Join Bill Shander for an in-depth discussion in this video Sketching and wireframing, part of Data Visualization for Data Analysts.
- Sketching and Wireframing is one of the very first things you should do in your process. It's really important. But, it's also the thing that data analysts are probably most reluctant to do. Does this sound familiar? You have some data, you do some analysis, and you're analysis includes looking at the data and a few different charts. So you use visualization as part of your analysis process. By the way, that's a very good thing. I'm sure you all agree that visualization is a critical part of the analysis itself. Most of you maybe already know about Anscombe's quartet.
Francis Anscombe was an English statistician who took four data sets with nearly identical properties. Each had 11 data points, and the same mean, and the variance of both x and y. The same correlation between x and y, as well as the same regression line. But, as you can see, by visualizing them, you can easily see how drastically different they are. This was a bit of a tangent, but let's go back to where we left off. You're doing visualization as part of your analysis. That's good. So, you finally find the chart that shows what you needed, and confirms or demonstrates your hypothesis.
And you've confirmed that it's not your bias that's getting you here, right? You're a good analyst. So what do you do? You know you need to make a PowerPoint slide about this. So, you export that chart, and you're done, right? I'm sure that is right, meaning I bet it's how you tend to do it. But I'm gonna suggest that it's wrong. You need to do what I suggested at the beginning of this section of the course. First, you need to come up with a story for that slide. What are you trying to explain? What's the beginning, middle and end? Then, think about your headlines. So, in this example, there's a positive correlation up to a certain point, at which point you get diminishing returns, and the correlation reverses.
So, something interesting is happening when x reaches 11. Aha! So our headline could be, "Correlation reverses and accelerates downward after 11" right? This is sort of like The New York Times version of your headline, very literal. Maybe a little bit more fun, "What goes up, must come down, dot, dot, dot, after 11" right? That makes a better headline. Or, The New York Post headline, "11 Condemned as Worst Integer in Study" right? Just a little bit more dramatic. Once you know your story, you've chosen the perfect headline.
Now I want you to sketch some ideas. So, you might ask yourself, "Why on earth would I do that when I have a perfectly good chart that I just created?" The reason is, that you created your chart without the benefit of a story or a headline. That's what we've been talking about. Yes, you can probably assume that the chart form will work for you. Although, I would suggest you consider alternatives at this stage, right? There's often more than one way to show your data, as you well know. For instance, maybe just showing the slope graph of the change and values could do the trick.
And by the way, I created this using a very unresponsive digital tablet in Photoshop. It's an ugly hot mess, right? But that's okay, it helps me get to an idea very quickly. I actually like this chart type for certain uses. I might've chosen this approach for this chart, after making this sketch. But, let's assume you do stick with that original chart type. Use that as your starting point for your sketch. Print it out, start writing on it, or just sketch a mock-up of it, and riff on it, right? Just a messy ugly thing like this one here. This takes 10 seconds to create. Then, start making adjustments to it.
For instance, can you think about how you would label your chart differently than how your software displayed the data? Maybe you add interesting call-outs for your audience. And that says, by the way, something is happening here, in case you can't read my awful handwriting. Can a background color for part of the chart, or a segmentation line, help communicate what you need to say? Spending literally 15 to 20 minutes sketching out your chart in the context of your story and headline will always result in a better visual depiction of the data. Maybe you can only do this for your most important charts in a presentation.
I know time is scarce. But, it's well worth it if you care about how you're communicating your brilliant data analysis. Again, this is about very quickly and easily validating ideas. Seeing if you can bring eyes, and your audience's attention to what's important. Then you can worry about how you can execute on it, in your tool of choice. And by the way, as I mention a lot in my Data Visualization Fundamentals course, I believe adamantly that when I say sketching, I mean you should sketch with pen and paper, or on a whiteboard. If you must work digitally, and that's okay, I do work digitally sometimes.
At least use a tablet and a stylus, or your finger. My point is that, working in an analog way, is often faster. It connects you more intimately to your subject matter, and allows for greater flexibility. If you can, check out that video from my other course and you might be convinced.
- Why visual communications matter, and how they work
- Communicating via story
- Communicating with color
- Using legends and sources
- Sketching and wireframing
- Rethinking slides, charts, and diagrams
- Rethinking your templates and brand guidelines