Triple Crown Framework for Data Visualization: Psychology

Use Psychological Schemas to Help Your Audience Process Your Visualization

Ryan explains psychological schemas and shows you how to tap into your end users expectations. You’ll see how spatial context and colors help end users read visuals faster or how they will create a starting point for an analysis.

Hi, this is Ryan with Playfair Data TV, and in this video, I’m going to be discussing my newest strategic framework, which I call the Triple Crown Framework for Data Visualization. These are the three principles that I just always have in the back of my mind whenever I’m doing a data visualization, whenever I’m designing something in Tableau Desktop. The genesis for this framework came from my experience with the Iron Viz Championship. If you’re not familiar with that contest, it’s held live at Tableau’s conference each year between three finalists. They’re up on stage in front of thousands of people. They’re each given the same dataset, and then they have 20 minutes to create the best visualization that they can.

The reason that this framework is rooted in that experience is I came to realize that almost every finalist had a very good balance of data, and then design to make it more engaging. But each year, what fascinates me is the three finalists always come up with something completely different between the three of them. It’s the exact same dataset, but all three of them have a very unique take on how to analyze that data. So I knew that there was some third ingredient. Something was missing that kind of put the winner over the top. They’re all very good with the data. They have started to realize we also need to make this engaging and balance in design, but there was some missing ingredient in there.

And that’s where that third and final principle comes from. So it starts with Data. We all know we can’t have data visualization without data. Then we’re balancing Design. It’s been also really interesting in my career to kind of watch the evolution of the adoption of design principles in data visualization. I’d say when I started a little over 10 years ago, that design was not really cared much about, especially from high level stakeholders. They really just wanted the facts. Just get straight to the data, even if it was just in a raw text table.

Today, I’d say we’re at about 50/50, where people are realizing it’s not enough just to have a raw text table. We need to make that data engaging in some way so that people want to look at it and act upon it, do something with it. And then the final ingredient that I came up with was Psychology. And even though I thought of this last, I actually believe that it comes first.

I discuss some of my psychology tips in other videos, so I’m not going to reiterate them here. But just to give you an example, it’s critical to understand who your audience is before you even open Tableau, before you do any data engineering, because your audience is very much going to inform the types of questions that you’re going to be asking and what your final product is going to look like.

So the reason I actually think psychology happens first or should be first in this framework is because of that focus or what should be a focus on the audience. That happens first before we get to the data or design. I’ve got another tip here at Playfair Data TV, which is to explicitly introduce the value of data visualization. I show you my exercise to do that in the video, Why Do We Visualize Data?.

So I won’t go back through that right now, but the reason it’s in this psychology section here is because that’s very much a psychological thing that happens when someone sees a raw text table versus a highlight table. Whether they like to admit it or not, something just kind of clicks, and they realize the value of data visualization.

What we will cover on this video is a principle that’s very specific to the science of psychology, which is called schemas. Schemas are patterns or expectations that we all have. We actually have thousands of these that we’ve developed over our lifetimes. They help us learn. They help us process the world around us, and they kind of help us standardize society, so we can navigate different situations very quickly.

A basic example is when a child learns what a horse is. They will commit some of that animal’s attributes to their mind. They’ll say, OK, I see a horse. I know it’s an animal. It’s hairy. It’s got four legs. It has a tail. So now they think that is what a horse is. The first time they come across a cow, they might mistake that cow for a horse, because it fits their expectation or those attributes that they had in place for what a horse was. After all, a cow is also an animal. It’s also hairy, has four legs, has a tail.

So you can also adjust these schemas over time as new information is introduced to you. The example that I like to share as an adult and why I have this animation that represents this topic on the screen here is when you go to a restaurant, even if you’ve never been to that restaurant before, you have an expectation on the order of events. You know how to go to a restaurant. You’re going to walk in and be greeted by the host. The host is going to sit you down. The waiter is going to come by, take your drink order, leave for a little while while you look at the food options. I don’t even need to finish this pattern. Everyone knows how to go to a restaurant. That’s another example of a schema.

If I were to walk into that restaurant, and the server showed up and says, here’s your ticket for the night, “How will you be paying?” I’d be very confused. That would disrupt my expectation, so much so, in fact, that I wouldn’t even know how to handle the situation. I argue that this concept of psychological schemas can help your data visualization in two major ways. To help illustrate, I’m going to show you an example from my Tableau Public Portfolio that looks at the lowest ticket price per section during Super Bowl 50.

This is a bar chart, which happens to be my favorite chart type. It’s even sorted in descending order, so it makes it even easier to analyze the length of these bars. If I were an employee at this stadium or maybe a big fan– by the way, this game was at Levi’s Stadium in Santa Clara, California. If I knew what those section names were, this chart would be even better. This is actually a really good chart, but I don’t work at the stadium. I’ve, in fact, never been to this stadium, and I was designing this visualization for a mainstream Tableau Public audience. Most of those people have never been to the stadium. Most of those people are probably not data people and especially not Tableau users, specifically.

So I wanted to try to make it easier for my end users to analyze this data. So I took this same data, and I mapped it on to Levi’s Stadium. And now even though I have never been to this stadium, I can leverage my schema, my expectation from having attended dozens of sporting events in my lifetime to be able to read this. As a sports fan, I would think that as I get closer to midfield and into those lower rows, I would expect those tickets to be more expensive.

As I move up and into the corners, into the nosebleed sections, I would expect those tickets to be less expensive. And sure enough, that’s exactly how the data played out when I mapped it onto the stadium. As you can see here, the lower rows, midfield game– midfield tickets, those are the most expensive. Up in the corners, the higher you get, the lower the ticket price.

So the first way that this concept can help you in the practice of data visualization is it can make it more efficient or faster for your end users to process. You can tap in and kind of reinforce what their expectation was. It’s going to be faster for them to read it. The other benefit to using these schemas is if there is ever a disruption to an expectation, what that does in the context of data visualization is it creates a starting point for an analysis.

If I were to map this data from the first chart onto this map we’re seeing on the screen now, and one of the corners was bright red, so the most expensive seat in the stadium was, in my opinion, the worst seat in the stadium, that would surprise me. It would disrupt my expectation or my schema, and what it would do in this case is it would make me want to dig deeper. I would know that there is something different about that section.

There’s no way that top corner could be the most expensive seat. Maybe they had some special where if you bought a seat there, you got season tickets for the next year. Or maybe there was some type of VIP party, or maybe you get a free car. Who knows? It’s the Super Bowl. But I would know that something different would have been happening if there was that much of a disruption to what I expected.

I also want to point out that I’m not arguing to do a crazy custom polygon map instead of a bar chart. What I would recommend is complementing those traditional chart types with something that kind of taps into your end user’s expectations. And in fact, in the real life version of this, there is a line graph underneath the stadium map. So I’m combining the two. The first one helps me quickly process it. I can then click on a section name or section, and then see the 14-day trend below it as a traditional line graph.

By the way, I used Tableau dashboard actions to accomplish that user experience. So I’m not arguing one or the other, but I do think that these two things work really well together. I also realize that this was a little bit of a radical example, so I want to share one more schema that I think will hit home for everybody watching this. And it relates to color. I bet for better or worse that everybody watching this right now has an idea of what these two colors mean. Green’s good. Red’s bad. For whatever reason, I personally don’t remember a single person ever telling me that green means good, and red means bad, but that is what I associate those colors with.

That’s another example of a schema. So at a minimum, because this one’s familiar to all of us, be aware that your end users do have expectations about what certain things mean. Don’t use green to represent bad and red to represent good. It will disrupt the schema, their expectation of what those mean. It will make it more confusing for them to analyze the data that you’re using.

This has been Ryan with Playfair Data TV – thanks for watching!