[MUSIC] Hi, as we have seen from the beginning of this course, simple forms of visual encoding can be used to tap into both systems one, and two. Although they are a powerful combination for enabling analytics with data, our perceptual and cognitive systems can also be mislead by visualizations. Even if that is not the creator's intent. As showed you through the Fox News and Florida gun visualizations in module two. As visualization designer, you need to recognize how to leverage people's perceptual and cognitive strengths for analytics, and also be aware of how these systems can be fooled. Depending on your design choices, the exact same data displayed in somewhat different ways can sometimes lead to dramatically different conclusions in your users mind. On the other hand, your users may have mental models or biases that may prevent them from interpreting data accurately. And part of your job is to help them break through those mental roadblocks. Now, I want to tell you a little story that illustrates the point I'm just talking about. In World War II, planes like this, the 24 Liberator, were being shot down at a very alarming rate. So they decided to protect the planes with more armor, which also means adding more weight. So they had to be very careful about where they placed that armor. One thought is to go and look at the planes that have come back, and see where the bullet holes are and then cover those places in the wings and other areas where the bullet holes are. That's a great thought, but it's also not the best approach. Can you say why? Well, it turns out that if the plane made it back with bullet holes in it, it made it back. So it survived. So by definition, it was able to withstand the damage to those areas. The damage to the areas of the planes that didn't come back was the catastrophic damage. And that's what needed to be seen and understood. That it's not what came back. It's what didn't come back that mattered or both. And so that's called survivorship bias. How can a visualization help people naturally inclined to think one way, think the reverse about a problem or consider the data in a different way? Let's take a look at how that can be done. Perhaps you've heard the phrase seeing is believing. Well, sometimes the reverse is equally true. What you believe can influence what you see. Although systems one and two, which you've learned about in previous lessons, perform a powerful combination for enabling analytics with data. They can also be mislead by a certain design choices. Even if that's not the creator's intention. Additionally, people's own beliefs, biases and mental models, can distort their interpretation of a visualization without them ever realizing it. Data visualization should reveal meaningful patterns as accurately and clearly as possible and ideally, let the data speak for itself. Part of your job is not to introduce distortions to a design and ideally help prevent people from unwittingly coming to false conclusion based on their own sets of biases. Your job as a visualization designer is to find ways to help clarify the data. So that the users or audiences can look at it and come to the correct conclusions. Just having the right data is simply not enough, and frankly even presenting it in a correct form isn't always enough either. It's really thinking about where users are coming from, how much knowledge they have. How good they are at statistics, and making design choices that will speak directly to them in the simplest most straightforward and complete way possible. Of course, in order to make your case, you need to capture the audiences attention. Let's take a look at a few examples. A multi-talented woman named Florence Nightingale, who gained fame in the 19th century as a nurse and social reformer, was also an early experimenter in statistical graphics. She wanted to move her audience which included Queen Victoria. With a data visualization that would prompt change in the health care system. Her goal was to, as she put it, affect through the eyes what we fail to convey to the public though their word proof ears. To do this, she tried various approaches. One is called the Bat's Wing Diagram. It shows the mortality rates of British soldiers from various causes over two years during the Crimean War. The main point of this visualization was to show that soldiers were dying from factors that were non-battle related as a much larger rate from the actual fighting. The circle on the right is divided into 12 sections, each representing 1 month of the year in a clock-wise manner. The length of the radial line in each month is proportional to the death rate, but there's a problem here that Nightingale caught and addressed quickly with subsequent redesigns. Can you explain what that problem is? Well, the gives the impression that the shaded area is proportional to the death rate rather than the actual case, which is the length of the radial lines, which makes this truly a gray area for visualization. Now, as I said, Florence Nightingale caught the problem and redesigned the diagram to make it more accurate. In Hera-Vision, the areas that are encoded in a blue-gray color represent the mortality rates from non-battle related causes. While the reddish-pink show the battle related mortality. The death rate is proportional to the area, rather than the length of the radii, as in the earlier back screen version. Now this is a very interesting and provocative visualization. And it continues to be controversial today. It may not necessarily be the best way to represent the information for everyone. There are certainly bar or line charts that could do a better technical job but with visually arresting design Nightingale hoped she would get through to her audience in a way that the more conventional chart types may not have done. It appears that her strategy and acceptance of the trade-offs may have worked and they downsized for a less critical in this particular incidence. Now there's much more to this story that goes beyond the scope of this lesson. That I'd highly recommend you read and think about. You can find references in the resources area. We've seen how design choices can create distortions and understanding. But they can also help counteract potential cognitive biases in your audiences, as well. As a visualization designer, you might consider and try to anticipate potential biases, especially common ones. And when possible, counteract them. Sometimes what might seem intuitive, can just be a matter of bias. Recall in the introductory video of this lesson. The survivorship bias example with attempts to protect Allied Aircraft in World War II. The key in that instance was to protect certain areas of the plane that were showing damage in the returning craft. The problem was they were forgetting about the damage on the ones that were not returning and that damage was really critical. How might you design visualizations that counter this tendency? Well, one approach is to use multiple views of the data to show what is known about damage patterns for returning planes. But also suggest and show areas of potential damage from the ones that did not return. The places that were not shown in the returning planes. At least somehow get their users to think about that issue. In doing so, you get to think about the problem in a different way, and that different way may make the difference between life and death. Let's examine another scenario. Nancy has just been informed by her position that she has a positive mammogram, but a positive mammogram doesn't necessarily mean Nancy has cancer. And she is wondering what exactly her chances are. The physician believes her chances are around 80%, but she's incorrect. Why is that? Well, because of the positive result, the doctor is already unconsciously tending to lump her patient into the cancer category. As a result, she's only selectively recalling a study about this issue, and in specific, only the first column of the study. So when that comes to mind. That is the column that people who have breast cancer. So with the people who have breast cancer, the number of people who have a positive mammogram is eight, whereas the people who have a negative mammogram is two. So eight out of ten or 80 % have breast cancer. The problem is, however, the doctor is not considering the other column of people who have had a positive mammogram and do not have breast cancer. Which is in fact quite a substantial number. Here's what the numbers look like in a simple diagram. As you can see by looking at all the numbers in this context and arranged in a certain way, that out of 103 people who tested positive from the mammogram, only 8 have cancer and 95 do not. It's all the same information, but it's presented and arranged in a different way. That makes it, hopefully, clearer that the chances that Nancy actually has cancer are about 8% not 80%. While 8% is not trivial, it's certainly much better than 80%. The point of all your efforts as a visualization designer is to transform raw data into clear, accurate, and meaningful insights in people's minds. There are many ways as process to go right or wrong, but ultimately the final transformations occur in your audience is minds. Your design choices from basic visual encoding options to anticipating encountering potential misinterpretations can all have an important impact in the success of your work. Well, thanks for joining me in this module on design for understanding. As we've learned, the human brain is hardwired for visualization. And there is a visual language made up of elements such as color, line length, shape, and interactivity that can be combined to communicate meaning in your data. Often is the expert use of these basic visual design elements that make a visualization go from good to great. It's important to fully consider the needs, abilities and your visualization's audience before deciding which design elements will suit them. Thanks very much for listening. Goodbye.