Editor's Note:

What would you think of the charts below? Colors and detailed numbers have all shown the efforts made by presenters. It's easy for the reaction to be "that's interesting". Except that, there is nothing more—no more ideas, no actions.

**Without exception, they are all ineffective, poor graphs! **

enter image description here

Cole will tell you with a large number of cases that truly effective charts are the same as stories that can produce certain suggested meaning. It invites a response from audiences and leads to productive conversations.

enter image description here

访谈嘉宾:Cole Knaflic,数据分析专家,前Google人力分析团队经理,曾任银行和私募基金分析师,曾在马里兰艺术学院教授信息可视化课程。目前专门研究定量信息的有效展示,并撰写热门博客storytellingwithdata.com。她的数据分析研讨会和演示深受世界各地受众追捧。

Cole 认为在“如何运用数据进行有效沟通”方面,需要着重考虑两个问题:

  • 抓住受众的视线
  • 把数据以故事的形式展现,包含情境、情节和结局

enter image description here

《用数据讲故事》里面,Cole 通过大量案例研究介绍数据可视化的基础知识,以及如何利用数据创造出吸引人的、信息量大的、有说服力的故事,进而达到有效沟通的目的。具体内容包括:如何充分理解上下文,如何选择合适的图表,如何消除杂乱,如何聚焦受众的视线,如何像设计师一样思考,以及如何用数据讲故事。

Interview Transcripts:

See Chinese Version

Cole is the lucky one who finds her goal that is to help rid the world of ineffective graphs. Could you tell us when and how did you find your love for data visualization?

Yes, I am very lucky to get to focus on an area around which I have so much passion! I first started to visualize data when I was working as an analyst in the banking industry. Looking back, that was a great place to start my analytical career because it is an area that has historically been very data-driven. For me, that was the first place where I saw how valuable data can be for helping people understand something better or make a smarter decision. By making data visual, we can make it more accessible to more people. I initially spent more time in this area of visualizing data because it was a way to bring some creativity into the process. Over time, I realized that when I paid more attention to how I showed the data, people paid more attention to the data, so it was a reinforcing cycle. At Google, I was able to further develop this area by building our training on data visualization. As part of this, I was able to do research to better understand why some of the things I'd arrived at through trial and error over time were effective and figure out how to teach this to others. I am incredibly passionate about the power of effectively communicating with data and very lucky that today, I get to spend my time teaching others how to do this, through my book, blog, and workshops.

What is the difference between simply presenting data and telling a story with it?

For me, presenting data is when you show a graph and maybe outline some findings. When an audience is faced with data like this, it's easy for the reaction to be "that's interesting" and then they move on to the next thing. Telling a story with data goes beyond this in several important ways. Stories have a shape, a narrative structure: they start with a plot, then there is some sort of tension introduced that reaches a point of climax, then there is an ending that resolves this tension. We can use these components of story when it comes to our data stories. The plot is: what context does our audience need in order to be in the right frame of mind for what we're about to tell them. The tension is: what is out of balance—what is important for our audience, why should they care? There may be twists or tension that you can show through your data. Then finally, the ending—the call to action—what your audience can do in order to resolve the tension that exists. When you end with a call to action, it invites a response from your audience, which can lead to productive conversations that often get missed when we simply present the data. That's why I encourage people to move beyond this and rather than simply show data, consider how they can make data a pivotal point in an overarching story.

Why would you say “3D pie chart is inherently evil”?

There aren't a lot of hard and fast rules when it comes to data visualization. It sits at an interesting intersection between art and science. But there are some rules. One that I often raise in my workshops is: never use 3D. The only exception is if you're actually plotting a third dimension (and even then it can get tricky quickly, so take care), but never for a single dimension like in a pie chart. In this case, 3D can visually skew the data, making it impossible to wrap our brains around what we're looking at and appropriately analyze or compare the segments. Pies themselves are also hard, because humans' eyes don't do a great job ascribing quantitative value to two-dimensional space. Pies are good if you want to show that one piece of the whole is really small, or one piece is really big. The challenge is that if you want to say anything much more nuanced than that, pies make it difficult because as soon as the segments approach similarity of size, it becomes difficult to assess which is bigger or by how much. In general, when it comes to visualizing data, I encourage people to be really clear on what they want their audience to be able to do with the graph and then design to try to make that as easy and intuitive as possible. The 3D pie chart can mislead more than it informs and should be avoided!

What are your top three principles for making an effective graph?

When communicating with data for explanatory purposes (in other words, there is a specific takeaway you want your audience to focus on or story you want to tell with the data), my top three tips are the following:

1 - Visualize the data in an accessible way (be clear on what you want your audience to do with the data and design to make that easy),

2 - Use color sparingly and strategically to direct your audience's attention, and

3 - Use words (titles, labels, annotation) to make it clear what your audience is looking at and why; put your key takeaway into words.

You've ever taught Introduction to Information Visualization at MICA. Could we get the idea that there is a natural relationship between data visualization and Art?

I do believe there is an overlap. I consider data visualization to sit at an interesting intersection between science and art. There is definitely science to it: guidelines and best practices to follow. But there's also an artistic component. This is one of the reasons I think this space is so much fun—two different people faced with the same data visualization challenge might come up with totally different solutions. There is room for diversity of approaches. There's no single "right" answer in my view when it comes to visualizing data effectively. It's important, though, to make sure that you use your artistic license to make the information easier for the audience to understand.

Sometimes, visualized data might be misleading or even biased. How to avoid it?

Yes, this can be a challenge. I think that the idea of truly unbiased data is unattainable. When we do anything with data—from what data we collect in the first place to what we choose to look at, to how we aggregate or disaggregate, to how we show it—bias is introduced at each of these steps. But it's important to take these actions with the idea of being true to the data, investigating alternative hypotheses, and painting a robust (and not misleading) picture. The golden rule of data visualization is "don't lie with data." And while sometimes there is malicious intent, I think more often this happens when people don't even realize it or mean to. There are some guidelines that should be followed to help us not mislead (e.g. bar charts must have a zero baseline, or as mentioned a moment ago, 3D should be avoided). Beyond that, to do good robust analysis, it's often helpful to solicit other view points and get feedback as a way to help ensure you're painting an appropriate picture of the data.

What inspires you to write a book? During the writing period, what's the most difficult task?

My main goal in writing the book was to be able to share what I've learned with more people. Writing is a difficult task. I was lucky, in that I've led so many workshops on this topic that I had a good idea of the main lessons I wanted to cover, and had said many of the words out loud, which made it easier to put pen to paper (and fingers to keyboard). Also, I used many of the strategies covered in the book, for example storyboarding to plan the content for a given section. Overall, I think most difficult was making the book accessible to a wide audience. My goal is that anyone can read the book and learn something, from those who don't have any experience visualizing data to those who work with data regularly and want to take the stories they tell with their data to the next level. To achieve this, I had a wide range of reviewers, from non-technical to a number of my respected colleagues in the field.

In your book, you've also mentioned several relevant luminaries. Which one gives you the greatest inspiration?

That's a great—and difficult—question! I'm not sure I can limit it to a single one. Stephen Few was certainly influential when I was first learning about best practices for communicating with data (my favorite book of his is "Show Me the Numbers"). Nancy Duarte has been an influence when it comes to better understanding and resonating with audiences (my favorite book of hers is "Resonate") and I often use her "Big Idea" as an exercise in my workshops to help people understand the benefit of getting clear and concise on their overall message. I also get a lot of day to day inspiration from others who blog about and work in the field of data visualization and visual communications.

Working at Google means a lot for your development. What's your life at Google and why those experiences are important?

Yes, Google played an important role for me (coincidentally, it is where I met my husband, so you can really say it was life changing!). Google is a company that is data-driven in everything they do—even how they manage the people function (traditionally, "Human Resources"). I had the good fortune to join the People Analytics team at Google in 2007, which was just as the team was forming so it was small and I had exposure to a ton of interesting work, learning about things like what makes a manager effective and how can we quantify that, or what drives attrition and how we might predict who is likely to leave the company. I also had the opportunity to build a course on data visualization. This started as part of an HR training program, but garnered such broad interest that we ended up rolling it out across all of Google and I was able to travel to a number of our offices throughout the world teaching people how to effectively visualize data. Over time, word started to spread and other organizations started reaching out to me, wondering if I could help them and their teams learn how to do this as well. Ultimately, I left Google to pursue my passion of teaching the world to better visualize data and use it to tell stories to increase understanding and drive action.

After leaving Google, what have you been doing to move your goal forward?

Since leaving Google, over the past few years I have been able to teach people at organizations around the world how to effectively tell their stories with data. I wrote the book, storytelling with data: a data visualization guide for business professionals (which has been translated into 9 languages), write the blog storytellingwithdata.com and spend a lot of time conducting public and custom workshops. My view is that anyone can improve their ability to communicate effectively with data and I very much enjoy getting to teach people about how to do this. Perhaps my work will even bring me to China one day! Thanks very much for inviting me to do this interview and helping me to share my work and ideas with your readers.