Ep. 79 | Visual Data Disasters with Alberto Cairo
Some people believe that well-visualized data will tell you exactly what you need to know. But there are many more visual data disasters than we realize out there. This week Alberto Cairo shares several examples he has discovered from weather to politics where some people really like what they see – it’s just not what the data is actually saying. How can you avoid this kind of data disaster? What should you do to avoid becoming a victim of it? This week learn the process Alberto Cairo uses to help others think clearly about data and what it should represent. Please help me spread the word about Customer Centric analytics. Rate and review my podcast on iTunes and write to Allison at info@ambitiondata.com or ambitiondata.com. Thanks for listening! Tell a friend!
Read Full Transcript
Allison Hartsoe: 00:00 Hello everyone. This is Allison. I’m away on vacation this week, but please enjoy this very popular episode from last year with Alberto Cairo on data visualization. We’ll see you next week.
Allison Hartsoe: 00:17 This is the customer equity accelerator, a weekly show for marketing executives who need to accelerate customer-centric thinking and digital maturity. I’m your host Allison Hartsoe of ambition data. This show features innovative guests who share quick wins on how to improve your bottom line while creating happier, more valuable customers. Ready to accelerate. Let’s go.
Allison Hartsoe: 00:46 Welcome everyone. Today’s show is about visual data disasters. Specifically, the many ways we misinterpret visual data and to help me discuss this topic is Alberto Cairo. Alberto is the fascinating and frequently funny leader of the visual trumpery tour. When he is not writing books or hosting workshops, he’s a professor at the University of Miami, specifically the Knight Chair of Visual Journalism. Alberto, welcome to the show.
Alberto Cairo: 01:17 Hi. Thanks for having me.
Allison Hartsoe: 01:19 Now. When I was getting my journalism degree, I can tell you visual journalism was not a specialty I had the option to choose. So could you start by telling us a little bit more about your background and how you gravitated to this topic?
Alberto Cairo: 01:34 Sure. It was, it was not an option when I studied journalism in Spain either 20 years ago. I’m originally from Spain, and I studied journalism there. Um, it all happened by happenstance. My original intent, when was studying journalism was to work in radio. I actually did internships in radio stations, the equivalent of NPR in Spain, reading the news in the morning. Uh, but, uh, at some point at the end of my, uh, in the fourth year of my BA, I believe I did, was a professor of mine knew that, uh, I can draw a little bit, I’m not a great artist, but I can, I can sketch things out. And she knew these about me, and she got a request from a local newspaper in Spain that was looking for a journalism student who could act or could work as a reporter, but at the same time, who also could have an eye for design. So she recommended me for that decision and I was hired as an intern in that newspaper with no experience whatsoever in design. I knew, I know nothing about design at the time, but then I started working in these, in these team, in this group, and I started educating myself about the convergence or the overlap that exists between, um, let’s say traditional storytelling or journalistic storytelling and also visual storytelling and how the two of them reinforce each other. And I have been in this field, ever since until these until today.
Allison Hartsoe: 03:04 Uh, just not to, uh, not to out your age or anything, but roughly what decade was that when your boss was asking you to do more visualization work?
Alberto Cairo: 03:14 Oh, this was 20 years ago. I am 43. I began my career when I was, Oh actually 21 years ago. I began my career when I was 22.
Allison Hartsoe: 03:23 Good. So from this point where you started to get much better, you bet, you self-educated, and you started to understand more about the crossover between journalism and data. How did that lead to your eventual position as the Knight Chair at the University of Miami?
Alberto Cairo: 03:42 Well that, that was, again, it’s a very long story, but I will try to make it short. So, uh, when I, when I started doing graphics to inform the public, I used to focus mostly on pictorial explanations. We used to call those infographics and then using the street. So imagine for example, that there is an accident happening, you know, in town and you do a visual explanation of how the, how the, how the accident happened, right? You draw the cars, and you draw the truck. Just to explain how the accident happened. We call that infographic in the news industry. I prefer to call it pictorial visualization or pictorial infographics. But then when I was like 10 years in my career, like this happened around 2007, no, actually before that 2005, 2006, I started getting interested in graphs and maps and data charts. And I started educating myself a little bit about cartography, learning about statistics and analytics, et Cetera. So I started reading a little bit about that and also practicing teaching myself the tools of the trade. And I gravitated more towards these fields a little bit later, around 2008 2009, a when I started writing my first book, the functional art and also the second one, the second one mostly eroded around 2011 2012 the truthful art, which focus exclusively on data visualization.
Allison Hartsoe: 05:07 And I should also call out that there is a website, I believe it’s the functionalart.com. Is that right?
Alberto Cairo: 05:12 Yeah, that’s my, that’s my personal web blog, which is devoted to visual communication in general. I tend to focus a little bit more on data visualization, but I sometimes also write about pictorial visualization, which is something that I still teach. I still teach people how to use, for example, 3D modeling and animation to the, to tell the stories, right? Uh, but I focus most of my career at the moment, most of my teaching on data visualization. But I teach both data visualization and pictorial visualization.
Allison Hartsoe: 05:42 Got It. So a lot of the folks who listen to this podcast have a background in marketing and marketing analytics, and there is a big push about pretty pictures. Tell me maybe a little bit about why should I care about what is conveyed behind those pretty pictures, you know, maybe shouldn’t I just be happy that I’ve created something that’s visually pleasing and that, you know, people are starting to talk about it. I’m starting to get uptake on the work that I’ve done. Should I care that there might be visual mischief in that infographic or in that visual presentation?
Alberto Cairo: 06:18 Absolutely. So creating a beautiful picture is a worthy goal. I’m not against it. I’m a fan of, you know, nice looking maps and charts and graphs, et cetera. Although I think that that is just one of the dimensions that need to be considered when we design a data visualization. In my second book, the truth for our, I say that any great graphic is made of at least five different elements, or it should pursue five or six different goals. And the first one is, one of them is that it needs to be beautiful. Right? So I would graphic is always a beautiful object that you can look at and enjoy, right? That’s one of the components, one of the elements of a great graphic, but the first one in the list of a thing that we should pursue is that any graphic should be truthful. And by meaning, by, by, uh, the meaning of truthful in there is that the graphic should reflect our best understanding of what the truth that hides behind the data are. It should be the best representation of those stories that sometimes hide behind numbers. And is a primary goal. So a beautiful graphic that is not truthful will never be a good graphic. In order for a graphic to be good first of all, it needs to be truthful. It needs to be honest. It needs to be deep enough to, to, to show us the realities that hide behind those numbers.
Allison Hartsoe: 07:36 And what are the other elements in those five?
Alberto Cairo: 07:39 Oh, I, I talk about many of them. So I, I say that a graphic needs to be truthful. The second is that it needs to be beautiful to attract people’s attention. I also say that it needs to be functional in the sense that the way that we shape the data needs to depend on the messages that we want to communicate. And I devoted a lot of pages in the book to explain for example, how to decide whether you need to use a bar graph or a line chart or a data map, et cetera. So each one of those ways of representing data is very suited for a particular goal. So sometimes, for example, when you have the example that I put, one example that I put in the book and I was explained to my students is that most beginners, uh, when, when, when they started doing, uh, an infograph data visualization and they are working with a data set that has a geographical component, let’s say that is for example, you know, state-level unemployment rate, they rush to the computer and they design a map immediately just because the data set has a geographic component.
Alberto Cairo: 08:40 And when I recommend people is like, wait for a second, stop for a second. Think about the nature of the data. And more importantly, think about what it is that you want to communicate, right? Because if the purpose of the, of the chart, of the visualization that you are doing, is to show geographic patterns in the data, more unemployment here, less unemployment over there, then the map is the right solution. But if the purpose of the graphic is not to show the geographic patterns of the data, if the purpose of your graphic is to let people rank and compare the states in the United States according to unemployment rate, then the map is not functional for that purpose. You need to do some sort of bar chart, for example, to compare the different, the different unemployment rates. So that’s what I mean by functional. It’s like the shape of the object.
Alberto Cairo: 09:26 It’s like the default form follows function, right? The ruling graphic design or although it’s a much more complicated than that and I spend a lot of space in the books explaining these, it all boils down to that. It’s like the form of a graphic should somehow be constrained or be guided by the purpose itself by the function that that graphic has. Then I also say that a graphic a good visualization needs to be insightful, meaning that it needs to provide interesting insights from the data. It needs to reveal patterns and some trends that are relevant and then that may go unnoticed if you don’t visualize them. And as a result of all these, a great visualization can be enlightening. It may be, it may change your mind about a particular topic, right? You didn’t know something about, about these data set. Then you visualize it, and you learn something about the data, and in that sense, the data or the visualization of the data is enlightening you. So those are the five elements.
Allison Hartsoe: 10:28 Now, I actually had four elements as you were talking. Did I miss one? I got a truthful.
Alberto Cairo: 10:32 It’s truthful. I May, I may need to consult, but I believe that you said truthful, beautiful, functional, insightful, and enlightening. So those are the five. Those are the five requirements for a great visualization effect.
Allison Hartsoe: 10:47 I love it. Okay, so tell us a little bit more about what is the problem now? You know, why the need for the visual Trumpery tour and, and I got to ask you, you know, the name trumpery there it seems like there’s a, a political angle in that. Can you, can you tell us a little bit about the tour that you have going on?
Alberto Cairo: 11:09 Sure, sure. When I explained at the beginning of the, of the visual trumpery talk, um, is that the title of the talk is a provocation because a better title for the talk would be how charts lie, or how visualizations lie or how visualizations mislead us. Right. Um, the thing is that, um, a, if I title at talk how Charts lie that is not as attractive or as eye-catching as Visual Trumpery. Now visual trumpery the title comes, uh, from, from a moment of revelation that I had after the 2016 presidential election when I was on Twitter following the results of the, of the November election and a, a person who I follow in social media and social media tweeted the meaning of the word trumpery because Trumpery is an actual word in the English language. It’s an old, old word that comes from French and a trumpery is something that lies something that deceives the eye.
Alberto Cairo: 12:08 So I thought this is perfect because it will help me attract more bigger audiences. It will help me also mislead them a little bit and make my first point in the talk, which is a chart or by extension, any piece of information will always mislead you if you don’t pay attention to it. And if you don’t read beyond the headline, if you read the headline alone, visual trumpery, you may think, well this is a highly partisan talk, right? But then you access the content, you see the actual content of the talk, and you will notice that the talk is, as I said in the talk itself, it’s very political, but it is not partisan. So I believe that the using charts well is a political issue in the sense that charts can inform public discourse or public conversations if they are well-designed, but the talk is not partisan in the sense that I have examples both from the right and from the left.
Allison Hartsoe: 13:05 Let’s talk about some of those examples? And I just, I love the name of that tour. That’s fantastic. And I, I, I think I know a fair amount of, um, unique words, but that was one I had not been familiar with. So thank you for adding to my lexicon there. So let’s talk about that political, uh, angle, the political party side around visual trumpery and obviously there’s such a, an interesting angle that could be played from both sides. What is it that the political parties are doing that is causing us to be, you know, perhaps misled or using the data incorrectly?
Alberto Cairo: 13:41 Well, it’s not just political parties. It’s, it’s partisan people on both sides of the political spectrum, right? So, uh, both in the talk and in the book that I’m writing around the talk, I explained that these, uh, the misuse of charts is not related, uh, to, uh, you are being on, on, on either end of the political spectrum. It’s more related to the fact that you are partisan and that we all like to see, what do we want to see? We all like to have our opinions confirmed, right? And corroborated, right? When we see a, yeah, when we see a chart that apparently confirmed for corroborates what we already believe, we feel really happy. Right? And when we see a chart that refutes what we believe, we tend to reject it rather than reading it carefully. Right? So I have examples from partisans, from both the left and the right and up and down from all over the political spectrum because this is a universal problem.
Alberto Cairo: 14:40 Uh, beginning on the left for example, they have an example coming from liberal pundits who a while ago were trying to defend Obama care, the affordable care act, using a chart of in the job market number of jobs created by private companies, right? If you plot the number of jobs in the private market between 2006 and 2000, I don’t remember the details, but imagine that it’s between 2006 and 2017, the shape of that curve will be, we’ll have the shape of a U, alright, it has a U shape, meaning that jobs are started, the creation of jobs started shrinking or dropping, uh, during the economic crisis. And then there is an inflection point, and then the job market recovered. The number of jobs is started increasing again, right? What these pundits were doing though is that on top of that chart showing the number of jobs created by private companies, they overlayed a point in time saying, well, this is the point in time when Obama care was approved, when Obamacare was passed, the affordable care act and the passing of the affordable care act happens to coincide very closely to the a, with the inflection point in the curve when the curve has started, start going up again the number of jobs, right?
Alberto Cairo: 16:02 So what these people were implying is that the affordable care act is great for the job market. Contrary to what Republicans say, Republicans like to say that the affordable care act is terrible for the job market because it’s hindering companies disability to hire people and so on and so forth because it’s so expensive to pay for healthcare, right? Um, in contrary to that, these parties were say, well, that is not true. Take a look at what happens. The job market is recovering and take a look at what happened on the inflection point. The affordable care act was passed. Well, I don’t have a strong opinion about this. I’m not an economist, and I don’t have an opinion as to whether Obamacare is good or bad for the, for the job market. But what I point out in the talk is that that is completely beside the point what you believe about Obamacare, whether you think that is good or bad, the chart itself doesn’t tell you anything about that.
Alberto Cairo: 16:54 Because a chart, this is a principle of good chart reading or good chart making, a chart only shows what it shows and nothing else. And all that that chart is showing is that there is a coincidence in time between two different events, the inflection point in the job market curve and the passing of Obamacare. But that doesn’t mean that Obama care is what caused the inflection in the job market curve. They could be completely unrelated, or they could be many other factors that also contributed to the recovery of the economy. The chart is not revealing them right. So the chart doesn’t prove anything. It’s completely, I think that that particular chart was completely useless to either attack or to defend Obamacare. The chart only zone
Allison Hartsoe: 17:40 We often talk about that in analytics, you know, correlation is not causation.
Alberto Cairo: 17:44 Yeah. That’s the common mantra in statistics, right? All the it needs to have an extension. Needs to have a caveat, which is the correlation or coincidence, temporal coincidence are usually the first clue to later finding causation. That’s also very important to remember. It’s not that all correlations are meaningless. Correlation is a very important or relationships because correlation has a very specific meaning in statistics. Relationship between variables is usually the first clue that you need to pay attention at, right? But the chart alone, that’s the key thing. The chart alone is often not enough to establish in a caution relationship between variables. You need more, right? More information.
Allison Hartsoe: 18:25 What percentage of the charts that you see that are displaying information but not necessarily matching the story that someone is telling about them? What percentage of them are in this camp of um, correlation not Causation?
Alberto Cairo: 18:43 It’s difficult to tell. I don’t know what percentage it could be, but uh, it’s, it’s, it’s quite a lot. I mean it’s, it’s, it’s very common. It’s very common to see, and it’s not just a problem with correlations. It’s a problem we’ve, and again, this is another good principle of good chart making and good chart reading. Don’t read too much into a chart. That’s another thing, right? We in relationship to a chart only shows what it shows and nothing else. Don’t read too much, too much into the, into a single chart. Try to think beyond the chart. A Chart is a visualization can be an extremely powerful means for communication, but only if you control yourself or you curve your own instincts to try to see too much into it or read too much into it. And we all do these, right? We are, again, the chart itself, there was nothing wrong with it. There is a coincidence in time between the affordable care act and the inflection point. That could be that perhaps it could be perhaps that Obamacare is actually not that bad for the job market is a possible clue to that. But again, the chart alone doesn’t prove that you need more information. You need more data in order to stop chain conversation? That’s what I try to, I try to explain, but we need to jump. We, we tend to jump to conclusions. That’s the articularly if those conclusions confirm what we want to believe.
Allison Hartsoe: 20:05 Now is there an example from the right, we’ve already taken the left to task. I imagine there’s one on the other side.
Alberto Cairo: 20:11 Yeah. Yeah. There are so many. Um, the most, uh, perhaps in most newsworthy or most current one is the way that a partisan Republicans, particularly those who support president Trump very strongly tend to misread the county-level election results map, right? Just picture in your brain the results of the 2016 presidential election at the county level, right. You will imagine a map that is mostly covering red with just a few spots of blue in new urban areas and in coastal areas like an ocean of red and we’ve, we’ve few with a few islands of blue here and there, right? Well, there’s nothing wrong with that map per se. The map is correctly designed. It’s just this plain devote at the county level, right? Red and blue, Republican, a Republican and Democrat. There’s nothing wrong with that map, but the inferences that people make from that map are completely wrong, right?
Alberto Cairo: 21:12 It’s like ah, because the map, what some people particularly again on the Republican side tend to see on that map is a victory by a landslide. They are using the map to infer popular vote, right? The number of people who voted for either president Trump or Hillary Clinton and they say, take a look at how much red there is on these map and how little blue there is on these map. Actually, we could, if you can calculate with something that I did, the percent of red area on that map in the county level map and the percent of blue area on that map, it’s 80% red and 20% blue, but that is not what happened in the popular vote, right? The popular vote was pleaded of 46 point something percent for Trump and 48% for Hillary Clinton. Trump, Trump lost the popular vote. So you can not infer that support for president Trump is huge based on that map. But that is what the map implies. If that is what you want to believe that president Trump has huge support and that is not, that is not true. Sorry.
Allison Hartsoe: 22:18 I think the proper pronunciation there is huge, right?
Alberto Cairo: 22:22 Huge. Yeah. Huge, huge, huge support, right? Again, and the point that I make during the talk, I tried to, you know, wrap up all these examples with a little bit of humor and I always proceed them with the, with a similar sentence saying, um, what I’m saying right now is a completely, besides of what you think about a politics at the moment, so objectively speaking, this map is misleading you for such and such reason regardless of whether you will post Trump or do you support Trump, you don’t, it misleads us anyway and we need to be careful. Yeah.
Allison Hartsoe: 22:56 Let’s talk about another example. I saw this one on your talk at Berkeley, and you went through the hurricane chart in the cone of uncertainty. Will you take us through this one because this is one we see all the time and you know, for those of us in this space, I never even thought to question it, but take us through why this one is misleading.
Alberto Cairo: 23:16 Wow. All right. Okay, so let’s get, let’s get started because explaining visuals in a podcast is always difficult. All right, so in that particular example, it’s quite complicated, but also fascinating. I, I in the, in the book that I’m writing, I devote like, like six or seven pages to it, but I’m going to try to be concise. So first of all, picture, oh your, all right, perfect. And I have an article online that we can link to where people can read about it. All right. So anyway, so picture how hurricane forecasts are usually displayed in the media, right? So when, the media or the National Hurricane Center, which is the, the, the center that produces all these maps, when they want to inform you of where a hurricane or a tropical storm could go, they usually create a map which you can see digital graphic area that may be affected by the hurricane.
Alberto Cairo: 24:07 And then they show you the path of the hurricane surrounded by a cone of increasing size, right? We call that the cone of uncertainty. Now, the right way to read that map is to imagine that that area contained in the cone is made of, let’s say dozens or hundreds of possible paths of this storm. Basically what scientists are trying to tell you is that this line here in the middle of the cone is what we estimate that is the most probable path of the storm, but this storm could be a little bit to the right or a little bit to the left within the boundaries of this cone. All right, so that’s how to read the, by the way, the path of the center of the storm. That’s a very important, very important thing to remember. Anyway, that’s not how many people read that map when and these has been proven experimentally, alright.
Alberto Cairo: 25:05 People have been shown these map and ask, what do you see in this map? When many people see the cone of uncertainty, what they see is an area of impact or they imagine that the cone is showing the possible areas that may be affected by this storm, right? That’s a reason why some people down here in Florida called the cone of uncertainty. The cone of death. If you are inside, either you are inside the cone, you are under threat. If you’re outside the cone, you are fine. But that is not how to read it obviously because again, the cone of uncertainty, just basically our range of possible paths of the center of the storm and the center of the storm is just a point, right? A storm is a huge object. So try to imagine, for example, that at the end the storm passes by the right edge of the, of the cone of uncertainty.
Alberto Cairo: 25:56 It could affect, it could still affect an area very far away from the uh, from the cone itself just because the storm is a huge object, but it gets even worse. It gets even worse when you get to the nitty-gritty of how they how that cone of uncertainty is created. So the first time that I saw it, I saw the map, I saw the cone, et Cetera. I ask myself, does the cone contain all forecasts, all probable or possible paths of the storm, right? And that was not my assumption, right? Remembers it, I remember Stats 101, right? I remember on satellite ends. Right, exactly. That’s what most people believe, right? That’s what if you know a little bit of a stats, that will be your inference, right? And it’s completely wrong. You, you may think that you know the cone contains 95% of possible paths, but in some strange cases, in some you know, outlier cases, it could be 5% of the cases.
Alberto Cairo: 26:54 The storm could be outside the boundaries of the cone of uncertainty. The actual path of the storm will be outside. But that is not true. When you read the footnotes of the map, or you go to the documentation of how the map is created, the cone contains only two out of three possible paths of the storm. That means that one out of three of the time, the actual path of the storm, the track, the track of the storm could be, will be outside the boundaries of the cone of uncertainty, right? So is the map bad, no, it is not bad, it is only bad if you don’t know how to read it. And that’s why it’s so relevant and so important that people in my profession, journalists, we learn to read charts better because if we don’t read the charts well, right, we will not be able to inform the public about how to read them well. I have never seen in a news media, TV casts, for example, newscasts explain someone, explain the cone of uncertainty correctly. It may happen, but I have not seen it. So, and we need to explain it better just because again, you could be affected by the storm even if you are outside the, uh, the cone of uncertainty. The probability of that happening is really high.
Allison Hartsoe: 28:11 Exactly, and I, I think it’s very interesting where you put the responsibility, you don’t put the responsibility on the receiver, but you put it on the communicator, on the journalists.
Alberto Cairo: 28:22 I put it in both. So I put it in both. Um, that’s a very, also another very relevant point. Well, the main responsibility is obviously on the communicator. The communicator needs to make an effort in explaining things as clearly as possible, right? That’s the primary responsibility. But, and this is very important, but there is also a responsibility on the part of the readers, right? We live in a time in which we don’t pay attention. We just browse through, browse through things very quickly. We don’t pay attention to them. We take a look at our chart, and we believe we understand it and you we, oh, I understand. I understand the chart. All right. I don’t need to take a look at it carefully because I have already understood it. I can move away to take a look at the next, next piece of content.
Alberto Cairo: 29:06 So there’s also a responsibility on the part of the audience to educate ourselves to be more attentive, to be more careful. Began. Remember what I said about the title of my talk. If you only read the title, you will be misled. You need to read beyond the title. You need to read the footnotes. You need to read the fine print in order to understand what it is that the chart is showing or what the talk is about, right? So we have a responsibility as people, as readers, as citizens, to be more attentive and to be more responsible. Uh, with the way that we absorb, absorb information, and we handle the information that we absorb.
Allison Hartsoe: 29:45 I’m going to push back on that for just a minute because I think there’s a natural human intent to see what you want to see. How do you know that you’re just not seeing what you want to see versus thinking critically?
Alberto Cairo: 29:56 Well, you can put yourself in the position of thinking in a, in terms of counterfactual. All right, so I can tell you about that. Let’s go back to the example about the affordable care act, right? Um, I, I’m part of the audience who may be misled by that chart into thinking that Obamacare is great for the job market because I happen to be from Europe, from Western Europe, I’m from Spain. We have so to speak, socialized medicine, which is a very funny term in the United States. Socialized medicine in Spain, right, pay through taxes, and it works wonderfully. It works really, really well. Health care, public health care, universal public healthcare in Spain is fantastic. So I am, and I tend to believe that that could be a great model for the United States, right? Why don’t we have universal health care here in the United States? I’m open to debate, but what I’m trying to say is that I can be someone who will be misled by that chart, but I was not misled by that chart.
Alberto Cairo: 30:55 Why? Because I, I thought about it carefully. I didn’t just take a look at it. Imagine in that is just an illustration. Take a quick look at it and say, okay, yeah, that is great. Obamacare is great for the job market. And I moved away from it. I paid attention to the chart. I thought about the chart, I started thinking about alternative explanations for the inflection point, right? For example, um, a, um, the recovery act, right? And it which happened before the inflection point, right? The Obama presidency injected billions of dollars in the private market. That could be one possible explanation for the job market picking up at that particular time, even more so than Obamacare. So there are ways in which you can curb, I believed your own ideological predispositions or biases. It can be done. You cannot do it 100% of the time obviously.
Alberto Cairo: 31:51 But if you only do it say 50 or 60 or 60 or 70% of the time, that’s progress. That’s progress. And there is another way. There is another way in which you can do this also in the long term, which is to create, and that’s another recommendation that I make over and over and over and over again. Create a varied media diet. So, um, try to try to identify media sources all over the ideological spectrum, um, that look trustworthy and that you can consult and that you read assuming that the people writing that content are not trying to lie to you, right? And that’s how that, that’s something that I have been done throughout the years. I have, I have curated at least of media sources that I can consult and that I can see the trustworthy, both on the right and on the left. I may expose myself actively to things that are very different to mine. It’s hard to do. I’m not going to say that it’s easy to do because again, we all love to have our own opinions confirm. But when we read an argument against our own opinions, we need to really carefully obviously given that that argument is well constructed and is honest
Allison Hartsoe: 32:58 Is it harder to have that variety and in age where, uh, there are so many recommendation engines driving the next article that we see?
Alberto Cairo: 33:09 It is harder in a sense, and again, this all comes back to the responsibility on the part of readers. It is harder in the sense that we have become our own curators of information, meaning that we are now responsible to create our own media diet. We can not just rely on, you know, forces above us to curate information for us. So we need to actively identify throughout time sources that are reliable, that don’t lie actively and use them for information in the future and only consult those sources. So I my own, um, my own bias when I read pay news in social media is that if that a particular piece of information comes from a source that I don’t know, um, that I have not vetted myself, all right, uh, in a certain period of time, immediately I don’t trust that piece of information. I will look more, I will look deeper into it before, before sharing it in social media, before tweeting about it. I will take a look at very, very close look at what the sources and what the information,
Allison Hartsoe: 34:14 I personally use a rule of triangulation. If I can find three different sources saying basically the same thing, then I start to trust that piece of information. Do you have a rule of thumb like that?
Alberto Cairo: 34:26 But do you need to be, yeah, that could be part of the process of vetting your sources, but he can also be dangerous because sources tend to link to each other. So again, at least one of those, one or two of the, uh, a or the corners of that triangle should be sources that you already trust, that have proven to you that they don’t lie actively, right? So it cannot be just any source. Those three sources cannot be any source. One or two of them need to be sources that can be trustworthy according to your own assessment. And again, it’s hard work because you need to follow sources throughout a particular period of time to see what they publish, what their orientation is, what they publish, good content or bad content, bias content or non-biased content. So it’s hard work. So yeah, we need to do for hard work I believe as, as readers, as consumers of information.
Allison Hartsoe: 35:17 When you’re trying to decide what’s a valid source, how important are the retractions?
Alberto Cairo: 35:23 Extremely important. Actually just, that’s one of the things that, that I use as, um, as a, as a sign for quality. So if I use a information, never publishes corrections visibly, I just erased it from my list. I only have, I only include sources that public that, that publish corrections. Um, when the, when they screw up, we all screw up. There is a difference between line and is screwing up. Everybody screws up. That stuff very important. Um, but not everybody lies in news media. Right. So, and that the key difference, you can identify when you focus, when you pay attention at corrections.
Allison Hartsoe: 36:04 Yeah, these are, these are great tips. Are there other tips that people should keep in mind when they’re looking at these visualizations? And we’ve talked a lot about thinking critically, we’ve talked about reliable sources. Um, what else should they be? You know, if they’re, if they’re convinced they really want to think more carefully about data and the visuals that they look at. Is there a hit list or a checklist or something that they could go through?
Alberto Cairo: 36:28 Yeah, I’m the sort of things that we need to pay attention to. All the tips that I give during the talk. So for an instance, take obviously take a look at the source of the, of the data. If you have time, take a look at the primary source of the data. So we have the chart, for example, is displaying the results of a pole or the results of a scientific study. You know devote one minute or a couple of minutes to go through the primary source and see what the, what the data is and whether they chart published in that particular source reflects the primary source as well. So that will be the first thing. If you have the time, obviously we don’t always have the time to do that, but if we do, it can take us a long way in identifying faulty visualizations. Second thing that we can pay attention to is as to whether the graphic is, um, well designed or not.
Alberto Cairo: 37:19 Meaning, whether the scales on a chart have been, uh, have been distorted or things like that, or the colors of the map have been distorted to convey a particular message, et cetera. Talk extensively about these. Um, uh, another thing that is to ask ourselves whether the chart contains a sufficient amount of information in order to support the message that the chart is intended to convey. Right? And this is a very, very common problem and it’s related to the principle that explained before about when I say that a chart shows only what it shows and nothing else, right? So is the chart providing enough data to support the claims? For example, that the title of the talk, of the chart is making, right? So we need to pay attention because many charts are extremely simplified, are over-simplified, and sometimes they need to include more information.
Allison Hartsoe: 38:16 Are you also recommending in that, that you not look at the title first, that you look at the chart first and then the title.
Alberto Cairo: 38:23 You know what I mean? I don’t mind. That could be one possible approach. But if you take a look at the title first, then you need to keep the title in mind and put it in, in comparison, compare it to the actual content of the chart itself because the title can bias your perception of the chart. So it’s like read the title by all means. That’s what I do. But then putting aside for just one second and just focus on the messages that the chart is conveying and then compare those messages to the title to see whether the title has any merit or not. Right? So that could be a possible strategy. Um, don’t read too much another principal or another thing I’ll do. So again, don’t read too much into a chart. That could be another good principle of chart reading. Uh, and try to put your as much as you can. Again, it’s not a 100% possible, but try to put your own cultural and ideological biases aside. When you, when you read a chart, when you assess a chart, tried to assess the chart based on its own merits, not on what you want to see on the chart. Again, this is hard work. It sounds easy when you say it. It’s hard, but it can be done.
Allison Hartsoe: 39:33 But I think the first step, like most biases, is being aware that it’s, that you might have a bias.
Alberto Cairo: 39:39 Yeah, that’s perhaps the hardest part, right? Acknowledging that we all come from somewhere, right? We all have values, and we all have, and we need to assume that other people’s values also have merit and those who have reasons to exist, right? If we want to have an informed conversation,
Allison Hartsoe: 39:57 Well, and I think that it’s especially hard in this industry when we’re trying to create data-driven cultures and we’re trying to give people actionable insights. You know, there’s pressure to show that there’s something you can do with the data. And so I think it’s very easy to leap into, um, the fact that something is, um, perhaps more important than it is. But in testing, which is something we also talk about on the show, can be a really great avenue to say this is an assumption. Now let’s test it and see if it’s,
Alberto Cairo: 40:29 Yeah, yeah. Testing is really important, but I believe that there’s something else that is losing wherein which is to assume or perhaps accept that charts are not conversation stoppers. Meaning a chart will not close a conversation. We need to change our mind. We need to change our frame of mind. The way that we approach charts. Charts are not the way to end conversations. Charts are a means to push conversation forward, to facilitate conversations. Charts alone are rarely arguments on their own when they are presented in isolations. But charts can be extremely powerful elements in conversations or assets in arguments about a particular topic, right? Uh, again, we tend to believe that a chart represents truth. That a Chart is subjective, that a chart that the data on the chart is scientific. And, and that is, I mean that’s not a bad thing to think, right. But at the same time, we also need to remember that charts are limited. They are limited models of reality, and therefore, they cannot capture reality itself in all its complexity. They need to be putting context. They need to be put again inside longer arguments about the topics that we need to discuss.
Allison Hartsoe: 41:48 Yeah, that’s a fantastic point. I love that. Thank you. So now you’ve got the tour coming up. Can you talk a little bit about how would people get in touch with you or how they might find your books or how they might catch your tour?
Alberto Cairo: 42:02 Sure. I mean the best way to to find me is through my own web blog, which is the title of my first book.com. So it’s like, uh, the functional, the functional art.com, thefunctionalart.com. That’s my web blog. Also on Twitter, I’m pretty active on Twitter, and I tweet mostly about visualization, it’s my first name and last name. So it’s @AlbertoCairo, the website of the, of the visual trumpery talk, uh, it has not been updated for a while, so I need to, to work on that in the next few months. But uh, it says trumperytour.com, and I usually post all the dates and places of future talks in there. Um, yes. I’m going to Nashville for example, in September or October. I don’t remember. I need to check that out. Um, I’m going to North Carolina in September, so I am visiting a Raleigh, Chapel Hill, a Durham and then Charlotte. So four different cities in North Carolina in September. Um, yeah.
Allison Hartsoe: 43:14 Now if somebody can’t get to one of those talks is there are particular talk you’re very fond of that we could link to in our show notes?
Alberto Cairo: 43:21 Sure. There is a, I gave a version. The trumpery talk has different versions, and I believe that one of the best ones is the one that I delivered in New Zealand. Um, during the IHakka a celebration. So IHakka is I H A K A. IHakka. So the IHakka lectures is a series of lectures honoring a Ross IHakka who is one of the creators of the R programming language. He was a professor, he was a professor in New Zealand, in Auckland, in the Department of Statistics of the University of Auckland. He retired, he’s still around, and they created this series of lectures, and they invited me to deliver the visual trumpery talk during the Ihaka lectures. And it’s available on Youtube. That’s one of my favorite versions of the Trumpery talk. But it changes every, every time that I deliver it. I update the examples. I changed it a little bit. The structure of the talk doesn’t change that much, but the examples that I showcase, uh, those I change.
Allison Hartsoe: 44:22 I, um, I have to say that if you, if you have the chance to catch Alberto Cairo on one of these talks, definitely go and see him. I, I caught one of them online before we spoke on this podcast. And you are just so entertaining. It’s so delightful to hear this topic, which can be fairly dry, delivered in such a funny, entertaining way. So I think that’s fantastic. Thank you for taking up the cod
Alberto Cairo: 44:50 Thank you for the kind of work.
Allison Hartsoe: 44:51 Good. Well, let’s summarize a little bit about what we’ve heard. Uh, we talked about why should I care about visual, temporary, you know, this, this visual mischief and, and deception that’s going on. And, and we came to the conclusion that there were five great elements that happen in most powerful charts or most powerful graphics from them being truthful, beautiful, functional, insightful, and enlightening.
Allison Hartsoe: 45:20 Those are the five key components that we’re really after. But you know, we can marry that with what you said at the end, Alberto, which is the chart isn’t just meant to be a stopping point. And I think that’s where the last piece enlightening becomes so valuable is if it is indeed enlightening, shouldn’t it provoke a conversation? Shouldn’t it cause people to say, oh, I didn’t know that. What about this? And, and indeed, that’s what we oftentimes like to hear when people are engaging with our work is they find six different ways that they want to twist it and turn it to explore it and understand it. A, it can be a powerful way to, to get hold of the data and the story behind the data.
Alberto Cairo: 46:00 That’s a fantastic summary. You should deliver the talk in the future.
Allison Hartsoe: 46:06 I don’t think I could do justice. Not like you. Thank you. And then we talked about the different examples we talked about, uh, the left and the right versions in the end at the edit or sorry, the electorial map. Uh, the Obamacare on the left and, uh, electorial map on the right. And what you said here that was really good was it, and, and I’ve heard this before from other people that we’ve interviewed, which is to think critically, to think beyond what the data is telling you to think beyond the chart. And, and especially here, we talked a lot about pausing, set your biases aside and really think that the chart is just the chart. It’s only showing you what it has. It’s not a, it’s not designed to tell you all the answers at once unless it’s extremely well designed.
Alberto Cairo: 46:55 And there’s another, I’m going to interrupt you in there but because there was another element to that which I forgot to mention, but I make this point in the talk and also in the, in my, in my third book, the one that I’m writing, which is that when we think alone we know, we don’t reason, we rationalize. That’s a very another very relevant point. So when we only talk to ourselves or to people who are likeminded people who already think like us, we tend to basically use our reasoning skills to confirm what we already believe. It is better, and again this is connected to the idea of charts as part of the, as part of a dialogue enabling process, right? So don’t reason about the chart on your own. Talk with other people who are not necessarily likeminded about the chart because every, every person will see something different in the chart and understanding and good reasoning may arise through the conversation about the data that is being shown to you and to that other person. We don’t think well alone. We are social creatures, and we only can reason well when we don’t think on our own or in collaboration with people who already believe what we believe. We reason better when we partner out with people who are not necessarily like us but who are a little bit different than we are.
Allison Hartsoe: 48:16 I love that, that is going to be our closing note today. Don’t, don’t reason on your own and definitely look for those opposing opinions to come to a proper, unbiased conclusion. Alberto, thank you so much for joining us today. It’s been an absolute pleasure to have you.
Alberto Cairo: 48:33 Thanks for having me.
Allison Hartsoe: 48:35 Links to everything we discussed are@ambitiondata.com/podcast and that will include links to the different visuals that we talked about today and links to the video that Alberto mentioned at the end of our talk. Now, remember when you use your data effectively, you can build customer equity. It is not magic. It’s just a very specific journey that you can follow to get results. Thank you for joining today’s show. This is Allison. Just a few things before you head out. Every Friday, I put together a short bulleted list of three to five things I’ve seen that represent customer equity signal, not noise, and believe me, there’s a lot of noise out there. I actually call this email The signal, things I include could be smart tools I’ve run across, articles I’ve shared, cool statistics, or people and companies I think are doing amazing work, building customer equity. If you’d like to receive this nugget of goodness each week, you can sign up@ambitiondata.com, and you’ll get the very next one. I hope you enjoy the signal. See you next week on the customer equity accelerator.