2024-11-14 05:20:20
Reflecting on my early days at Google, I recall a project that involved analyzing thousands of free-text comments from our internal survey, Googlegeist. The task was straightforward in concept: distill this qualitative data into actionable insights by identifying common themes, categorizing feedback, and quantifying sentiments. But in practice, it was a tedious and time-consuming process, taking countless hours to produce anything meaningful. At the time, we didn’t have effective tools for this. There were options, but they were often clunky, lacked the nuance required to understand context, and frequently missed the mark on detecting tone and sentiment accurately.
Fast forward to today, and the tools for qualitative analysis have come a long way, largely due to advancements in artificial intelligence and sophisticated natural language models. I’m often asked about AI's role in data storytelling and whether and in what instances it can replace human judgment. The answer isn’t black-and-white; AI isn’t a perfect solution, nor does it eliminate the need for human insight. However, it has advanced enough to solve challenges that seemed insurmountable in the past—for example, in the realm of qualitative analysis.
Recently, I saw a demo of a tool created by a friend of mine—also a former Googler—who founded a company called Insight7. As he walked my team at storytelling with data through the tool’s capabilities, I found myself thinking back to that early Google project and how much time it could have saved me. Insight7’s tool automatically categorizes and analyzes text, drawing out themes and insights in seconds. What struck me wasn’t just the speed, but the quality of the analysis. It was a reminder that as the technology around us evolves, so should our openness to exploring new possibilities. Tools like this can augment our work, help us uncover insights we might have missed, and ultimately free us to focus on what we do best—crafting stories and building understanding.
It’s easy to hold onto previous beliefs about what’s possible, especially when we’ve experienced the frustrations of earlier versions. For me, one example is my long-held view of pie charts as suboptimal for most data storytelling purposes. But I’ve found that staying open to revisiting these beliefs keeps me open to innovation (sometimes, to using even pie charts in the right context!). In the same way, we may hold outdated beliefs about AI’s capabilities, but if we remain curious, we’ll be better positioned to make use of advancements that are made.
As data storytelling and analysis in the age of AI continue to evolve, let’s not be limited by past assumptions. Who knows what new possibilities might emerge if we stay open to tools and ideas that can help us tell even richer, more compelling stories with data?
If you’re curious to explore Insight7, they’ve just launched new product features, and my friend has offered a special discount for SWD readers. Use the code SWD40 at Insight7.io to get 40% off all paid plans through November 15. If you work with qualitative data—or have been hesitant due to the time investment involved—this tool could be worth a look.
2024-11-12 04:28:44
After Halloween, my social media fills up with pictures of candy graphs. They’re fun, colorful, and a creative way to introduce kids to graphs using physical objects. Candy graphs can show quantities, allow kids to compare categories, and generally make graphs feel approachable. But despite the sweetness, they fall short in a crucial way.
Candy graphs may look great on social media—bold, flashy, and neatly arranged. But they’re often meaningless, and sometimes even misleading. For instance, there are more Snickers in the picture above than Reese’s cups, whereas it looks like the opposite is true.
September was a nice month here in Wisconsin, but we don’t need a graph to tell us that: we experienced it.
Candy graphs, dice rolls, weather charts—these types of visualizations, while fun, don’t truly tap into the power of graphs. They don’t help kids understand something meaningful about their world or make any decisions based on the information. The problem with candy graphs highlights a broader issue in how we typically teach data to kids. We miss the chance to show them that data can be a powerful tool for critical thinking, decision-making, and positive change.
The best graphs—for kids and adults alike—are those that help us make better decisions, spark thoughtful discussions, or inspire us to improve our habits. But so often, we use examples that are disconnected from kids' experiences. Instead of guiding them to ask questions that they care about, we give them prepackaged data from topics that might be colorful or fun, but ultimately lack meaning for them.
Kids are naturally curious—they love asking questions, drawing, and exploring topics that matter to them. By engaging them in collecting and graphing data on subjects they care about, we’re teaching critical problem-solving skills in a hands-on, interactive way. When we spark their curiosity and empower them to visualize data to answer their own questions, graphs become more than just tools—they become superpowers.
Let’s take a fresh approach to teaching graphs. Instead of focusing on data that’s "nice to look at," let’s focus on data that’s actionable. For instance, I had my kids track their sleep hours and mood each morning for a week, then plot it on a scatterplot. My 11-year old decided he should try to get to sleep a little earlier so he’d feel more energized in the mornings. During a recent school assembly activity, we surveyed kids and then had them graph the results to questions like what the favorite new option for the hot lunch menu would be. These are merely a couple of examples: there are so many questions kids can ask and answer to learn about themselves and have input on decisions that impact them.
Begin by involving kids in choosing questions they want answered. Then help them collect data from their own lives, and let them plot their own points on the graph. This way, they’re not just learning what a graph is—they’re learning what a graph can do.
Join me at the free Make Math Moments Virtual Summit (November 15–17) to explore a hands-on classroom activity on meaningful data collection and graphing. My session airs Saturday, November 16th at 5:30pm ET. You can also listen to my chat with Make Math Moments cofounder Jon Orr in the latest episode of the SWD podcast:
For more ideas and inspiration on creating meaningful graphing activities, check out our Daphne Draws Data YouTube channel, where you’ll find short videos with graphing activities designed for use in the classroom, video read-aloud companions, and more.
2024-10-31 00:25:00
Sometimes, we encounter folks who are reluctant to add words to their data tables or visualizations. This hesitation might stem from having been taught to “let the data speak for itself,” as if data were a wise, all-knowing entity rather than a structured set of measurements. Measurements that, ultimately, are shaped by resource constraints and the choices of others along the way.
Occasionally, though, this reluctance tips into something closer to disdain. When it does, it’s often due to a belief that data analysis is a “hard skill,” while communication and presentation are merely “soft skills”—less worth the effort to develop or refine. This perspective is not only limiting but ironic, especially when we consider the true nature of these skill types.
Generally speaking, “hard skills” deal with things like numeric precision, algorithmic development, engineering, and analytical investigation. They’re areas where your work can be objectively tested, measured, and evaluated. In other words, with hard skills, you’re likely to receive clear feedback on whether you’ve succeeded or failed. You know where you stand.
On the flip side, “soft skills” include things like interpersonal communication, emotional intelligence, adaptability, and collaboration. All of these are essential in any group setting, but you rarely get a clean “yes” or “no” on whether you’ve applied them well in any given situation. Yet despite the perceived ambiguity, soft skills are crucial for effectiveness in the workplace. In fact, they’re often much more challenging to develop, as they require ongoing self-awareness, practice, and responsiveness to others.
All of this underscores that the hard skill versus soft skill divide is an arbitrary distinction that often works against both individuals and teams. A talented engineer who can communicate technical insights effectively to management is just as valuable as a team lead who can discuss SAS with the statisticians. Skills, at their core, are simply skills—tools that allow us to solve problems, create connections, and share insights.
2024-10-11 02:33:00
My fellow data storytellers and I have lost count of the number of times we’ve been asked questions like, “What’s the right number of slides to have in your presentation,” “How many graphs should you have on one slide,” or “What’s the right number of words to include on a single page?”
I’ve worked with many folks over the years who have quantitative backgrounds, and they tend to crave the reliability of specific numeric guidelines whenever possible. In asking questions like these, people hope, or even expect, that we’ll have prescriptive and undeniable numeric answers foro them. If you do too, then unfortunately you’re not going to like what I have to say next.
It’s time to throw another couple of dollars into the “it depends” jar, because like so many other questions in communication and data visualization, the so-called “correct” answers to these specific questions will vary depending on context.
Nearly 20 years ago, the “10/20/30 Rule” for PowerPoint presentations became popular:
Use no more than 10 slides.
Speak for no longer than 20 minutes.
Only include text with a font size 30-points or larger.
This pithy, memorable, almost magic-sounding guideline has persisted ever since. After all, for self-identifying “hard skills people,” this rule provides an objective metric that can be applied to a “soft skills” task that feels inherently subjective. (As an aside, the concept of some skills being “hard” and some being “soft” is a pet peeve of mine…but that’s a topic for another day.)
Take note, though, that the 10/20/30 Rule as originally conceived came out of the world of venture capital. Those communications are taking place within a unique and specific context. Given the customs and norms associated with a pitch deck, a VC audience will expect particular structures, formats, and progressions from a presentation, for which the 10/20/30 Rule would be well-suited. But that doesn’t mean we should export these guidelines globally to all of our presentations. Committing to using 10 slides/20 minutes/30-point type in every scenario is to overly constrain yourself.
The motivations and intentions of 10/20/30 are sound: in many cases, our aim should be to make our communications as short and as focused as possible, pairing clear imagery with words written in legible and high-contrast type. This is particularly true when your slides are meant to support your live, spoken presentation.
Take the 30-point rule, for instance. I’ve been looking at computer screens for decades, and as such my eyes aren’t what they used to be. I appreciate it when people use larger text in a slide meant for being projected on a screen. Thirty-point type? That’s fantastic, as a minimum size. If the text is sparse enough, it might even feel too small.
However, the 10/20/30 rule doesn’t limit us to ONLY 30-point type. We can bump up the text size as much as we want. Let’s go wild and use 180-point type.
Isn’t that great? It’s attention-grabbing, it’s legible, it’s dynamic, and it won’t distract my audience if I’m trying to talk while this slide is up on the screen. It supports me without competing with me.
That large, bold type looks even better when I find ways to incorporate it thoughtfully into full-bleed imagery, like in the photo below:
Imagine this slide comes up as I’m transitioning between sections of a live presentation. While I introduce the next topic in my talk—hopefully something related to beaches, San Sebastian, or landscape photography, since that’s what the image depicts—my audience gets to associate my words with a meaningfully related and aesthetically pleasing visual. The picture’s impact is greater for taking up the full slide; a smaller image cut-and-pasted into a default content window of a slide template doesn’t have nearly the same dramatic effect.
The common theme among the examples we’ve seen so far are the relatively short chunks of text we have to work with. In these situations, the 10/20/30 Rule shines.
What do we do, though, when we have more than a sentence or two’s worth of text to deal with? After all, which challenge do we find ourselves facing more often when building presentations: the case where we have too little text on one slide, or the case where we have too much? I’m guessing it’s the latter.
More often than not, the best way to deal with a slide containing too much text isn’t just to shrink the font size until everything fits. The thoughtful and elegant solution would be to edit the copy so that the total word count gets much smaller. What are the chances that every word on a slide is truly necessary, anyway? Editing takes time—and sometimes, confidence—but your audience will appreciate not having to wade through a ton of text. It’s good practice to be concise and be willing to cut words (or graphs, or images) that aren’t essential to your message or critical to include.
Sometimes, though, editing down the copy isn’t going to be a realistic option. For example:
Who would dare take the position that the right visual solution here would be to edit down the Gettysburg Address? Clearly, a different solution is called for. Even if we find ourselves working with a slide that’s somewhat less historically significant than the most famous Presidential speech ever, we still will run up against prohibitions on removing text from time to time. This may be due to compliance or regulatory reasons, or just because the person with authority prohibits it from happening.
When omitting even a single word is non-negotiable, sticking with the “30-point or larger” rule won’t do. Ideally, we want to create something that is more instantly scannable than the block of cramped text above. By “scannable,” we mean:
the text has been laid out so that anybody seeing the page for the first time will know what the key point is;
the secondary topics and key phrases or details are easy to find and place within the larger structure of the document; and
the document should be easy to navigate, such that skipping to and reading any particular section of the document in context should be trivial.
This visual hierarchy can be created with the thoughtful use of varying font sizes, weights, and families; color, position on the page, and other typographical choices also come into play. Look how we can take the same text of the Gettysburg Address, but instead of presenting it as a uniformly difficult to read 30-point block of words, create a scannable hierarchy:
The improvement in legibility and approachability are obvious. How did we subdivide this text into a scannable hierarchy?
EMPHASIZE A PITHY, MEMORABLE PHRASE: The biggest type on the page, 40-point type, goes to the most recognizable line of the Address, “Four score and seven years ago.”
LEAD WITH THE MAIN IDEA: The lines that follow are in 28-point Garamond gray, until the key phrase “dedicated to the proposition that all men are created equal,” which are the same size but set in blue Gill Sans SemiBold.
ELEVATE SECONDARY THEMES: The topic phrases in each of the four following paragraphs of the address are written in solid black and set in 24-point Gill Sans Bold, with the paragraph text that follows in 18-point gray Garamond.
POINT OUT KEY DETAILS: Other important callout phrases are in blue Gill Sans SemiBold.
CRAFT A LAYOUT TO SUPPORT THE HIERARCHY: The introductory sentence stands alone, cutting across two columns of text; each of the four following blocks of text are roughly equal in proportion, and are aligned neatly vertically and horizontally in a 2x2 square.
While there is plenty of text on this slide that is smaller than the 10/20/30 Rule would endorse, the document itself is easily legible and scannable for anyone, regardless of their level of interest in the material. You can glance at it and immediately know it’s the Gettysburg Address from the very first line; you can grasp the narrative structure of the Address by scanning the bolded black phrases; you can get a sense of the “so what?” of the speech by connecting the three blue bold phrases; and you can dive deeper into the actual content.
Sadly, I don’t have a great slogan like “10/20/30” that can apply equally to all presentations. The reality is that the number of slides, the time we speak, and the size of our fonts all depend too much on the exact communication we’re building, and the situation in which it will take place.
Some presentations only need four slides; some need 400.
Sometimes you’ll only talk for a few minutes, and sometimes for a few hours (and if you do, PLEASE take breaks!).
And, yes, while you’ll ideally have large to very large text on your slides, sometimes you’ll need to include smaller font sizes as well, particularly if your slides will be sent around and/or read independently.
In the absence of relying on a memorable one-size-fits-all metric, let’s instead just agree that we should take as much care in designing a text-heavy slide as we would if it had graphs or images…even if they aren’t always going to be as historically significant as the Gettysburg Address.
As long as we use intentional formatting and visual structure to make that text easy to scan and understand, we’ll be more likely to connect with our audience, and therefore more likely to achieve the positive outcome we’re seeking.
2024-09-27 03:13:28
We’ve all heard—or maybe even said—the phrase before, “I’m not a numbers person.” What does this mean? What are the origins of this unfortunate and limiting mindset?
I believe it starts young.
My latest book, Daphne Draws Data was published earlier this month. I’ve been on the publicity circuit, talking with various people about the book for podcasts, live events, and the like. One line of questions that has come up repeatedly is around when children’s aversion to math begins, and what causes it. I don’t know the answer. I suspect that it is the way that math is typically portrayed in our culture: difficult and boring. On TV, in movies, and in books, the kid who likes math is often stereotyped as the glasses-wearing, friendless nerd.
My attention was recently drawn to a book called I’m Trying to Love Math. I had an immediate and visceral negative reaction to this. I fear that a book titled in this manner only perpetuates the misconception that math is something that should be loathed or feared, propelling the common myth rather than debunking it.
I tried to suspend full judgment, thinking that perhaps it was simply an unfortunate title. After doing a little more digging, however, that is disappointingly not the case. On Amazon, the description of the book begins “Do multiplication tables give you hives? Do you break out in a sweat when you see more than a few numbers hanging out together?”
The author’s other titles include I’m Trying to Love Spiders.
Seriously? Math is scary like spiders?
The first line of the book is “If you ask me, math is not very lovable.” The book continues on the following page with, “I know I’m not alone here either. 4 in 10 Americans hate math.” This is accompanied by a pie chart, where an annotation pointing to the segment representing 40% states “That’s like 40%!” (incorrect, it is precisely 40%, but that’s not even my biggest issue).
I’m Trying to Like Math won the Children’s Choice Award, which is an award that school librarians look to in order to determine which titles to carry in school libraries.
I am speechless. (Practically speechless; I’m actually full of words about how incredibly frustrating I find all of this to be.)
While I’ve focused my rant on this particular book, I actually don’t mean to isolate it. What I’m angered by is much bigger than a single book (or even the triad it appears with in “frequently purchased together,” which also includes Math Curse and The Math Allergy). It’s the message that is reinforced all sorts of different ways with children that math is hard and boring and for nerds.
THIS IS THE WRONG MESSAGE.
(I tend to avoid all caps because it reads like yelling, but in this case that’s exactly what I’m going for; you might note that I even bolded it for additional emphasis.)
I don’t know if a single delightful dragon who loves drawing data can counter this. But I’m certainly going to have her try. Let’s celebrate math—encouraging it and the logical thinking and problem solving it builds—as a super power, not demonize it as something we should “try to love.”
