From the blog