Turning Bad Charts into Compelling Data Stories | Dominic Bohan | TEDxYouth@Singapore

so which slice of pie is largest let's do a quick show of hands who thinks red is the largest slice any takers he thinks yellow a couple of votes for yellow he thinks blue is the larger slice few votes any takers for green Green is a popular choice and what about purple no votes for purple and it's not purple and one more choice who thinks they're all the same very popular so the room is divided between yellow green and all the same let's visualize the exact same data as a simple bar chart and now the answer is instantly obvious even if you had a hunch that it was Green how confident would you have been in saying which color comes second third and so on the pie chart has failed us miserably and this is far from the worst chart that's out there I'm a data storytelling trainer and I see bad charts everywhere the world is full of charts that look like this maybe you see them in your workplace or infographics meant to look fun and cute like this or worse still the creme de la creme of bad charts I give you this my vision for the future of us is a world in which no human being shall ever again have to suffer the indignity of trying to piece together a confusing abomination of a chart like this decorated like a Christmas tree why am I so passionate about data visualization and why should you care well Humanity is creating more data faster than ever you've probably heard a bunch of hype about it slogans like daughter is the new oil a little factoids like we create more data in a year now than in millennia of human history combined and the thing about this hype is it's actually justified Dada really is transforming our world but Dada is useless unless human beings can interpret analyze and understand it and use it to drive action to make sense of data we need visualization and for our visualizations to land and make an impact they need to have a message that our audience cares about in other words we need to tell a story I believe the data storytelling can change the world the most impactful data stories can even save lives okay so I know that's a big call so I want to prove it to you and today I want to share with you three simple techniques that you can all use to tell compelling stories with data you don't need any specialist expertise you don't need to be a statistician or a data science dying scientist to apply these principles so three simple principles the first is choose a human friendly chart type what do I mean by that well let's take a look at an example so we've got some data here from a personality test that I took and my friend took and we want to compare our results across each of these five major personality dimensions what do we think the differences are quite difficult to do with this pie chart and we saw earlier with our little experiment that pie charts have some pretty severe limitations and this is because we're the pie chart we're forced to decode angle and area and human beings are much better at perceiving numbers that are encoded using simple bars using simple lengths so this experiment that I performed on you earlier is very similar to a series of experiments that was performed back in 1984 by two researchers named Cleveland and McGill and Cleveland McGill were fascinated by this question of which charts of human beings good at interpreting and which charts do we struggle with so they showed their participants series of lines and bars and shapes that encoded numbers and recall these options up on the screen elementary perceptual tasks and they measured how good the participants were at deciphering each of these tasks and they ranked them from the tasks that were worse done to the tasks that we're best at and we have some clear winners human beings are best at perceiving numbers encoded by lengthened position there I'll go to two choices for human friendly chart types so let's visualize our personality test data using position not much of an improvement right so this is called the radar chart and it uses position but it uses it randomly there's no reason that these personality dimensions should appear in the particular order they do or why they should form a Pentagon it pains me to say it but if you actually take a personality test you're very likely to see a mess of a chart like this this chart is becoming popular consultants are even using it so why does this train wreck of a chart proliferate well it could be that the consultants want to distract you from how much they're charging but it could also be that this chart does look kind of interesting I have to admit some people would even use the most dangerous word in the English language when it comes to data visualization to describe this chart word is cool whenever I hear this word whenever someone runs up and tells me Dom I've got this cool new chart I want you to see I shudder in fear because I'm about to see a disgrace of a chart like this once again this chart uses position randomly and it uses area which we've seen from our pie chart example it's not a good way to encode numeric values and we can make it even worse we can make the bubbles dance around and light up modern software packages allow us to do more new and exciting things with data visualization than ever before but just because we can doesn't mean we should so I want to convince you that simpler is better and I want to come back to a simple human friendly chart type that uses position and now we're going to use position properly we're going to align these positions that show our values on a common scale we're going to change our data to a dot plot suddenly the insights are immediately obvious I am much more extroverted than my friend and she's much more agreeable I'm not the most agreeable person and on the other personality dimensions we're almost the same which is maybe why we're friends or to move on to the next key to effective data storytelling which is to be a ruthless minimalist and to explain this concept I want to use a personal example so I've been in a relationship for about two years now and it's often around this time that your partner starts to ask about some of your previous relationships so I did what any good daughter storyteller would do and I put together a chart so here it is and so thank you I'm gonna need your help to fix up this chart and so I wouldn't wouldn't blame her if she left me on the spot just for the bad chart design here so we're gonna fix this up this chart shows on the x-axis my aging years and I've estimated the intensity of some of these past relationships over time but before my audience which is my current girlfriend can understand what's going on with this chart and understand my message we need to remove all these distracting components which we call chart junk let's start with the worst component first this background color which serves no purpose now this charts just an estimate I've put together we don't need a huge degree of precision so let's get rid of the gridlines and we can also get rid of this chart title and simplify our y-axis labels now there's one more piece of chart junk that needs to be eliminated from this can you spot what it is I'm gonna go against what we might have learned in school and university and say that every chart does not need a legend this legend is forcing our audience to do work our eyes have to track up and down between the legend and the serious labels and we have to hold the series labels in our short-term memory it's diverting our attention from the message of the chart fortunately there's a better way human beings naturally perceive objects that are close together as belonging together and we can take advantage of this and just label our series directly by putting the labels close to the series and we can further reinforce and strengthen this connection by using similarity of color so let's do that now this charts still a little busy for my liking we've got a lot of color here and color is a devastatingly effective way to focus it in Dardis storytelling and because it's so effective we want to use it sparingly so I'm gonna push everything to gray for now to create a blank canvas for storytelling and that leads us into our third and final key principle of data storytelling which is that everything we put in front of our audience needs to contribute to a clear key takeaway that our audience cares about so with our all gray blank canvas let's take our audience through a story and highlight them the pieces of this story one at a time so we start with my lackluster first attempt at dating consistently pretty low intensity here let's call this a practice and in my mid-20s I thought I could handle two relationships at once the data shows this was not an effective strategy with both of them pretty quickly plummeting down toward zero and now we've got the most dangerous and volatile territory of all the ups and downs of this almost marriage and my current girlfriend might notice that the peak intensity of this relationship is higher than the highest peak of my relationship with her so I need to add a little bit of reassurance here and just let her know that I no longer speak with Liana and of course the names have been changed in this example so after that ordeal we have the recovery period I estimated the intensity of these relationships on quarterly intervals so sadly none of these dots made it into a line and finally we've got my current relationship with a forecast of strong growth expected to continue into the future thank you now I promise you at the start of this talk that I was going to convince you that data storytelling can change the world and can even save lives now this chart might not even be good enough to save my relationship so I want to show you one of the most profound examples of effective data storytelling in history and to do that we have to go back to 1854 in the crowded and squalid streets of Soho in central London back then there'd been a drop dramatic and rapid outbreak of deadly cholera in a street called Broad Street and the prevailing view amongst physicians at the time was that cholera was caused by a mysterious foul stench in the air a force they called miasma so they were wrong but there was one physician that dissented a man called Jon Snow and Jon Snow believed that cholera was being transmitted via drinking water and he shared his theory with his colleagues and they replied you know nothing Jon Snow and so Jon Snow collected some data he collected data on the locations the exact addresses of every cholera case in London and he marked down the locations where they had occurred and the more deaths there were at each location the larger these black bars that I'll show you grew and we can see when we zoom in on the Broad Street pump that we have the highest concentration of cases around this pump and back in those days people would work to their closest pump to gather water so what we see when we look at some pumps a little bit further from the Broad Street pump is the cases start to drop off and when we go very far from the Broad Street pump where no one could possibly be walking there the cholera cases appear entirely so John snow took his findings to the local parish Commission and they finally said you know something John snow and they agreed to remove the handle from the Broad Street pump and within the day the deaths had stopped now the epidemic had already peaked by this point so we don't know how many lives were saved but John snows contribution would go on to shape the field of Epidemiology and make a massive contribution to the germ theory of disease John snow was able to change the world with data storytelling because he had a simple human friendly chart that used length to encode his values he was a ruthless minimalist maybe because that back then he didn't have the choice to add chart junk to his visualizations and he had a clear and powerful take away that his audience cared about and I want to convince you that you too can change the world and tell impactful stories with data and all you need at your disposal is the simple tools that were available back in 1854 [Applause]

Channel: TEDx Talks

Views: 24,107

Rating: 4.9458413 out of 5

Keywords: TEDxTalks, English, Social Sciences, Data, Data Science, Relationships, Social Science, Statistics, Youth

Id: edAf1jx1wh8

Length: 16min 33sec (993 seconds)

Published: Tue Nov 26 2019