Tom segura bellco theatre september 28

Identify a chart type that could be used to display different editorial perspectives of your dataset and explain why you felt it to be appropriate.
Identify two other chart types that could show something about your subject matter, though maybe not confined to the data you are looking at. In other words, chart types that could incorporate data not already included in your selected dataset.
Review the classifying chart families in Chapter 6 of your textbook. Select at least one chart type from each of the classifying chart families (CHRTS) that could portray different editorial perspectives about your subject. This may include additional data, not already included in your selected dataset.

Data Visualisation

2

3

Data Visualisation A Handbook for Data Driven Design

Andy Kirk

4

SAGE Publications Ltd

1 Oliver’s Yard

55 City Road

London EC1Y 1SP

SAGE Publications Inc.

2455 Teller Road

Thousand Oaks, California 91320

SAGE Publications India Pvt Ltd

B 1/I 1 Mohan Cooperative Industrial Area

Mathura Road

New Delhi 110 044

SAGE Publications Asia-Pacific Pte Ltd

3 Church Street

#10-04 Samsung Hub

Singapore 049483

5

© Andy Kirk 2016

First published 2016

Apart from any fair dealing for the purposes of research or private study, or criticism or review, as permitted under the Copyright, Designs and Patents Act, 1988, this publication may be reproduced, stored or transmitted in any form, or by any means, only with the prior permission in writing of the publishers, or in the case of reprographic reproduction, in accordance with the terms of licences issued by the Copyright Licensing Agency. Enquiries concerning reproduction outside those terms should be sent to the publishers.

Library of Congress Control Number: 2015957322

British Library Cataloguing in Publication data

A catalogue record for this book is available from the British Library

ISBN 978-1-4739-1213-7

ISBN 978-1-4739-1214-4 (pbk)

Editor: Mila Steele

Editorial assistant: Alysha Owen

Production editor: Ian Antcliff

Marketing manager: Sally Ransom

Cover design: Shaun Mercier

Typeset by: C&M Digitals (P) Ltd, Chennai, India

Printed and bound in Great Britain by Bell and Bain Ltd, Glasgow

6

Contents List of Figures with Source Notes Acknowledgements About the Author INTRODUCTION PART A FOUNDATIONS

1 Defining Data Visualisation 2 Visualisation Workflow

PART B THE HIDDEN THINKING 3 Formulating Your Brief 4 Working With Data 5 Establishing Your Editorial Thinking

PART C DEVELOPING YOUR DESIGN SOLUTION 6 Data Representation 7 Interactivity 8 Annotation 9 Colour 10 Composition

PART D DEVELOPING YOUR CAPABILITIES 11 Visualisation Literacy

References Index

7

List of Figures with Source Notes 1.1 A Definition for Data Visualisation 19 1.2 Per Capita Cheese Consumption in the U.S., by Sarah Slobin (Fortune magazine) 20 1.3 The Three Stages of Understanding 22 1.4–6 Demonstrating the Process of Understanding 24–27 1.7 The Three Principles of Good Visualisation Design 30 1.8 Housing and Home Ownership in the UK, by ONS Digital Content Team 33 1.9 Falling Number of Young Homeowners, by the Daily Mail 33 1.10 Gun Deaths in Florida (Reuters Graphics) 34 1.11 Iraq’s Bloody Toll, by Simon Scarr (South China Morning Post) 34 1.12 Gun Deaths in Florida Redesign, by Peter A. Fedewa (@pfedewa) 35 1.13 If Vienna would be an Apartment, by NZZ (Neue Zürcher Zeitung) [Translated] 45 1.14 Asia Loses Its Sweet Tooth for Chocolate, by Graphics Department (Wall Street Journal) 45 2.1 The Four Stages of the Visualisation Workflow 54 3.1 The ‘Purpose Map’ 76 3.2 Mizzou’s Racial Gap Is Typical On College Campuses, by FiveThirtyEight 77 3.3 Image taken from ‘Wealth Inequality in America’, by YouTube user ‘Politizane’ (www.youtube.com/watch?v=QPKKQnijnsM) 78 3.4 Dimensional Changes in Wood, by Luis Carli (luiscarli.com) 79 3.5 How Y’all, Youse and You Guys Talk, by Josh Katz (The New York Times) 80 3.6 Spotlight on Profitability, by Krisztina Szücs 81 3.7 Countries with the Most Land Neighbours 83 3.8 Buying Power: The Families Funding the 2016 Presidential Election, by Wilson Andrews, Amanda Cox, Alicia DeSantis, Evan Grothjan, Yuliya Parshina-Kottas, Graham Roberts, Derek Watkins and Karen Yourish (The New York Times) 84 3.9 Image taken from ‘Texas Department of Criminal Justice’ Website (www.tdcj.state.tx.us/death_row/dr_executed_offenders.html) 86

8

3.10 OECD Better Life Index, by Moritz Stefaner, Dominikus Baur, Raureif GmbH 89 3.11 Losing Ground, by Bob Marshall, The Lens, Brian Jacobs and Al Shaw (ProPublica) 89 3.12 Grape Expectations, by S. Scarr, C. Chan, and F. Foo (Reuters Graphics) 91 3.13 Keywords and Colour Swatch Ideas from Project about Psychotherapy Treatment in the Arctic 92 3.14 An Example of a Concept Sketch, by Giorgia Lupi of Accurat 92 4.1 Example of a Normalised Dataset 99 4.2 Example of a Cross-tabulated Dataset 100 4.3 Graphic Language: The Curse of the CEO, by David Ingold and Keith Collins (Bloomberg Visual Data), Jeff Green (Bloomberg News) 101 4.4 US Presidents by Ethnicity (1789 to 2015) 114 4.5 OECD Better Life Index, by Moritz Stefaner, Dominikus Baur, Raureif GmbH 116 4.6 Spotlight on Profitability, by Krisztina Szücs 117 4.7 Example of ‘Transforming to Convert’ Data 119 4.8 Making Sense of the Known Knowns 123 4.9 What Good Marathons and Bad Investments Have in Common, by Justin Wolfers (The New York Times) 124 5.1 The Fall and Rise of U.S. Inequality, in Two Graphs Source: World Top Incomes Database; Design credit: Quoctrung Bui (NPR) 136 5.2–4 Why Peyton Manning’s Record Will Be Hard to Beat, by Gregor Aisch and Kevin Quealy (The New York Times) 138–140 C.1 Mockup Designs for ‘Poppy Field’, by Valentina D’Efilippo (design); Nicolas Pigelet (code); Data source: The Polynational War Memorial, 2014 (poppyfield.org) 146 6.1 Mapping Records and Variables on to Marks and Attributes 152 6.2 List of Mark Encodings 153 6.3 List of Attribute Encodings 153 6.4 Bloomberg Billionaires, by Bloomberg Visual Data (Design and development), Lina Chen and Anita Rundles (Illustration) 155 6.5 Lionel Messi: Games and Goals for FC Barcelona 156 6.6 Image from the Home page of visualisingdata.com 156 6.7 How the Insane Amount of Rain in Texas Could Turn Rhode Island Into a Lake, by Christopher Ingraham (The Washington Post) 156

9

6.8 The 10 Actors with the Most Oscar Nominations but No Wins 161 6.9 The 10 Actors who have Received the Most Oscar Nominations 162 6.10 How Nations Fare in PhDs by Sex Interactive, by Periscopic; Research by Amanda Hobbs; Published in Scientific American 163 6.11 Gender Pay Gap US, by David McCandless, Miriam Quick (Research) and Philippa Thomas (Design) 164 6.12 Who Wins the Stanley Cup of Playoff Beards? by Graphics Department (Wall Street Journal) 165 6.13 For These 55 Marijuana Companies, Every Day is 4/20, by Alex Tribou and Adam Pearce (Bloomberg Visual Data) 166 6.14 UK Public Sector Capital Expenditure, 2014/15 167 6.15 Global Competitiveness Report 2014–2015, by Bocoup and the World Economic Forum 168 6.16 Excerpt from a Rugby Union Player Dashboard 169 6.17 Range of Temperatures (°F) Recorded in the Top 10 Most Populated Cities During 2015 170 6.18 This Chart Shows How Much More Ivy League Grads Make Than You, by Christopher Ingraham (The Washington Post) 171 6.19 Comparing Critics Scores (Rotten Tomatoes) for Major Movie Franchises 172 6.20 A Career in Numbers: Movies Starring Michael Caine 173 6.21 Comparing the Frequency of Words Used in Chapter 1 of this Book 174 6.22 Summary of Eligible Votes in the UK General Election 2015 175 6.23 The Changing Fortunes of Internet Explorer and Google Chrome 176 6.24 Literarcy Proficiency: Adult Levels by Country 177 6.25 Political Polarization in the American Public’, Pew Research Center, Washington, DC (February, 2015) (http://www.people- press.org/2014/06/12/political-polarization-in-the-american-public/) 178 6.26 Finviz (www.finviz.com) 179 6.27 This Venn Diagram Shows Where You Can Both Smoke Weed and Get a Same-Sex Marriage, by Phillip Bump (The Washington Post) 180 6.28 The 200+ Beer Brands of SAB InBev, by Maarten Lambrechts for Mediafin: www.tijd.be/sabinbev (Dutch),

10

www.lecho.be/service/sabinbev (French) 181 6.29 Which Fossil Fuel Companies are Most Responsible for Climate Change? by Duncan Clark and Robin Houston (Kiln), published in the Guardian, drawing on work by Mike Bostock and Jason Davies 182 6.30 How Long Will We Live – And How Well? by Bonnie Berkowitz, Emily Chow and Todd Lindeman (The Washington Post) 183 6.31 Crime Rates by State, by Nathan Yau 184 6.32 Nutrient Contents – Parallel Coordinates, by Kai Chang (@syntagmatic) 185 6.33 How the ‘Avengers’ Line-up Has Changed Over the Years, by Jon Keegan (Wall Street Journal) 186 6.34 Interactive Fixture Molecules, by @experimental361 and @bootifulgame 187 6.35 The Rise of Partisanship and Super-cooperators in the U.S. House of Representatives. Visualisation by Mauro Martino, authored by Clio Andris, David Lee, Marcus J. Hamilton, Mauro Martino, Christian E. Gunning, and John Armistead Selde 188 6.36 The Global Flow of People, by Nikola Sander, Guy J. Abel and Ramon Bauer 189 6.37 UK Election Results by Political Party, 2010 vs 2015 190 6.38 The Fall and Rise of U.S. Inequality, in Two Graphs. Source: World Top Incomes Database; Design credit: Quoctrung Bui (NPR) 191 6.39 Census Bump: Rank of the Most Populous Cities at Each Census, 1790–1890, by Jim Vallandingham 192 6.40 Coal, Gas, Nuclear, Hydro? How Your State Generates Power. Source: U.S. Energy Information Administration, Credit: Christopher Groskopf, Alyson Hurt and Avie Schneider (NPR) 193 6.41 Holdouts Find Cheapest Super Bowl Tickets Late in the Game, by Alex Tribou, David Ingold and Jeremy Diamond (Bloomberg Visual Data) 194 6.42 Crude Oil Prices (West Texas Intermediate), 1985–2015 195 6.43 Percentage Change in Price for Select Food Items, Since 1990, by Nathan Yau 196 6.44 The Ebb and Flow of Movies: Box Office Receipts 1986–2008, by Mathew Bloch, Lee Byron, Shan Carter and Amanda Cox (The New York Times) 197 6.45 Tracing the History of N.C.A.A. Conferences, by Mike Bostock,

11

Shan Carter and Kevin Quealy (The New York Times) 198 6.46 A Presidential Gantt Chart, by Ben Jones 199 6.47 How the ‘Avengers’ Line-up Has Changed Over the Years, by Jon Keegan (Wall Street Journal) 200 6.48 Native and New Berliners – How the S-Bahn Ring Divides the City, by Julius Tröger, André Pätzold, David Wendler (Berliner Morgenpost) and Moritz Klack (webkid.io) 201 6.49 How Y’all, Youse and You Guys Talk, by Josh Katz (The New York Times) 202 6.50 Here’s Exactly Where the Candidates Cash Came From, by Zach Mider, Christopher Cannon, and Adam Pearce (Bloomberg Visual Data) 203 6.51 Trillions of Trees, by Jan Willem Tulp 204 6.52 The Racial Dot Map. Image Copyright, 2013, Weldon Cooper Center for Public Service, Rector and Visitors of the University of Virginia (Dustin A. Cable, creator) 205 6.53 Arteries of the City, by Simon Scarr (South China Morning Post) 206 6.54 The Carbon Map, by Duncan Clark and Robin Houston (Kiln) 207 6.55 Election Dashboard, by Jay Boice, Aaron Bycoffe and Andrei Scheinkman (Huffington Post). Statistical model created by Simon Jackman 208 6.56 London is Rubbish at Recycling and Many Boroughs are Getting Worse, by URBS London using London Squared Map © 2015 www.aftertheflood.co 209 6.57 Automating the Design of Graphical Presentations of Relational Information. Adapted from McKinlay, J. D. (1986). ACM Transactions on Graphics, 5(2), 110–141. 213 6.58 Comparison of Judging Line Size vs Area Size 213 6.59 Comparison of Judging Related Items Using Variation in Colour (Hue) vs Variation in Shape 214 6.60 Illustrating the Correct and Incorrect Circle Size Encoding 216 6.61 Illustrating the Distortions Created by 3D Decoration 217 6.62 Example of a Bullet Chart using Banding Overlays 218 6.63 Excerpt from What’s Really Warming the World? by Eric Roston and Blacki Migliozzi (Bloomberg Visual Data) 218 6.64 Example of Using Markers Overlays 219 6.65 Why Is Her Paycheck Smaller? by Hannah Fairfield and Graham Roberts (The New York Times) 219

12

6.66 Inside the Powerful Lobby Fighting for Your Right to Eat Pizza, by Andrew Martin and Bloomberg Visual Data 220 6.67 Excerpt from ‘Razor Sales Move Online, Away From Gillette’, by Graphics Department (Wall Street Journal) 220 7.1 US Gun Deaths, by Periscopic 225 7.2 Finviz (www.finviz.com) 226 7.3 The Racial Dot Map: Image Copyright, 2013, Weldon Cooper Center for Public Service, Rector and Visitors of the University of Virginia (Dustin A. Cable, creator) 227 7.4 Obesity Around the World, by Jeff Clark 228 7.5 Excerpt from ‘Social Progress Index 2015’, by Social Progress Imperative, 2015 228 7.6 NFL Players: Height & Weight Over Time, by Noah Veltman (noahveltman.com) 229 7.7 Excerpt from ‘How Americans Die’, by Matthew C. Klein and Bloomberg Visual Data 230 7.8 Model Projections of Maximum Air Temperatures Near the Ocean and Land Surface on the June Solstice in 2014 and 2099: NASA Earth Observatory maps, by Joshua Stevens 231 7.9 Excerpt from ‘A Swing of Beauty’, by Sohail Al-Jamea, Wilson Andrews, Bonnie Berkowitz and Todd Lindeman (The Washington Post) 231 7.10 How Well Do You Know Your Area? by ONS Digital Content team 232 7.11 Excerpt from ‘Who Old Are You?’, by David McCandless and Tom Evans 233 7.12 512 Paths to the White House, by Mike Bostock and Shan Carter (The New York Times) 233 7.13 OECD Better Life Index, by Moritz Stefaner, Dominikus Baur, Raureif GmbH 233 7.14 Nobel Laureates, by Matthew Weber (Reuters Graphics) 234 7.15 Geography of a Recession, by Graphics Department (The New York Times) 234 7.16 How Big Will the UK Population be in 25 Years Time? by ONS Digital Content team 234 7.17 Excerpt from ‘Workers’ Compensation Reforms by State’, by Yue Qiu and Michael Grabell (ProPublica) 235 7.18 Excerpt from ‘ECB Bank Test Results’, by Monica Ulmanu, Laura Noonan and Vincent Flasseur (Reuters Graphics) 236 7.19 History Through the President’s Words, by Kennedy Elliott, Ted

13

Mellnik and Richard Johnson (The Washington Post) 237 7.20 Excerpt from ‘How Americans Die’, by Matthew C. Klein and Bloomberg Visual Data 237 7.21 Twitter NYC: A Multilingual Social City, by James Cheshire, Ed Manley, John Barratt, and Oliver O’Brien 238 7.22 Killing the Colorado: Explore the Robot River, by Abrahm Lustgarten, Al Shaw, Jeff Larson, Amanda Zamora and Lauren Kirchner (ProPublica) and John Grimwade 238 7.23 Losing Ground, by Bob Marshall, The Lens, Brian Jacobs and Al Shaw (ProPublica) 239 7.24 Excerpt from ‘History Through the President’s Words’, by Kennedy Elliott, Ted Mellnik and Richard Johnson (The Washington Post) 240 7.25 Plow, by Derek Watkins 242 7.26 The Horse in Motion, by Eadweard Muybridge. Source: United States Library of Congress’s Prints and Photographs division, digital ID cph.3a45870. 243 8.1 Titles Taken from Projects Published and Credited Elsewhere in This Book 248 8.2 Excerpt from ‘The Color of Debt: The Black Neighborhoods Where Collection Suits Hit Hardest’, by Al Shaw, Annie Waldman and Paul Kiel (ProPublica) 249 8.3 Excerpt from ‘Kindred Britain’ version 1.0 © 2013 Nicholas Jenkins – designed by Scott Murray, powered by SUL-CIDR 249 8.4 Excerpt from ‘The Color of Debt: The Black Neighborhoods Where Collection Suits Hit Hardest’, by Al Shaw, Annie Waldman and Paul Kiel (ProPublica) 250 8.5 Excerpt from ‘Bloomberg Billionaires’, by Bloomberg Visual Data (Design and development), Lina Chen and Anita Rundles (Illustration) 251 8.6 Excerpt from ‘Gender Pay Gap US?’, by David McCandless, Miriam Quick (Research) and Philippa Thomas (Design) 251 8.7 Excerpt from ‘Holdouts Find Cheapest Super Bowl Tickets Late in the Game’, by Alex Tribou, David Ingold and Jeremy Diamond (Bloomberg Visual Data) 252 8.8 Excerpt from ‘The Life Cycle of Ideas’, by Accurat 252 8.9 Mizzou’s Racial Gap Is Typical On College Campuses, by FiveThirtyEight 253 8.10 Excerpt from ‘The Infographic History of the World’, Harper Collins (2013); by Valentina D’Efilippo (co-author and designer);

14

James Ball (co-author and writer); Data source: The Polynational War Memorial, 2012 254 8.11 Twitter NYC: A Multilingual Social City, by James Cheshire, Ed Manley, John Barratt, and Oliver O’Brien 255 8.12 Excerpt from ‘US Gun Deaths’, by Periscopic 255 8.13 Image taken from Wealth Inequality in America, by YouTube user ‘Politizane’ (www.youtube.com/watch?v=QPKKQnijnsM) 256 9.1 HSL Colour Cylinder: Image from Wikimedia Commons published under the Creative Commons Attribution-Share Alike 3.0 Unported license 265 9.2 Colour Hue Spectrum 265 9.3 Colour Saturation Spectrum 266 9.4 Colour Lightness Spectrum 266 9.5 Excerpt from ‘Executive Pay by the Numbers’, by Karl Russell (The New York Times) 267 9.6 How Nations Fare in PhDs by Sex Interactive, by Periscopic; Research by Amanda Hobbs; Published in Scientific American 268 9.7 How Long Will We Live – And How Well? by Bonnie Berkowitz, Emily Chow and Todd Lindeman (The Washington Post) 268 9.8 Charting the Beatles: Song Structure, by Michael Deal 269 9.9 Photograph of MyCuppa mug, by Suck UK (www.suck.uk.com/products/mycuppamugs/) 269 9.10 Example of a Stacked Bar Chart Based on Ordinal Data 270 9.11 Rim Fire – The Extent of Fire in the Sierra Nevada Range and Yosemite National Park, 2013: NASA Earth Observatory images, by Robert Simmon 270 9.12 What are the Current Electricity Prices in Switzerland [Translated], by Interactive things for NZZ (the Neue Zürcher Zeitung) 271 9.13 Excerpt from ‘Obama’s Health Law: Who Was Helped Most’, by Kevin Quealy and Margot Sanger-Katz (The New York Times) 272 9.14 Daily Indego Bike Share Station Usage, by Randy Olson (@randal_olson) (http://www.randalolson.com/2015/09/05/visualizing-indego-bike- share-usage-patterns-in-philadelphia-part-2/) 272 9.15 Battling Infectious Diseases in the 20th Century: The Impact of Vaccines, by Graphics Department (Wall Street Journal) 273 9.16 Highest Max Temperatures in Australia (1st to 14th January 2013), Produced by the Australian Government Bureau of

15

Meteorology 274 9.17 State of the Polar Bear, by Periscopic 275 9.18 Excerpt from Geography of a Recession by Graphics Department (The New York Times) 275 9.19 Fewer Women Run Big Companies Than Men Named John, by Justin Wolfers (The New York Times) 276 9.20 NYPD, Council Spar Over More Officers by Graphics Department (Wall Street Journal) 277 9.21 Excerpt from a Football Player Dashboard 277 9.22 Elections Performance Index, The Pew Charitable Trusts © 2014 278 9.23 Art in the Age of Mechanical Reproduction: Walter Benjamin by Stefanie Posavec 279 9.24 Casualties, by Stamen, published by CNN 279 9.25 First Fatal Accident in Spain on a High-speed Line [Translated], by Rodrigo Silva, Antonio Alonso, Mariano Zafra, Yolanda Clemente and Thomas Ondarra (El Pais) 280 9.26 Lunge Feeding, by Jonathan Corum (The New York Times); whale illustration by Nicholas D. Pyenson 281 9.27 Examples of Common Background Colour Tones 281 9.28 Excerpt from NYC Street Trees by Species, by Jill Hubley 284 9.29 Demonstrating the Impact of Red-green Colour Blindness (deuteranopia) 286 9.30 Colour-blind Friendly Alternatives to Green and Red 287 9.31 Excerpt from, ‘Pyschotherapy in The Arctic’, by Andy Kirk 289 9.32 Wind Map, by Fernanda Viégas and Martin Wattenberg 289 10.1 City of Anarchy, by Simon Scarr (South China Morning Post) 294 10.2 Wireframe Sketch, by Giorgia Lupi for ‘Nobels no degree’ by Accurat 295 10.3 Example of the Small Multiples Technique 296 10.4 The Glass Ceiling Persists Redesign, by Francis Gagnon (ChezVoila.com) based on original by S. Culp (Reuters Graphics) 297 10.5 Fast-food Purchasers Report More Demands on Their Time, by Economic Research Service (USDA) 297 10.6 Stalemate, by Graphics Department (Wall Street Journal) 297 10.7 Nobels No Degrees, by Accurat 298 10.8 Kasich Could Be The GOP’s Moderate Backstop, by FiveThirtyEight 298

16

10.9 On Broadway, by Daniel Goddemeyer, Moritz Stefaner, Dominikus Baur, and Lev Manovich 299 10.10 ER Wait Watcher: Which Emergency Room Will See You the Fastest? by Lena Groeger, Mike Tigas and Sisi Wei (ProPublica) 300 10.11 Rain Patterns, by Jane Pong (South China Morning Post) 300 10.12 Excerpt from ‘Pyschotherapy in The Arctic’, by Andy Kirk 301 10.13 Gender Pay Gap US, by David McCandless, Miriam Quick (Research) and Philippa Thomas (Design) 301 10.14 The Worst Board Games Ever Invented, by FiveThirtyEight 303 10.15 From Millions, Billions, Trillions: Letters from Zimbabwe, 2005−2009, a book written and published by Catherine Buckle (2014), table design by Graham van de Ruit (pg. 193) 303 10.16 List of Chart Structures 304 10.17 Illustrating the Effect of Truncated Bar Axis Scales 305 10.18 Excerpt from ‘Doping under the Microscope’, by S. Scarr and W. Foo (Reuters Graphics) 306 10.19 Record-high 60% of Americans Support Same-sex Marriage, by Gallup 306 10.20 Images from Wikimedia Commons, published under the Creative Commons Attribution-Share Alike 3.0 Unported license 308 11.1–7 The Pursuit of Faster’ by Andy Kirk and Andrew Witherley 318–324

17

Acknowledgements

This book has been made possible thanks to the unwavering support of my incredible wife, Ellie, and the endless encouragement from my Mum and Dad, the rest of my brilliant family and my super group of friends.

From a professional standpoint I also need to acknowledge the fundamental role played by the hundreds of visualisation practitioners (no matter under what title you ply your trade) who have created such a wealth of brilliant work from which I have developed so many of my convictions and formed the basis of so much of the content in this book. The people and organisations who have provided me with permission to use their work are heroes and I hope this book does their rich talent justice.

18

About the Author

Andy Kirk is a freelance data visualisation specialist based in Yorkshire, UK. He is a visualisation design consultant, training provider, teacher, researcher, author, speaker and editor of the award-winning website visualisingdata.com After graduating from Lancaster University in 1999 with a BSc (hons) in Operational Research, Andy held a variety of business analysis and information management positions at organisations including West Yorkshire Police and the University of Leeds. He discovered data visualisation in early 2007 just at the time when he was shaping up his proposal for a Master’s (MA) Research Programme designed for members of staff at the University of Leeds. On completing this programme with distinction, Andy’s passion for the subject was unleashed. Following his graduation in December 2009, to continue the process of discovering and learning the subject he launched visualisingdata.com, a blogging platform that would chart the ongoing development of the data visualisation field. Over time, as the field has continued to grow, the site too has reflected this, becoming one of the most popular in the field. It features a wide range of fresh content profiling the latest projects and contemporary techniques, discourse about practical and theoretical matters, commentary about key issues, and collections of valuable references and resources. In 2011 Andy became a freelance professional focusing on data visualisation consultancy and training workshops. Some of his clients include CERN, Arsenal FC, PepsiCo, Intel, Hershey, the WHO and McKinsey. At the time of writing he has delivered over 160 public and private training events across the UK, Europe, North America, Asia, South Africa and Australia, reaching well over 3000 delegates. In addition to training workshops Andy also has two academic teaching positions. He joined the highly respected Maryland Institute College of Art (MICA) as a visiting lecturer in 2013 and has been teaching a module on the Information Visualisation Master’s Programme since its inception. In January 2016, he began teaching a data visualisation module as part of the MSc in Business Analytics at the Imperial College Business School in London.

19

Between 2014 and 2015 Andy was an external consultant on a research project called ‘Seeing Data’, funded by the Arts & Humanities Research Council and hosted by the University of Sheffield. This study explored the issues of data visualisation literacy among the general public and, among many things, helped to shape an understanding of the human factors that affect visualisation literacy and the effectiveness of design.

20

Introduction

I.1 The Quest Begins In his book The Seven Basic Plots, author Christopher Booker investigated the history of telling stories. He examined the structures used in biblical teachings and historical myths through to contemporary storytelling devices used in movies and TV. From this study he found seven common themes that, he argues, can be identifiable in any form of story.

One of these themes was ‘The Quest’. Booker describes this as revolving around a main protagonist who embarks on a journey to acquire a treasured object or reach an important destination, but faces many obstacles and temptations along the way. It is a theme that I feel shares many characteristics with the structure of this book and the nature of data visualisation.

You are the central protagonist in this story in the role of the data visualiser. The journey you are embarking on involves a route along a design workflow where you will be faced with a wide range of different conceptual, practical and technical challenges. The start of this journey will be triggered by curiosity, which you will need to define in order to accomplish your goals. From this origin you will move forward to initiating and planning your work, defining the dimensions of your challenge. Next, you will begin the heavy lifting of working with data, determining what qualities it contains and how you might share these with others. Only then will you be ready to take on the design stage. Here you will be faced with the prospect of handling a spectrum of different design options that will require creative and rational thinking to resolve most effectively.

The multidisciplinary nature of this field offers a unique opportunity and challenge. Data visualisation is not an especially difficult capability to acquire, it is largely a game of decisions. Making better decisions will be your goal but sometimes clear decisions will feel elusive. There will be occasions when the best choice is not at all visible and others when there will be many seemingly equal viable choices. Which one to go with? This book aims to be your guide, helping you navigate efficiently through these

21

difficult stages of your journey.

You will need to learn to be flexible and adaptable, capable of shifting your approach to suit the circumstances. This is important because there are plenty of potential villains lying in wait looking to derail progress. These are the forces that manifest through the imposition of restrictive creative constraints and the pressure created by the relentless ticking clock of timescales. Stakeholders and audiences will present complex human factors through the diversity of their needs and personal traits. These will need to be astutely accommodated. Data, the critical raw material of this process, will dominate your attention. It will frustrate and even disappoint at times, as promises of its treasures fail to materialise irrespective of the hard work, love and attention lavished upon it.

Your own characteristics will also contribute to a certain amount of the villainy. At times, you will find yourself wrestling with internal creative and analytical voices pulling against each other in opposite directions. Your excitably formed initial ideas will be embraced but will need taming. Your inherent tastes, experiences and comforts will divert you away from the ideal path, so you will need to maintain clarity and focus.

The central conflict you will have to deal with is the notion that there is no perfect in data visualisation. It is a field with very few ‘always’ and ‘nevers’. Singular solutions rarely exist. The comfort offered by the rules that instruct what is right and wrong, good and evil, has its limits. You can find small but legitimate breaking points with many of them. While you can rightly aspire to reach as close to perfect as possible, the attitude of aiming for good enough will often indeed be good enough and fundamentally necessary.

In accomplishing the quest you will be rewarded with competency in data visualisation, developing confidence in being able to judge the most effective analytical and design solutions in the most efficient way. It will take time and it will need more than just reading this book. It will also require your ongoing effort to learn, apply, reflect and develop. Each new data visualisation opportunity poses a new, unique challenge. However, if you keep persevering with this journey the possibility of a happy ending will increase all the time.

I.2 Who is this Book Aimed at? 22

The primary challenge one faces when writing a book about data visualisation is to determine what to leave in and what to leave out. Data visualisation is big. It is too big a subject even to attempt to cover it all, in detail, in one book. There is no single book to rule them all because there is no one book that can cover it all. Each and every one of the topics covered by the chapters in this book could (and, in several cases, do) exist as whole books in their own right.

The secondary challenge when writing a book about data visualisation is to decide how to weave all the content together. Data visualisation is not rocket science; it is not an especially complicated discipline. Lots of it, as you will see, is rooted in common sense. It is, however, certainly a complex subject, a semantic distinction that will be revisited later. There are lots of things to think about and decide on, as well as many things to do and make. Creative and analytical sensibilities blend with artistic and scientific judgments. In one moment you might be checking the statistical rigour of your calculations, in the next deciding which tone of orange most elegantly contrasts with an 80% black. The complexity of data visualisation manifests itself through how these different ingredients, and many more, interact, influence and intersect to form the whole.

The decisions I have made in formulating this book‘s content have been shaped by my own process of learning about, writing about and practising data visualisation for, at the time of writing, nearly a decade. Significantly – from the perspective of my own development – I have been fortunate to have had extensive experience designing and delivering training workshops and postgraduate teaching. I believe you only truly learn about your own knowledge of a subject when you have to explain it and teach it to others.

I have arrived at what I believe to be an effective and proven pedagogy that successfully translates the complexities of this subject into accessible, practical and valuable form. I feel well qualified to bridge the gap between the large population of everyday practitioners, who might identify themselves as beginners, and the superstar technical, creative and academic minds that are constantly pushing forward our understanding of the potential of data visualisation. I am not going to claim to belong to that latter cohort, but I have certainly been the former – a beginner – and most of my working hours are spent helping other beginners start their journey. I know the things that I would have valued when I was starting out and I

23

know how I would have wished them to be articulated and presented for me to develop my skills most efficiently.

There is a large and growing library of fantastic books offering many different theoretical and practical viewpoints on the subject of data visualisation. My aim is to bring value to this existing collection of work by taking on a particular perspective that is perhaps under-represented in other texts – exploring the notion and practice of a visualisation design process. As I have alluded to in the opening, the central premise of this book is that the path to mastering data visualisation is achieved by making better decisions: effective choices, efficiently made. The book’s central goal is to help develop your capability and confidence in facing these decisions.

Just as a single book cannot cover the whole of this subject, it stands that a single book cannot aim to address directly the needs of all people doing data visualisation. In this section I am going to run through some of the characteristics that shape the readers to whom this book is primarily targeted. I will also put into context the content the book will and will not cover, and why. This will help manage your expectations as the reader and establish its value proposition compared with other titles.

Domain and Duties The core audiences for whom this book has been primarily written are undergraduate and postgraduate-level students and early career researchers from social science subjects. This reflects a growing number of people in higher education who are interested in and need to learn about data visualisation.

Although aimed at social sciences, the content will also be relevant across the spectrum of academic disciplines, from the arts and humanities right through to the formal and natural sciences: any academic duty where there is an emphasis on the use of quantitative and qualitative methods in studies will require an appreciation of good data visualisation practices. Where statistical capabilities are relevant so too is data visualisation.

Beyond academia, data visualisation is a discipline that has reached mainstream consciousness with an increasing number of professionals and organisations, across all industry types and sizes, recognising the

24

importance of doing it well for both internal and external benefit. You might be a market researcher, a librarian or a data analyst looking to enhance your data capabilities. Perhaps you are a skilled graphic designer or web developer looking to take your portfolio of work into a more data- driven direction. Maybe you are in a managerial position and not directly involved in the creation of visualisation work, but you need to coordinate or commission others who will be. You require awareness of the most efficient approaches, the range of options and the different key decision points. You might be seeking generally to improve the sophistication of the language you use around commissioning visualisation work and to have a better way of expressing and evaluating work created for you.

Basically, anyone who is involved in whatever capacity with the analysis and visual communication of data as part of their professional duties will need to grasp the demands of data visualisation and this book will go some way to supporting these needs.

Subject Neutrality One of the important aspects of the book will be to emphasise that data visualisation is a portable practice. You will see a broad array of examples of work from different industries, covering very different topics. What will become apparent is that visualisation techniques are largely subject-matter neutral: a line chart that displays the ebb and flow of favourable opinion towards a politician involves the same techniques as using a line chart to show how a stock has changed in value over time or how peak temperatures have changed across a season in a given location. A line chart is a line chart, regardless of the subject matter. The context of the viewers (such as their needs and their knowledge) and the specific meaning that can be drawn will inevitably be unique to each setting, but the role of visualisation itself is adaptable and portable across all subject areas.

Data visualisation is an entirely global concern, not focused on any defined geographic region. Although the English language dominates the written discourse (books, websites) about this subject, the interest in it and visible output from across the globe are increasing at a pace. There are cultural matters that influence certain decisions throughout the design process, especially around the choices made for colour usage, but otherwise it is a discipline common to all.

25

Level and Prerequisites The coverage of this book is intended to serve the needs of beginners and those with intermediate capability. For most people, this is likely to be as far as they might ever need to go. It will offer an accessible route for novices to start their learning journey and, for those already familiar with the basics, there will be content that will hopefully contribute to fine- tuning their approaches.

For context, I believe the only distinction between beginner and intermediate is one of breadth and depth of critical thinking rather than any degree of difficulty. The more advanced techniques in visualisation tend to be associated with the use of specific technologies for handling larger, complex datasets and/or producing more bespoke and feature-rich outputs.

This book is therefore not aimed at experienced or established visualisation practitioners. There may be some new perspectives to enrich their thinking, some content that will confirm and other content that might constructively challenge their convictions. Otherwise, the coverage in this book should really echo the practices they are likely to be already observing.

As I have already touched on, data visualisation is a genuinely multidisciplinary field. The people who are active in this field or profession come from all backgrounds – everyone has a different entry point and nobody arrives with all constituent capabilities. It is therefore quite difficult to define just what are the right type and level of pre- existing knowledge, skills or experiences for those learning about data visualisation. As each year passes, the savvy-ness of the type of audience this book targets will increase, especially as the subject penetrates more into the mainstream. What were seen as bewilderingly new techniques several years ago are now commonplace to more people.

That said, I think the following would be a fair outline of the type and shape of some of the most important prerequisite attributes for getting the most out of this book:

Strong numeracy is necessary as well as a familiarity with basic statistics. While it is reasonable to assume limited prior knowledge of data

26

visualisation, there should be a strong desire to want to learn it. The demands of learning a craft like data visualisation take time and effort; the capabilities will need nurturing through ongoing learning and practice. They are not going to be achieved overnight or acquired alone from reading this book. Any book that claims to be able magically to inject mastery through just reading it cover to cover is over-promising and likely to under-deliver. The best data visualisers possess inherent curiosity. You should be the type of person who is naturally disposed to question the world around them or can imagine what questions others have. Your instinct for discovering and sharing answers will be at the heart of this activity. There are no expectations of your having any prior familiarity with design principles, but a desire to embrace some of the creative aspects presented in this book will heighten the impact of your work. Unlock your artistry! If you are somebody with a strong creative flair you are very fortunate. This book will guide you through when and crucially when not to tap into this sensibility. You should be willing to increase the rigour of your analytical decision making and be prepared to have your creative thinking informed more fundamentally by data rather than just instinct. A range of technical skills covering different software applications, tools and programming languages is not expected for this book, as I will explain next, but you will ideally have some knowledge of basic Excel and some experience of working with data.

I.3 Getting the Balance

Handbook vs Tutorial Book The description of this book as being a ‘handbook’ positions it as being of practical help and presented in accessible form. It offers direction with comprehensive reference – more of a city guidebook for a tourist than an instruction manual to fix a washing machine. It will help you to know what things to think about, when to think about them, what options exist and how best to resolve all the choices involved in any data-driven design.

Technology is the key enabler for working with data and creating

27

visualisation design outputs. Indeed, apart from a small proportion of artisan visualisation work that is drawn by hand, the reliance on technology to create visualisation work is an inseparable necessity. For many there is a understandable appetite for step-by-step tutorials that help them immediately to implement data visualisation techniques via existing and new tools.

However, writing about data visualisation through the lens of selected tools is a bit of a minefield, given the diversity of technical options out there and the mixed range of skills, access and needs. I greatly admire those people who have authored tutorial-based texts because they require astute judgement about what is the right level, structure and scope.

The technology space around visualisation is characterised by flux. There are the ongoing changes with the enhancement of established tools as well as a relatively high frequency of new entrants offset by the decline of others. Some tools are proprietary, others are open source; some are easier to learn, others require a great deal of understanding before you can even consider embarking on your first chart. There are many recent cases of applications or services that have enjoyed fleeting exposure before reaching a plateau: development and support decline, the community of users disperses and there is a certain expiry of value. Deprecation of syntax and functions in programming languages requires the perennial updating of skills.

All of this perhaps paints a rather more chaotic picture than is necessarily the case but it justifies the reasons why this book does not offer teaching in the use of any tools. While tutorials may be invaluable to some, they may also only be mildly interesting to others and possibly of no value to most. Tools come and go but the craft remains. I believe that creating a practical, rather than necessarily a technical, text that focuses on the underlying craft of data visualisation with a tool-agnostic approach offers an effective way to begin learning about the subject in appropriate depth. The content should be appealing to readers irrespective of the extent of their technical knowledge (novice to advanced technicians) and specific tool experiences (e.g. knowledge of Excel, Tableau, Adobe Illustrator).

There is a role for all book types. Different people want different sources of insight at different stages in their development. If you are seeking a text that provides in-depth tutorials on a range of tools or pages of programmatic instruction, this one will not be the best choice. However, if

28

you consult only tutorial-related books, the chances are you will likely fall short on the fundamental critical thinking that will be needed in the longer term to get the most out of the tools with which you develop strong skills.

To substantiate the book’s value, the digital companion resources to this book will offer a curated, up-to-date collection of visualisation technology resources that will guide you through the most common and valuable tools, helping you to gain a sense of what their roles are and where these fit into the design workflow. Additionally, there will be recommended exercises and many further related digital materials available for exploring.

Useful vs Beautiful Another important distinction to make is that this book is not intended to be seen as a beauty pageant. I love flicking through those glossy ‘coffee table’ books as much as the next person; such books offer great inspiration and demonstrate some of the finest work in the field. This book serves a very different purpose. I believe that, as a beginner or relative beginner on this learning journey, the inspiration you need comes more from understanding what is behind the thinking that makes these amazing works succeed and others not.

My desire is to make this the most useful text available, a reference that will spend more time on your desk than on your bookshelf. To be useful is to be used. I want the pages to be dog-eared. I want to see scribbles and annotated notes made across its pages and key passages underlined. I want to see sticky labels peering out above identified pages of note. I want to see creases where pages have been folded back or a double-page spread that has been weighed down to keep it open. In time I even want its cover reinforced with wallpaper or wrapping paper to ensure its contents remain bound together. There is every intention of making this an elegantly presented and packaged book but it should not be something that invites you to ‘look, but don’t touch’.

Pragmatic vs Theoretical The content of this book has been formed through many years of absorbing knowledge from all manner of books, generations of academic papers, thousands of web articles, hundreds of conference talks, endless online and

29

personal discussions, and lots of personal practice. What I present here is a pragmatic translation and distillation of what I have learned down the years.

It is not a deeply academic or theoretical book. Where theoretical context and reference is relevant it will be signposted as I do want to ground this book in as much evidenced-based content as possible; it is about judging what is going to add most value. Experienced practitioners will likely have an appetite for delving deeper into theoretical discourse and the underlying sciences that intersect in this field but that is beyond the scope of this particular text.

Take the science of visual perception, for example. There is no value in attempting to emulate what has already been covered by other books in greater depth and quality than I could achieve. Once you start peeling back the many different layers of topics like visual and cognitive science the boundaries of your interest and their relevance to data visualisation never seem to arrive. You get swallowed up by the depth of these subjects. You realise that you have found yourself learning about what the very concept of light and sight is and at that point your brain begins to ache (well, mine does at least), especially when all you set out to discover was if a bar chart would be better than a pie chart.

An important reason for giving greater weight to pragmatism is because of people: people are the makers, the stakeholders, the audiences and the critics in data visualisation. Although there are a great deal of valuable research-driven concepts concerning data visualisation, their practical application can be occasionally at odds with the somewhat sanitised and artificial context of the research methods employed. To translate them into real-world circumstances can sometimes be easier said than done as the influence of human factors can easily distort the significance of otherwise robust ideas.

I want to remove the burden from you as a reader having to translate relevant theoretical discourse into applicable practice. Critical thinking will therefore be the watchword, equipping you with the independence of thought to decide rationally for yourself what the solutions are that best fit your context, your data, your message and your audience. To do this you will need an appreciation of all the options available to you (the different things you could do) and a reliable approach for critically determining what choices you should make (the things you will do and why).

30

Contemporary vs Historical This book is not going to look too far back into the past. We all respect the ancestors of this field, the great names who, despite primitive means, pioneered new concepts in the visual display of statistics to shape the foundations of the field being practised today. The field’s lineage is decorated by the influence of William Playfair’s first ever bar chart, Charles Joseph Minard’s famous graphic about Napoleon’s Russian campaign, Florence Nightingale’s Coxcomb plot and John Snow’s cholera map. These are some of the totemic names and classic examples that will always be held up as the ‘firsts’. Of course, to many beginners in the field, this historical context is of huge interest. However, again, this kind of content has already been superbly covered by other texts on more than enough occasions. Time to move on.

I am not going to spend time attempting to enlighten you about how we live in the age of ‘Big Data’ and how occupations related to data are or will be the ‘sexiest jobs’ of our time. The former is no longer news, the latter claim emerged from a single source. I do not want to bloat this book with the unnecessary reprising of topics that have been covered at length elsewhere. There is more valuable and useful content I want you to focus your time on.

The subject matter, the ideas and the practices presented here will hopefully not date a great deal. Of course, many of the graphic examples included in the book will be surpassed by newer work demonstrating similar concepts as the field continues to develop. However, their worth as exhibits of a particular perspective covered in the text should prove timeless. As more research is conducted in the subject, without question there will be new techniques, new concepts, new empirically evidenced principles that emerge. Maybe even new rules. There will be new thought- leaders, new sources of reference, new visualisers to draw insight from. New tools will be created, existing tools will expire. Some things that are done and can only be done by hand as of today may become seamlessly automated in the near future. That is simply the nature of a fast-growing field. This book can only be a line in the sand.

Analysis vs Communication

31

A further important distinction to make concerns the subtle but significant difference between visualisations which are used for analysis and visualisations used for communication.

Before a visualiser can confidently decide what to communicate to others, he or she needs to have developed an intimate understanding of the qualities and potential of the data. This is largely achieved through exploratory data analysis. Here, the visualiser and the viewer are the same person. Through visual exploration, different interrogations can be pursued ‘on the fly’ to unearth confirmatory or enlightening discoveries about what insights exist.

Visualisation techniques used for analysis will be a key component of the journey towards creating visualisation for communication but the practices involved differ. Unlike visualisation for communication, the techniques used for visual analysis do not have to be visually polished or necessarily appealing. They are only serving the purpose of helping you to truly learn about your data. When a data visualisation is being created to communicate to others, many careful considerations come into play about the requirements and interests of the intended or expected audience. This has a significant influence on many of the design decisions you make that do not exist alone with visual analysis.

Exploratory data analysis is a huge and specialist subject in and of itself. In its most advanced form, working efficiently and effectively with large complex data, topics like ‘machine learning’, using self-learning algorithms to help automate and assist in the discovery of patterns in data, become increasingly relevant. For the scope of this book the content is weighted more towards methods and concerns about communicating data visually to others. If your role is in pure data science or statistical analysis you will likely require a deeper treatment of the exploratory data analysis topic than this book can reasonably offer. However, Chapter 4 will cover the essential elements in sufficient depth for the practical needs of most people working with data.

Print vs Digital The opportunity to supplement the print version of this book with an e- book and further digital companion resources helps to cushion the agonising decisions about what to leave out. This text is therefore

32

enhanced by access to further digital resources, some of which are newly created, while others are curated references from the endless well of visualisation content on the Web. Included online (book.visualisingdata.com) will be:

a completed case-study project that demonstrates the workflow activities covered in this book, including full write-ups and all related digital materials; an extensive and up-to-date catalogue of over 300 data visualisation tools; a curated collection of tutorials and resources to help develop your confidence with some of the most common and valuable tools; practical exercises designed to embed the learning from each chapter; further reading resources to continue learning about the subjects covered in each chapter.

I.4 Objectives Before moving on to an outline of the book’s contents, I want to share four key objectives that I hope to accomplish for you by the final chapter. These are themes that will run through the entire text: challenge, enlighten, equip and inspire.

To challenge you I will be encouraging you to recognise that your current thinking about visualisation may need to be reconsidered, both as a creator and as a consumer. We all arrive in visualisation from different subject and domain origins and with that comes certain baggage and prior sensibilities that can distort our perspectives. I will not be looking to eliminate these, rather to help you harness and align them with other traits and viewpoints.

I will ask you to relentlessly consider the diverse decisions involved in this process. I will challenge your convictions about what you perceive to be good or bad, effective or ineffective visualisation choices: arbitrary choices will be eliminated from your thinking. Even if you are not necessarily a beginner, I believe the content you read in this book will make you question some of your own perspectives and assumptions. I will encourage you to reflect on your previous work, asking you to consider how and why you have designed visualisations in the way that you have: where do you need to improve? What can you do better?

33

It is not just about creating visualisations, I will also challenge your approach to reading visualisations. This is not something you might usually think much about, but there is an important role for more tactical approaches to consuming visualisations with greater efficiency and effectiveness.

To enlighten you will be to increase your awareness of the possibilities in data visualisation. As you begin your discovery of data visualisation you might not be aware of the whole: you do not entirely know what options exist, how they are connected and how to make good choices. Until you know, you don’t know – that is what the objective of enlightening is all about.

As you will discover, there is a lot on your plate, much to work through. It is not just about the visible end-product design decisions. Hidden beneath the surface are many contextual circumstances to weigh up, decisions about how best to prepare your data, choices around the multitude of viable ways of slicing those data up into different angles of analysis. That is all before you even reach the design stage, where you will begin to consider the repertoire of techniques for visually portraying your data – the charts, the interactive features, the colours and much more besides.

This book will broaden your visual vocabulary to give you more ways of expressing your data visually. It will enhance the sophistication of your decision making and of visual language for any of the challenges you may face.

To equip is to ensure you have robust tactics for managing your way through the myriad options that exist in data visualisation. The variety it offers makes for a wonderful prospect but, equally, introduces the burden of choice. This book aims to make the challenge of undertaking data visualisation far less overwhelming, breaking down the overall prospect into smaller, more manageable task chunks.

The structure of this book will offer a reliable and flexible framework for thinking, rather than rules for learning. It will lead to better decisions. With an emphasis on critical thinking you will move away from an over- reliance on gut feeling and taste. To echo what I mentioned earlier, its role as a handbook will help you know what things to think about, when to think about them and how best to resolve all the thinking involved in any data-driven design challenge you meet.

34

To inspire is to give you more than just a book to read. It is the opening of a door into a subject to inspire you to step further inside. It is about helping you to want to continue to learn about it and expose yourself to as much positive influence as possible. It should elevate your ambition and broaden your capability.

It is a book underpinned by theory but dominated by practical and accessible advice, including input from some of the best visualisers in the field today. The range of print and digital resources will offer lots of supplementary material including tutorials, further reading materials and suggested exercises. Collectively this will hopefully make it one of the most comprehensive, valuable and inspiring titles out there.

I.5 Chapter Contents The book is organised into four main parts (A, B, C and D) comprising eleven chapters and preceded by the ‘Introduction’ sections you are reading now.

Each chapter opens with an introductory outline that previews the content to be covered and provides a bridge between consecutive chapters. In the closing sections of each chapter the most salient learning points will be summarised and some important, practical tips and tactics shared. As mentioned, online there will be collections of practical exercises and further reading resources recommended to substantiate the learning from the chapter.

Throughout the book you will see sidebar captions that will offer relevant references, aphorisms, good habits and practical tips from some of the most influential people in the field today.

Introduction This introduction explains how I have attempted to make sense of the complexity of the subject, outlining the nature of the audience I am trying to reach, the key objectives, what topics the book will be covering and not covering, and how the content has been organised.

35

Part A: Foundations Part A establishes the foundation knowledge and sets up a key reference of understanding that aids your thinking across the rest of the book. Chapter 1 will be the logical starting point for many of you who are new to the field to help you understand more about the definitions and attributes of data visualisation. Even if you are not a complete beginner, the content of the chapter forms the terms of reference that much of the remaining content is based on. Chapter 2 prepares you for the journey through the rest of the book by introducing the key design workflow that you will be following.

Chapter 1: Defining Data Visualisation

Defining data visualisation: outlining the components of thinking that make up the proposed definition for data visualisation. The importance of conviction: presenting three guiding principles of good visualisation design: trustworthy, accessible and elegant. Distinctions and glossary: explaining the distinctions and overlaps with other related disciplines and providing a glossary of terms used in this book to establish consistency of language.

Chapter 2: Visualisation Workflow

The importance of process: describing the data visualisation design workflow, what it involves and why a process approach is required. The process in practice: providing some useful tips, tactics and habits that transcend any particular stage of the process but will best prepare you for success with this activity.

Part B: The Hidden Thinking Part B discusses the first three preparatory stages of the data visualisation design workflow. ‘The hidden thinking’ title refers to how these vital activities, that have a huge influence over the eventual design solution, are somewhat out of sight in the final output; they are hidden beneath the surface but completely shape what is visible. These stages represent the often neglected contextual definitions, data wrangling and editorial challenges that are so critical to the success or otherwise of any

36

visualisation work – they require a great deal of care and attention before you switch your attention to the design stage.

Chapter 3: Formulating Your Brief

What is a brief?: describing the value of compiling a brief to help initiate, define and plan the requirements of your work. Establishing your project’s context: defining the origin curiosity or motivation, identifying all the key factors and circumstances that surround your work, and defining the core purpose of your visualisation. Establishing your project’s vision: early considerations about the type of visualisation solution needed to achieve your aims and harnessing initial ideas about what this solution might look like.

Chapter 4: Working With Data

Data literacy: establishing a basic understanding with this critical literacy, providing some foundation understanding about datasets and data types and some observations about statistical literacy. Data acquisition: outlining the different origins of and methods for accessing your data. Data examination: approaches for acquainting yourself with the physical characteristics and meaning of your data. Data transformation: optimising the condition, content and form of your data fully to prepare it for its analytical purpose. Data exploration: developing deeper intimacy with the potential qualities and insights contained, and potentially hidden, within your data.

Chapter 5: Establishing Your Editorial Thinking

What is editorial thinking?: defining the role of editorial thinking in data visualisation. The influence of editorial thinking: explaining how the different dimensions of editorial thinking influence design choices.

Part C: Developing Your Design Solution

37

Part C is the main part of the book and covers progression through the data visualisation design and production stage. This is where your concerns switch from hidden thinking to visible thinking. The individual chapters in this part of the book cover each of the five layers of the data visualisation anatomy. They are treated as separate affairs to aid the clarity and organisation of your thinking, but they are entirely interrelated matters and the chapter sequences support this. Within each chapter there is a consistent structure beginning with an introduction to each design layer, an overview of the many different possible design options, followed by detailed guidance on the factors that influence your choices.

The production cycle: describing the cycle of development activities that take place during this stage, giving a context for how to work through the subsequent chapters in this part.

Chapter 6: Data Representation

Introducing visual encoding: an overview of the essentials of data representation looking at the differences and relationships between visual encoding and chart types. Chart types: a detailed repertoire of 49 different chart types, profiled in depth and organised by a taxonomy of chart families: categorical, hierarchical, relational, temporal, and spatial. Influencing factors and considerations: presenting the factors that will influence the suitability of your data representation choices.

Chapter 7: Interactivity

The features of interactivity:

Data adjustments: a profile of the options for interactively interrogating and manipulating data. View adjustments: a profile of the options for interactively configuring the presentation of data.

Influencing factors and considerations: presenting the factors that will influence the suitability of your interactivity choices.

Chapter 8: Annotation

38

The features of annotation:

Project annotation: a profile of the options for helping to provide viewers with general explanations about your project. Chart annotation: a profile of the annotated options for helping to optimise viewers’ understanding your charts.

Influencing factors and considerations: presenting the factors that will influence the suitability of your annotation choices.

Chapter 9: Colour

The features of colour:

Data legibility: a profile of the options for using colour to represent data. Editorial salience: a profile of the options for using colour to direct the eye towards the most relevant features of your data. Functional harmony: a profile of the options for using colour most effectively across the entire visualisation design.

Influencing factors and considerations: presenting the factors that will influence the suitability of your colour choices.

Chapter 10: Composition

The features of composition:

Project composition: a profile of the options for the overall layout and hierarchy of your visualisation design. Chart composition: a profile of the options for the layout and hierarchy of the components of your charts.

Influencing factors and considerations: presenting the factors that will influence the suitability of your composition choices.

Part D: Developing Your Capabilities Part D wraps up the book’s content by reflecting on the range of capabilities required to develop confidence and competence with data

39

visualisation. Following completion of the design process, the multidisciplinary nature of this subject will now be clearly established. This final part assesses the two sides of visualisation literacy – your role as a creator and your role as a viewer – and what you need to enhance your skills with both.

Chapter 11: Visualisation Literacy

Viewing: Learning to see: learning about the most effective strategy for understanding visualisations in your role as a viewer rather than a creator. Creating: The capabilities of the visualiser: profiling the skill sets, mindsets and general attributes needed to master data visualisation design as a creator.

40

Part A Foundations

41

1 Defining Data Visualisation

This opening chapter will introduce you to the subject of data visualisation, defining what data visualisation is and is not. It will outline the different ingredients that make it such an interesting recipe and establish a foundation of understanding that will form a key reference for all of the decision making you are faced with.

Three core principles of good visualisation design will be presented that offer guiding ideals to help mould your convictions about distinguishing between effective and ineffective in data visualisation.

You will also see how data visualisation sits alongside or overlaps with other related disciplines, and some definitions about the use of language in this book will be established to ensure consistency in meaning across all chapters.

1.1 The Components of Understanding To set the scene for what is about to follow, I think it is important to start this book with a proposed definition for data visualisation (Figure 1.1). This definition offers a critical term of reference because its components and their meaning will touch on every element of content that follows in this book. Furthermore, as a subject that has many different proposed definitions, I believe it is worth clarifying my own view before going further:

Figure 1.1 A Definition for Data Visualisation

42

At first glance this might appear to be a surprisingly short definition: isn’t there more to data visualisation than that, you might ask? Can nine words sufficiently articulate what has already been introduced as an eminently complex and diverse discipline?

I have arrived at this after many years of iterations attempting to improve the elegance of my definition. In the past I have tried to force too many words and too many clauses into one statement, making it cumbersome and rather undermining its value. Over time, as I have developed greater clarity in my own convictions, I have in turn managed to establish greater clarity about what I feel is the real essence of this subject. The definition above is, I believe, a succinct and practically useful description of what the pursuit of visualisation is truly about. It is a definition that largely informs the contents of this book. Each chapter will aim to enlighten you about different aspects of the roles of and relationships between each component expressed. Let me introduce and briefly examine each of these one by one, explaining where and how they will be discussed in the book.

Firstly, data, our critical raw material. It might appear a formality to mention data in the definition for, after all, we are talking about data visualisation as opposed to, let’s say, cheese visualisation (though visualisation of data using cheese has happened, see Figure 1.2), but it needs to be made clear the core role that data has in the design process. Without data there is no visualisation; indeed there is no need for one. Data plays the fundamental role in this work, so you will need to give it your undivided attention and respect. You will discover in Chapter 4 the importance of developing an intimacy with your data to acquaint yourself with its physical properties, its meaning and its potential qualities.

43

Figure 1.2 Per Capita Cheese Consumption in the US

Data is names, amounts, groups, statistical values, dates, comments, locations. Data is textual and numeric in format, typically held in datasets in table form, with rows of records and columns of different variables.

This tabular form of data is what we will be considering as the raw form of data. Through tables, we can look at the values contained to precisely read them as individual data points. We can look up values quite efficiently, scanning across many variables for the different records held. However, we cannot easily establish the comparative size and relationship between multiple data points. Our eyes and mind are not equipped to translate easily the textual and numeric values into quantitative and qualitative meaning. We can look at the data but we cannot really see it without the context of relationships that help us compare and contrast them effectively with other values. To derive understanding from data we need to see it represented in a different, visual form. This is the act of data representation.

This word representation is deliberately positioned near the front of the definition because it is the quintessential activity of data visualisation design. Representation concerns the choices made about the form in which your data will be visually portrayed: in lay terms, what chart or charts you will use to exploit the brain’s visual perception capabilities most effectively.

When data visualisers create a visualisation they are representing the data they wish to show visually through combinations of marks and attributes. Marks are points, lines and areas. Attributes are the appearance properties

44

of these marks, such as the size, colour and position. The recipe of these marks and their attributes, along with other components of apparatus, such as axes and gridlines, form the anatomy of a chart.

In Chapter 6 you will gain a deeper and more sophisticated appreciation of the range of different charts that are in common usage today, broadening your visual vocabulary. These charts will vary in complexity and composition, with each capable of accommodating different types of data and portraying different angles of analysis. You will learn about the key ingredients that shape your data representation decisions, explaining the factors that distinguish the effective from the ineffective choices.

Beyond representation choices, the presentation of data concerns all the other visible design decisions that make up the overall visualisation anatomy. This includes choices about the possible applications of interactivity, features of annotation, colour usage and the composition of your work. During the early stages of learning this subject it is sensible to partition your thinking about these matters, treating them as isolated design layers. This will aid your initial critical thinking. Chapters 7–10 will explore each of these layers in depth, profiling the options available and the factors that influence your decisions.

However, as you gain in experience, the interrelated nature of visualisation will become much more apparent and you will see how the overall design anatomy is entirely connected. For instance, the selection of a chart type intrinsically leads to decisions about the space and place it will occupy; an interactive control may be included to reveal an annotated caption; for any design property to be even visible to the eye it must possess a colour that is different from that of its background.

The goal expressed in this definition states that data visualisation is about facilitating understanding. This is very important and some extra time is required to emphasise why it is such an influential component in our thinking. You might think you know what understanding means, but when you peel back the surface you realise there are many subtleties that need to be acknowledged about this term and their impact on your data visualisation choices. Understanding ‘understanding’ (still with me?) in the context of data visualisation is of elementary significance.

When consuming a visualisation, the viewer will go through a process of understanding involving three stages: perceiving, interpreting and

45

comprehending (Figure 1.3). Each stage is dependent on the previous one and in your role as a data visualiser you will have influence but not full control over these. You are largely at the mercy of the viewer – what they know and do not know, what they are interested in knowing and what might be meaningful to them – and this introduces many variables outside of your control: where your control diminishes the influence and reliance on the viewer increases. Achieving an outcome of understanding is therefore a collective responsibility between visualiser and viewer.

These are not just synonyms for the same word, rather they carry important distinctions that need appreciating. As you will see throughout this book, the subtleties and semantics of language in data visualisation will be a recurring concern.

Figure 1.3 The Three Stages of Understanding

Let’s look at the characteristics of the different stages that form the process of understanding to help explain their respective differences and mutual dependencies.

Firstly, perceiving. This concerns the act of simply being able to read a chart. What is the chart showing you? How easily can you get a sense of the values of the data being portrayed?

Where are the largest, middle-sized and smallest values? What proportion of the total does that value hold? How do these values compare in ranking terms? To which other values does this have a connected relationship?

The notion of understanding here concerns our attempts as viewers to

46

efficiently decode the representations of the data (the shapes, the sizes and the colours) as displayed through a chart, and then convert them into perceived values: estimates of quantities and their relationships to other values.

Interpreting is the next stage of understanding following on from perceiving. Having read the charts the viewer now seeks to convert these perceived values into some form of meaning:

Is it good to be big or better to be small? What does it mean to go up or go down? Is that relationship meaningful or insignificant? Is the decline of that category especially surprising?

The viewer’s ability to form such interpretations is influenced by their pre- existing knowledge about the portrayed subject and their capacity to utilise that knowledge to frame the implications of what has been read. Where a viewer does not possess that knowledge it may be that the visualiser has to address this deficit. They will need to make suitable design choices that help to make clear what meaning can or should be drawn from the display of data. Captions, headlines, colours and other annotated devices, in particular, can all be used to achieve this.

Comprehending involves reasoning the consequence of the perceiving and interpreting stages to arrive at a personal reflection of what all this means to them, the viewer. How does this information make a difference to what was known about the subject previously?

Why is this relevant? What wants or needs does it serve? Has it confirmed what I knew or possibly suspected beforehand or enlightened me with new knowledge? Has this experience impacted me in an emotional way or left me feeling somewhat indifferent as a consequence? Does the context of what understanding I have acquired lead me to take action – such as make a decision or fundamentally change my behaviour – or do I simply have an extra grain of knowledge the consequence of which may not materialise until much later?

Over the page is a simple demonstration to further illustrate this process of understanding. In this example I play the role of a viewer working with a sample isolated chart (Figure 1.4). As you will learn throughout the design

47

chapters, a chart would not normally just exist floating in isolation like this one does, but it will serve a purpose for this demonstration.

Figure 1.4 shows a clustered bar chart that presents a breakdown of the career statistics for the footballer Lionel Messi during his career with FC Barcelona.

The process commences with perceiving the chart. I begin by establishing what chart type is being used. I am familiar with this clustered bar chart approach and so I quickly feel at ease with the prospect of reading its display: there is no learning for me to have to go through on this occasion, which is not always the case as we will see.

I can quickly assimilate what the axes are showing by examining the labels along the x- and y-axes and by taking the assistance provided by colour legend at the top. I move on to scanning, detecting and observing the general physical properties of the data being represented. The eyes and brain are working in harmony, conducting this activity quite instinctively without awareness or delay, noting the most prominent features of variation in the attributes of size, shape, colour and position.

Figure 1.4 Demonstrating the Process of Understanding

I look across the entire chart, identifying the big, small and medium values

48

(these are known as stepped magnitude judgements), and form an overall sense of the general value rankings (global comparison judgements). I am instinctively drawn to the dominant bars towards the middle/right of the chart, especially as I know this side of the chart concerns the most recent career performances. I can determine that the purple bar – showing goals – has been rising pretty much year-on-year towards a peak in 2011/12 and then there is a dip before recovery in his most recent season.

My visual system is now working hard to decode these properties into estimations of quantities (amounts of things) and relationships (how different things compare with each other). I focus on judging the absolute magnitudes of individual bars (one bar at a time). The assistance offered by the chart apparatus, such as the vertical axis (or y- axis) values and the inclusion of gridlines, is helping me more quickly estimate the quantities with greater assurance of accuracy, such as discovering that the highest number of goals scored was around 73.

I then look to conduct some relative higher/lower comparisons. In comparing the games and goals pairings I can see that three out of the last four years have seen the purple bar higher than the blue bar, in contrast to all the rest. Finally I look to establish proportional relationships between neighbouring bars, i.e. by how much larger one is compared with the next. In 2006/07 I can see the blue bar is more than twice as tall as the purple one, whereas in 2011/12 the purple bar is about 15% taller.

By reading this chart I now have a good appreciation of the quantities displayed and some sense of the relationship between the two measures, games and goals.

The second part of the understanding process is interpreting. In reality, it is not so consciously consecutive or delayed in relationship to the perceiving stage but you cannot get here without having already done the perceiving. Interpreting, as you will recall, is about converting perceived ‘reading’ into meaning. Interpreting is essentially about orientating your assessment of what you’ve read against what you know about the subject.

As I mentioned earlier, often a data visualiser will choose to – or have the opportunity to – share such insights via captions, chart overlays or summary headlines. As you will learn in Chapter 3, the visualisations that present this type of interpretation assistance are commonly described as offering an ‘explanatory’ experience. In this particular demonstration it is

49

an example of an ‘exhibitory’ experience, characterised by the absence of any explanatory features. It relies on the viewer to handle the demands of interpretation without any assistance.

As you will read about later, many factors influence how well different viewers will be able to interpret a visualisation. Some of the most critical include the level of interest shown towards the subject matter, its relevance and the general inclination, in that moment, of a viewer to want to read about that subject through a visualisation. It is also influenced by the knowledge held about a subject or the capacity to derive meaning from a subject even if a knowledge gap exists.

Returning to the sample chart, in order to translate the quantities and relationships I extracted from the perceiving stage into meaning, I am effectively converting the reading of value sizes into notions of good or bad and comparative relationships into worse than or better than etc. To interpret the meaning of this data about Lionel Messi I can tap into my passion for and knowledge of football. I know that for a player to score over 25 goals in a season is very good. To score over 35 is exceptional. To score over 70 goals is frankly preposterous, especially at the highest level of the game (you might find plenty of players achieving these statistics playing for the Dog and Duck pub team, but these numbers have been achieved for Barcelona in La Liga, the Champions League and other domestic cup competitions). I know from watching the sport, and poring over statistics like this for 30 years, that it is very rare for a player to score remotely close to a ratio of one goal per game played. Those purple bars that exceed the height of the blue bars are therefore remarkable. Beyond the information presented in the chart I bring knowledge about the periods when different managers were in charge of Barcelona, how they played the game, and how some organised their teams entirely around Messi’s talents. I know which other players were teammates across different seasons and who might have assisted or hindered his achievements. I also know his age and can mentally compare his achievements with the traditional football career arcs that will normally show a steady rise, peak, plateau, and then decline.

Therefore, in this example, I am not just interested in the subject but can bring a lot of knowledge to aid me in interpreting this analysis. That helps me understand a lot more about what this data means. For other people they might be passingly interested in football and know how to read what

50

is being presented, but they might not possess the domain knowledge to go deeper into the interpretation. They also just might not care. Now imagine this was analysis of, let’s say, an NHL ice hockey player (Figure 1.5) – that would present an entirely different challenge for me.

In this chart the numbers are irrelevant, just using the same chart as before with different labels. Assuming this was real analysis, as a sports fan in general I would have the capacity to understand the notion of a sportsperson’s career statistics in terms of games played and goals scored: I can read the chart (perceiving) that shows me this data and catch the gist of the angle of analysis it is portraying. However, I do not have sufficient domain knowledge of ice hockey to determine the real meaning and significance of the big–small, higher–lower value relationships. I cannot confidently convert ‘small’ into ‘unusual’ or ‘greater than’ into ‘remarkable’. My capacity to interpret is therefore limited, and besides I have no connection to the subject matter, so I am insufficiently interested to put in the effort to spend much time with any in-depth attempts at interpretation.

Figure 1.5 Demonstrating the Process of Understanding

Imagine this is now no longer analysis about sport but about the sightings in the wild of Winglets and Spungles (completely made up words). Once again I can still read the chart shown in Figure 1.6 but now I have

51

absolutely no connection to the subject whatsoever. No knowledge and no interest. I have no idea what these things are, no understanding about the sense of scale that should be expected for these sightings, I don’t know what is good or bad. And I genuinely don’t care either. In contrast, for those who do have a knowledge of and interest in the subject, the meaning of this data will be much more relevant. They will be able to read the chart and make some sense of the meaning of the quantities and relationships displayed.

To help with perceiving, viewers need the context of scale. To help with interpreting, viewers need the context of subject, whether that is provided by the visualiser or the viewer themself. The challenge for you and I as data visualisers is to determine what our audience will know already and what they will need to know in order to possibly assist them in interpreting the meaning. The use of explanatory captions, perhaps positioned in that big white space top left, could assist those lacking the knowledge of the subject, possibly offering a short narrative to make the interpretations – the meaning – clearer and immediately accessible.

We are not quite finished, there is one stage left. The third part of the understanding process is comprehending. This is where I attempt to form some concluding reasoning that translates into what this analysis means for me. What can I infer from the display of data I have read? How do I relate and respond to the insights I have drawn out as through interpretation? Does what I’ve learnt make a difference to me? Do I know something more than I did before? Do I need to act or decide on anything? How does it make me feel emotionally?

Figure 1.6 Demonstrating the Process of Understanding

52

Through consuming the Messi chart, I have been able to form an even greater appreciation of his amazing career. It has surprised me just how prolific he has been, especially having seen his ratio of goals to games, and I am particularly intrigued to see whether the dip in 2013/14 was a temporary blip or whether the bounce back in 2014/15 was the blip. And as he reaches his late 20s, will injuries start to creep in as they seem to do for many other similarly prodigious young talents, especially as he has been playing relentlessly at the highest level since his late teens?

My comprehension is not a dramatic discovery. There is no sudden inclination to act nor any need – based on what I have learnt. I just feel a heightened impression, formed through the data, about just how good and prolific Lionel Messi has been. For Barcelona fanatics who watch him play every week, they will likely have already formed this understanding. This kind of experience would only have reaffirmed what they already probably knew.

And that is important to recognise when it comes to managing expectations about what we hope to achieve amongst our viewers in terms of their final comprehending. One person’s ‘I knew that already’ is another person’s ‘wow’. For every ‘wow, I need to make some changes’ type of reflection there might be another ‘doesn’t affect me’. A compelling visualisation about climate change presented to Sylvie might affect her

53

significantly about the changes she might need to make in her lifestyle choices that might reduce her carbon footprint. For Robert, who is already familiar with the significance of this situation, it might have substantially less immediate impact – not indifference to the meaning of the data, just nothing new, a shrug of the shoulders. For James, the hardened sceptic, even the most indisputable evidence may have no effect; he might just not be receptive to altering his views regardless.

What these scenarios try to explain is that, from your perspective of the visualiser, this final stage of understanding is something you will have relatively little control over because viewers are people and people are complex. People are different and as such they introduce inconsistencies. You can lead a horse to water but you cannot make it drink: you cannot force a viewer to be interested in your work, to understand the meaning of a subject or get that person to react exactly how you would wish.

Visualising data is just an agent of communication and not a guarantor for what a viewer does with the opportunity for understanding that is presented. There are different flavours of comprehension, different consequences of understanding formed through this final stage. Many visualisations will be created with the ambition to simply inform, like the Messi graphic achieved for me, perhaps to add just an extra grain to the pile of knowledge a viewer has about a subject. Not every visualisation results in a Hollywood moment of grand discoveries, surprising insights or life-saving decisions. But that is OK, so long as the outcome fits with the intended purpose, something we will discuss in more depth in Chapter 3.

Furthermore, there is the complexity of human behaviour in how people make decisions in life. You might create the most compelling visualisation, demonstrating proven effective design choices, carefully constructed with very a specific audience type and need in mind. This might clearly show how a certain decision really needs to be taken by those in the audience. However, you cannot guarantee that the decision maker in question, while possibly recognising that there is a need to act, will be in a position to act, and indeed will know how to act.

It is at this point that one must recognise the ambitions and – more importantly – realise the limits of what data visualisation can achieve. Going back again, finally, to the components of the definition, all the reasons outlined above show why the term to facilitate is the most a visualiser can reasonably aspire to achieve.

54

It might feel like a rather tepid and unambitious aim, something of a cop- out that avoids scrutiny over the outcomes of our work: why not aim to ‘deliver’, ‘accomplish’, or do something more earnest than just ‘facilitate’? I deliberately use ‘facilitate’ because as we have seen we can only control so much. Design cannot change the world, it can only make it run a little smoother. Visualisers can control the output but not the outcome: at best we can expect to have only some influence on it.

1.2 The Importance of Conviction The key structure running through this book is a data visualisation design process. By following this process you will be able to decrease the size of the challenge involved in making good decisions about your design solution. The sequencing of the stages presented will help reduce the myriad options you have to consider, which makes the prospect of arriving at the best possible solution much more likely to occur.

Often, the design choices you need to make will be clear cut. As you will learn, the preparatory nature of the first three stages goes a long way to securing that clarity later in the design stage. On other occasions, plain old common sense is a more than sufficient guide. However, for more nuanced situations, where there are several potentially viable options presenting themselves, you need to rely on the guiding value of good design principles.

‘I say begin by learning about data visualisation’s “black and whites”, the rules, then start looking for the greys. It really then becomes quite a personal journey of developing your conviction.’ Jorge Camoes, Data Visualization Consultant

For many people setting out on their journey in data visualisation, the major influences that shape their early beliefs about data visualisation design tend to be influenced by the first authors they come across. Names like Edward Tufte, unquestionably one of the most important figures in this field whose ideas are still pervasive, represent a common entry point into the field, as do people like Stephen Few, David McCandless, Alberto Cairo, and Tamara Munzner, to name but a few. These are authors of prominent works that typically represent the first books purchased and

55

read by many beginners.

Where you go from there – from whom you draw your most valuable enduring guidance –will be shaped by many different factors: taste, the industry you are working in, the topics on which you work, the types of audiences you produce for. I still value much of what Tufte extols, for example, but find I can now more confidently filter out some of his ideals that veer towards impractical ideology or that do not necessarily hold up against contemporary technology and the maturing expectations of people.

‘My key guiding principle? Know the rules, before you break them.’ Gregor Aisch, Graphics Editor, The New York Times

The key guidance that now most helpfully shapes and supports my convictions comes from ideas outside the boundaries of visualisation design in the shape of the work of Dieter Rams. Rams was a German industrial and product designer who was most famously associated with the Braun company.

In the late 1970s or early 1980s, Rams was becoming concerned about the state and direction of design thinking and, given his prominent role in the industry, felt a responsibility to challenge himself, his own work and his own thinking against a simple question: ‘Is my design good design?’. By dissecting his response to this question he conceived 10 principles that expressed the most important characteristics of what he considered to be good design. They read as follows:

1. Good design is innovative. 2. Good design makes a product useful. 3. Good design is aesthetic. 4. Good design makes a product understandable. 5. Good design is unobtrusive. 6. Good design is honest. 7. Good design is long lasting. 8. Good design is thorough down to the last detail. 9. Good design is environmentally friendly.

10. Good design is as little design as possible.

Inspired by the essence of these principles, and considering their applicability to data visualisation design, I have translated them into three

56

high-level principles that similarly help me to answer my own question: ‘Is my visualisation design good visualisation design?’ These principles offer me a guiding voice when I need to resolve some of the more seemingly intangible decisions I am faced with (Figure 1.7).

Figure 1.7 The Three Principles of Good Visualisation Design

In the book Will it Make the Boat Go Faster?, co-author Ben Hunt-Davis provides details of the strategies employed by him and his team that led to their achieving gold medal success in the Men’s Rowing Eight event at the Sydney Olympics in 2000. As the title suggests, each decision taken had to pass the ‘will it make the boat go faster?’ test. Going back to the goal of data visualisation as defined earlier, these design principles help me judge whether any decision I make will better aid the facilitation of understanding: the equivalence of ‘making the boat go faster’.

I will describe in detail the thinking behind each of these principles and explain how Rams’ principles map onto them. Before that, let me briefly explain why there are three principles of Rams’ original ten that do not entirely fit, in my view, as universal principles for data visualisation.

‘I’m always the fool looking at the sky who falls off the cliff. In other words, I tend to seize on ideas because I’m excited about them without thinking through the consequences of the amount of work they will entail. I find tight deadlines energizing. Answering the question of “what is the graphic trying to do?” is always helpful. At minimum the work I create needs to speak to this. Innovation doesn’t have to be a wholesale out-of-the box approach. Iterating on a previous idea, moving it forward, is innovation.’ Sarah Slobin, Visual Journalist

Good design is innovative: Data visualisation does not need always to be innovative. For the majority of occasions the solutions being created call upon the tried and tested approaches that have been used for generations. Visualisers are not conceiving new forms of representation or implementing new design techniques in every

57

project. Of course, there are times when innovation is required to overcome a particular challenge; innovation generally materialises when faced with problems that current solutions fail to overcome. Your own desire for innovation may be aligned to personal goals about the development of your skills or through reflecting on previous projects and recognising a desire to rethink a solution. It is not that data visualisation is never about innovation, just that it is not always and only about innovation. Good design is long lasting: The translation of this principle to the context of data visualisation can be taken in different ways. ‘Long lasting’ could be related to the desire to preserve the ongoing functionality of a digital project, for example. It is quite demoralising how many historic links you visit online only to find a project has now expired through a lack of sustained support or is no longer functionally supported on modern browsers. Another way to interpret ‘long lasting’ is in the durability of the technique. Bar charts, for example, are the old reliables of the field – always useful, always being used, always there when you need them (author wipes away a respectful tear). ‘Long lasting’ can also relate to avoiding the temptation of fashion or current gimmickry and having a timeless approach to design. Consider the recent design trend moving away from skeuomorphism and the emergence of so-called flat design. By the time this book is published there will likely be a new movement. ‘Long lasting’ could apply to the subject matter. Expiry in the relevance of certain angles of analysis or out-of-date data is inevitable in most of our work, particularly with subjects that concern current matters. Analysis about the loss of life during the Second World War is timeless because nothing is now going to change the nature or extent of the underlying data (unless new discoveries emerge). Analysis of the highest grossing movies today will change as soon as new big movies are released and time elapses. So, once again, this idea of long lasting is very context specific, rather than being a universal goal for data visualisation. Good design is environmentally friendly: This is, of course, a noble aim but the relevance of this principle has to be positioned again at the contextual level, based on the specific circumstances of a given project. If your work is to be printed, the ink and paper usage immediately removes the notion that it is an environmentally friendly activity. Developing a powerful interactive that is being hammered constantly and concurrently by hundreds of thousands of users puts an

58

extra burden on the hosting server, creating more demands on energy supply. The specific judgements about issues relating to the impact of a project on the environment realistically reside with the protagonists and stakeholders involved.

A point of clarity is that, while I describe them as design principles, they actually provide guidance long before you reach the design thinking at the final stage of this workflow. Design choices encapsulate the critical thinking undertaken throughout. Think of it like an iceberg: the design is the visible consequences of lots of hidden preparatory thinking formed through earlier stages.

Finally, a comment is in order about something often raised in discussions about the principles for this subject: that is, the idea that visualisations need to be memorable. This is, in my view, not relevant as a universal principle. If something is memorable, wonderful, that will be a terrific by- product of your design thinking, but in itself the goal of achieving memorability has to be isolated, again, to a contextual level based on the specific goals of a given task and the capacity of the viewer. A politician or a broadcaster might need to recall information more readily in their work than a group of executives in a strategy meeting with permanent access to endless information at the touch of a button via their iPads.

Principle 1: Good Data Visualisation is Trustworthy The notion of trust is uppermost in your thoughts in this first of the three principles of good visualisation design. This maps directly onto one of Dieter Rams’ general principles of good design, namely that good design is honest.

Trust vs Truth

This principle is presented first because it is about the fundamental integrity, accuracy and legitimacy of any data visualisation you produce. This should always exist as your primary concern above all else. There should be no compromise here. Without securing trust the entire purpose of doing the work is undermined.

59

There is an important distinction to make between trust and truth. Truth is an obligation. You should never create work you know to be misleading in content, nor should you claim something presents the truth if it evidently cannot be supported by what you are presenting. For most people, the difference between a truth and an untruth should be beyond dispute. For those unable or unwilling to be truthful, or who are ignorant of how to differentiate, it is probably worth putting this book away now: my telling you how this is a bad thing is not likely to change your perspective.

If the imperative for being truthful is clear, the potential for there being multiple different but legitimate versions of ‘truth’ within the same data- driven context muddies things. In data visualisation there is rarely a singular view of the truth. The glass that is half full is also half empty. Both views are truthful, but which to choose? Furthermore, there are many decisions involved in your work whereby several valid options may present themselves. In these cases you are faced with choices without necessarily having the benefit of theoretical influence to draw out the right option. You decide what is right. This creates inevitable biases – no matter how seemingly tiny – that ripple through your work. Your eventual solution is potentially comprised of many well-informed, well-intended and legitimate choices – no doubt – but they will reflect a subjective perspective all the same. All projects represent the outcome of an entirely unique pathway of thought.

You can mitigate the impact of these subjective choices you make, for example, by minimising the amount of assumptions applied to the data you are working with or by judiciously consulting your audience to best ensure their requirements are met. However, pure objectivity is not possible in visualisation.

‘Every number we publish is wrong but it is the best number there is.’ Andrew Dilnott, Chair of the UK Statistics Authority

Rather than view the unavoidability of these biases as an obstruction, the focus should instead be on ensuring your chosen path is trustworthy. In the absence of an objective truth, you need to be able to demonstrate that your truth is trustable.

Trust has to be earned but this is hard to secure and very easy to lose. As the translation of a Dutch proverb states, ‘trust arrives on foot and leaves

60

on horseback’. Trust is something you can build by eliminating any sense that your version of the truth can be legitimately disputed. Yet, visualisers only have so much control and influence in the securing of trust. A visualisation can be truthful but not viewed as trustworthy. You may have done something with the best of intent behind your decision making, but it may ultimately fail to secure trust among your viewers for different reasons. Conversely a visualisation can be trustworthy in the mind of the viewer but not truthful, appearing to merit trust yet utterly flawed in its underlying truth. Neither of these are satisfactory: the latter scenario is a choice we control, the former is a consequence we must strive to overcome.

‘Good design is honest. It does not make a product appear more innovative, powerful or valuable than it really is. It does not attempt to manipulate the consumer with promises that cannot be kept.’ Dieter Rams, celebrated Industrial Designer

Let’s consider a couple of examples to illustrate this notion of trustworthiness. Firstly, think about the trust you might attach respectively to the graphics presented in Figure 1.8 and Figure 1.9. For the benefit of clarity both are extracted from articles discussing issues about home ownership, so each would be accompanied with additional written analysis at their published location. Both charts are portraying the same data and the same analysis; they even arrive at the same summary finding. How do the design choices make you feel about the integrity of each work?

Figure 1.8 Housing and Home Ownership in the UK (ONS)

61

Both portrayals are truthful but in my view the first visualisation, produced by the UK Office for National Statistics (ONS), commands greater credibility and therefore far more trust than the second visualisation, produced by the Daily Mail. The primary reason for this begins with the colour choices. They are relatively low key in the ONS graphic: colourful but subdued, yet conveying a certain assurance. In contrast, the Daily Mail’s colour palette feels needy, like it is craving my attention with sweetly coloured sticks. I don’t care for the house key imagery in the background but it is relatively harmless. Additionally, the typeface, font size and text colour feel more gimmicky in the second graphic. Once again, it feels like it is wanting to shout at me in contrast to the more polite nature of the ONS text. Whereas the Daily Mail piece refers to the ONS as the source of the data, it fails to include further details about the data source, which is included on the ONS graphic alongside other important explanatory features such as the subtitle, clarity about the yearly periods and the option to access and download the associated data. The ONS graphic effectively ‘shows all its workings’ and overall earns, from me at least, significantly more trust.

62

Figure 1.9 Falling Number of Young Homeowners (Daily Mail)

Another example about the fragility of trust concerns the next graphic, which plots the number of murders committed using firearms in Florida over a period of time. This frames the time around the enactment of the ‘Stand your ground’ law in the Florida. The area chart in Figure 1.10 shows the number of murders over time and, as you can see, the chart uses an inverted vertical y-axis with the red area going lower down as the number of deaths increases, with peak values at about 1990 and 2007. However, some commentators felt the inversion of the y-axis was deceptive and declared the graphic not trustworthy based on the fact they were perceiving the values as represented by an apparent rising ‘white mountain’. They mistakenly observed peak values around 1999 and 2005 based on them seeing these as the highest points. This confusion is caused by an effect known as figure-ground perception whereby a background form (white area) can become inadvertently recognised as the foreground form, and vice versa (with the red area seen as the background).

Figure 1.10 Gun Deaths in Florida

63

Figure 1.11 Iraq’s Bloody Toll

64

The key point here is that there was no intention to mislead. Although the approach to inverting the y-axis may not be entirely conventional, it was technically legitimate. Creatively speaking, the effect of dribbling blood was an understandably tempting metaphor to pursue. Indeed, the graphic attempts to emulate a notable infographic from several years ago showing the death toll during the Iraq conflict (Figure 1.11). In the case of the

65

Florida graphic, on reflection maybe the data was just too ‘smooth’ to convey the same dribbling effect achieved in the Iraq piece. However, being inspired and influenced by successful techniques demonstrated by others is to be encouraged. It is one way of developing our skills.

Figure 1.12 Reworking of ‘Gun Deaths in Florida’

Unfortunately, given the emotive nature of the subject matter – gun deaths – this analysis would always attract a passionate reaction regardless of its form. In this case the lack of trust expressed by some was an unintended

66

consequence of a single, innocent design: by reverting the y-axis to an upward direction, as shown in the reworked version in Figure 1.12, you can see how a single subjective design choice can have a huge influence on people’s perception.

The creator of the Florida chart will have made hundreds of perfectly sound visualisations and will make hundreds more, and none of them will ever carry the intent of being anything other than truthful. However, you can see how vulnerable perceived trust is when disputes about motives can so quickly surface as a result of the design choice made. This is especially the case within the pressured environment of a newsroom where you have only a single opportunity to publish a work to a huge and widespread audience. Contrast this setting with a graphic published within an organisation that can be withdrawn and reissued far more easily.

Trust Applies Throughout the Process

Trustworthiness is a pursuit that should guide all your decisions, not just the design ones. As you will see in the next chapter, the visualisation design workflow involves a process with many decision junctions – many paths down which you could pursue different legitimate options. Obviously, design is the most visible result of your decision making, but you need to create and demonstrate complete integrity in the choices made across the entire workflow process. Here is an overview of some of the key matters where trust must be at the forefront of your concern.

‘My main goal is to represent information accurately and in proper context. This spans from data reporting and number crunching to designing human-centered, intuitive and clear visualizations. This is my sole approach, although it is always evolving.’ Kennedy Elliott, Graphics Editor, The Washington Post

Formulating your brief: As mentioned in the discussion about the ‘Gun Crimes in Florida’ graphic, if you are working with potentially emotive subject matter, this will heighten the importance of demonstrating trust. Rightly or wrongly, your topic will be more exposed to the baggage of prejudicial opinion and trust will be precarious. As you will learn in Chapter 3, part of the thinking involved in ‘formulating your brief’ concerns defining your audience,

67

considering your subject and establishing your early thoughts about the purpose of your work, and what you are hoping to achieve. There will be certain contexts that lend themselves to exploiting the emotive qualities of your subject and/or data but many others that will not. Misjudge these contextual factors, especially the nature of your audience’s needs, and you will jeopardise the trustworthiness of your solution. As I have shown, matters of trust are often outside of your immediate influence: cynicism, prejudice or suspicion held by viewers through their beliefs or opinions is a hard thing to combat or accommodate. In general, people feel comfortable with visualisations that communicate data in a way that fits with their world view. That said, at times, many are open to having their beliefs challenged by data and evidence presented through a visualisation. The platform and location in which your work is published (e.g. website or source location) will also influence trust. Visualisations encountered in already-distrusted media will create obstacles that are hard to overcome. Working with data: As soon as you begin working with data you have a great responsibility to be faithful to this raw material. To be transparent to your audience you need to consider sharing as much relevant information about how you have handled the data that is being presented to them:

How was it collected: from where and using what criteria? What calculations or modifications have you applied to it? Explain your approach. Have you made any significant assumptions or observed any special counting rules that may not be common? Have you removed or excluded any data? How representative it is? What biases may exist that could distort interpretations?

Editorial thinking: Even with the purest of intent, your role as the curator of your data and the creator of its portrayal introduces subjectivity. When you choose to do one thing you are often choosing to not do something else. The choice to focus on analysis that shows how values have changed over time is also a decision to not show the same data from other viewpoints such as, for example, how it looks on a map. A decision to impose criteria on your analysis, like setting date parameters or minimum value thresholds, in order to reduce clutter, might be sensible and indeed legitimate, but is still a subjective choice.

68

‘Data and data sets are not objective; they are creations of human design. Hidden biases in both the collection and analysis stages present considerable risks [in terms of inference].’ Kate Crawford, Principal Researcher at Microsoft Research NYC

Data representation: A fundamental tenet of data visualisation is to never deceive the receiver. Avoiding possible misunderstandings, inaccuracies, confusions and distortions is of primary concern. There are many possible features of visualisation design that can lead to varying degrees of deception, whether intended or not. Here are a few to list now, but note that these will be picked up in more detail later:

The size of geometric areas can sometimes be miscalculated resulting in the quantitative values being disproportionately perceived. When data is represented in 3D, on the majority of occasions this represents nothing more than distracting – and distorting – decoration. 3D should only be used when there are legitimately three dimensions of data variables being displayed and the viewer is able to change his or her point of view to navigate to see different 2D perspectives. The bar chart value axis should never be ‘truncated’ – the origin value should always be zero – otherwise this approach will distort the bar size judgements. The aspect ratio (height vs width) of a line chart’s display is influential as it affects the perceived steepness of connecting lines which are key to reading the trends over time – too narrow and the steepness will be embellished; too wide and the steepness is dampened. When portraying spatial analysis through a thematic map representation, there are many different mapping projections to choose from as the underlying apparatus for presenting and orienting the geographical position of the data. There are many different approaches to flatten the spherical globe, translating it into a two-dimensional map form. The mathematical treatment applied can alter significantly the perceived size or shape of regions, potentially distorting their perception. Sometimes charts are used in a way that is effectively corrupt, like using pie charts for percentages that add up to more, or less, than 100%.

69

Data presentation: The main rule here is: if it looks significant, it should be, otherwise you are either misleading or creating unnecessary obstacles for your viewer. The undermining of trust can also be caused by what you decline to explain: restricted or non- functioning features of interactivity.

Absent annotations such as introduction/guides, axis titles and labels, footnotes, data sources that fail to inform the reader of what is going on. Inconsistent or inappropriate colour usage, without explanation. Confusing or inaccessible layouts. Thoroughness in delivering trust extends to the faith you create through reliability and consistency in the functional experience, especially for interactive projects. Does the solution work and, specifically, does it work in the way it promises to do?

Principle 2: Good Data Visualisation is Accessible This second of the three principles of good visualisation design helps to inform judgments about how best to facilitate your viewers through the process of understanding. It is informed by three of Dieter Rams’ general principles of good design:

2 Good design makes a product useful. 4 Good design makes a product understandable. 5 Good design is unobtrusive.

Reward vs Effort

The opening section of this chapter broke down the stages a viewer goes through when forming their understanding about, and from, a visualisation. This process involved a sequence of perceiving, interpreting and then comprehending. It was emphasised that a visualiser’s control over the viewer’s pursuit of understanding diminishes after each stage. The objective, as stated by the presented definition, of ‘facilitating’ understanding reflects the reality of what can be controlled. You can’t force viewers to understand, but you can smooth the way.

To facilitate understanding for an audience is about delivering accessibility. That is the essence of this principle: to remove design-related

70

obstacles faced by your viewers when undertaking this process of understanding. Stated another way, a viewer should experience minimum friction between the act of understanding (effort) and the achieving of understanding (reward).

This ‘minimising’ of friction has to be framed by context, though. This is key. There are many contextual influences that will determine whether what is judged inaccessible in one situation could be seen as entirely accessible in another. When people are involved, diverse needs exist. As I have already discussed, varying degrees of knowledge emerge and irrational characteristics come to the surface. You can only do so much: do not expect to get all things right in the eyes of every viewer.

‘We should pay as much attention to understanding the project’s goal in relation to its audience. This involves understanding principles of perception and cognition in addition to other relevant factors, such as culture and education levels, for example. More importantly, it means carefully matching the tasks in the representation to our audience’s needs, expectations, expertise, etc. Visualizations are human-centred projects, in that they are not universal and will not be effective for all humans uniformly. As producers of visualizations, whether devised for data exploration or communication of information, we need to take into careful consideration those on the other side of the equation, and who will face the challenges of decoding our representations.’ Isabel Meirelles, Professor, OCAD University (Toronto)

That is not to say that attempts to accommodate the needs of your audience should just be abandoned, quite the opposite. This is hard but it is essential. Visualisation is about human-centred design, demonstrating empathy for your audiences and putting them at the heart of your decision making.

There are several dimensions of definition that will help you better understand your audiences, including establishing what they know, what they do not know, the circumstances surrounding their consumption of your work and their personal characteristics. Some of these you can accommodate, others you may not be able to, depending on the diversity and practicality of the requirements. Again, in the absence of perfection optimisation is the name of the game, even if this means that sometimes the least worst is best.

71

The Factors Your Audiences Influence

Many of the factors presented here will occur when you think about your project context, as covered in Chapter 3. For now, it is helpful to introduce some of the factors that specifically relate to this discussion about delivering accessible design.

Subject-matter appeal: This was already made clear in the earlier illustration, but is worth logging again here: the appeal of the subject matter is a fundamental junction right at the beginning of the consumption experience. If your audiences are not interested in the subject – i.e. they are indifferent towards the topic or see no need or relevance to engage with it there and then – then they will not likely stick around. They will probably not be interested in putting in the effort to work through the process of understanding for something that might be ultimately irrelevant. For those to whom the subject matter is immediately appealing, they are significantly more likely to engage with the data visualisation right the way through.

‘Data visualization is like family photos. If you don’t know the people in the picture, the beauty of the composition won’t keep your attention.’ Zach Gemignani, CEO/Founder of Juice Analytics

Many of the ideas for this principle emerged from the Seeing Data visualisation literacy research project (seeingdata.org) on which I collaborated.

Dynamic of need: Do they need to engage with this work or is it entirely voluntary? Do they have a direct investment in having access to this information, perhaps as part of their job and they need this information to serve their duties? Subject-matter knowledge: What might your audiences know and not know about this subject? What is their capacity to learn or potential motivation to develop their knowledge of this subject? A critical component of this issue, blending existing knowledge with the capacity to acquire knowledge, concerns the distinctions between complicated, complex, simple and simplified. This might seem to be more about the semantics of language but is of significant influence in data visualisation – indeed in any form of communication:

72

Complicated is generally a technical distinction. A subject might be difficult to understand because it involves pre-existing – and probably high-level – knowledge and might be intricate in its detail. The mathematics that underpinned the Moon landings are complicated. Complicated subjects are, of course, surmountable – the knowledge and skill are acquirable – but only achieved through time and effort, hard work and learning (or extraordinary talent), and, usually, with external assistance. Complex is associated with problems that have no perfect conclusion or maybe even no end state. Parenting is complex; there is no rulebook for how to do it well, no definitive right or wrong, no perfect way of accomplishing it. The elements of parenting might not be necessarily complicated – cutting Emmie’s sandwiches into star shapes – but there are lots of different interrelated pressures always influencing and occasionally colliding. Simple, for the purpose of this book, concerns a matter that is inherently easy to understand. It may be so small in dimension and scope that it is not difficult to grasp, irrespective of prior knowledge and experience. Simplified involves transforming a problem context from either a complex or complicated initial state to a reduced form, possibly by eliminating certain details or nuances.

Understanding the differences in these terms is vital. When considering your subject matter and the nature of your analysis you will need to assess whether your audience will be immediately able to understand what you are presenting or have the capacity to learn how to understand it. If it is a subject that is inherently complex or complicated, will it need to be simplified? If you are creating a graphic about taxation, will you need to strip it down to the basics or will this process of simplification risk the subject being oversimplified? The final content may be obscured by the absence of important subtleties. Indeed, the audience may have felt sufficiently sophisticated to have had the capacity to work out and work with a complicated topic, but you denied them that opportunity. You might reasonably dilute/reduce a complex subject for kids, but generally my advice is don’t underestimate the capacity of your audience. Accordingly, clarity trumps simplicity as the most salient concern about data visualisation design.

73

‘Strive for clarity, not simplicity. It’s easy to “dumb something down,” but extremely difficult to provide clarity while maintaining complexity. I hate the word “simplify.” In many ways, as a researcher, it is the bane of my existence. I much prefer “explain,” “clarify,” or “synthesize.” If you take the complexity out of a topic, you degrade its existence and malign its importance. Words are not your enemy. Complex thoughts are not your enemy. Confusion is. Don’t confuse your audience. Don’t talk down to them, don’t mislead them, and certainly don’t lie to them.’ Amanda Hobbs, Researcher and Visual Content Editor

What do they need to know? The million-dollar question. Often, the most common frustration expressed by viewers is that the visualisation ‘didn’t show them what they were most interested in’. They wanted to see how something changed over time, not how it looked on a map. If you were them what would you want to know? This is a hard thing to second-guess with any accuracy. We will be discussing it further in Chapter 5. Unfamiliar representation: In the final chapter of this book I will cover the issue of visualisation literacy, discussing the capabilities that go into being the most rounded creator of visualisation work and the techniques involved in being the most effective consumer also. Many people will perhaps be unaware of a deficit in their visualisation literacy with regard to consuming certain chart types. The bar, line and pie chart are very common and broadly familiar to all. As you will see in Chapter 6, there are many more ways of portraying data visually. This deficit in knowing how to read a new or unfamiliar chart type is not a failing on the part of the viewer, it is simply a result of their lack of prior exposure to these different methods. For visualisers a key challenge lies with situations when the deployment of an uncommon chart may be an entirely reasonable and appropriate choice – indeed perhaps even the ‘simplest’ chart that could have been used – but it is likely to be unfamiliar to the intended viewers. Even if you support it with plenty of ‘how to read’ guidance, if a viewer is overwhelmed or simply unwilling to make the effort to learn how to read a different chart type, you have little control in overcoming this. Time: At the point of consuming a visualisation is the viewer in a pressured situation with a lot at stake? Are viewers likely to be impatient and intolerant of the need to spend time learning how to read a display? Do they need quick insights or is there some capacity for them to take on exploring or reading in more depth? If it is the

74

former, the immediacy of the presented information will therefore be a paramount requirement. If they have more time to work through the process of perceiving, interpreting and comprehending, this could be a more conducive situation to presenting complicated or complex subject matter – maybe even using different, unfamiliar chart types. Format: What format will your viewers need to consume your work? Are they going to need work created for a print output or a digital one? Does this need to be compatible with a small display as on a smartphone or a tablet? If what you create is consumed away from its intended native format, such as viewing a large infographic with small text on a mobile phone, that will likely result in a frustrating experience for the viewer. However, how and where your work is consumed may be beyond your control. You can’t mitigate for every eventuality. Personal tastes: Individual preferences towards certain colours, visual elements and interaction features will often influence (enabling or inhibiting) a viewer’s engagement. The semiotic conventions that visualisers draw upon play a part in determining whether viewers are willing to spend time and expend effort looking at a visualisation. Be aware though that accommodating the preferences of one person may not cascade, with similar appeal, to all, and might indeed create a rather negative reaction. Attitude and emotion: Sometimes we are tired, in a bad mood, feeling lazy, or having a day when we are just irrational. And the prospect of working on even the most intriguing and well-designed project sometimes feels too much. I spend my days looking at visualisations and can sympathise with the narrowing of mental bandwidth when I am tired or have had a bad day. Confidence is an extension of this. Sometimes our audiences may just not feel sufficiently equipped to embark on a visualisation if it is about an unknown subject or might involve pushing them outside their comfort zone in terms of the demands placed on their interpretation and comprehension.

The Factors You Can Influence

Flipping the coin, let’s look at the main ways we, as visualisers, can influence (positively or negatively) the accessibility of the designs created. In effect, this entire book is focused on minimising the likelihood that your solution demonstrates any of these negative attributes. Repeating the

75

mantra from earlier, you must avoid doing anything that will cause the boat to go slower.

‘The key difference I think in producing data visualisation/infographics in the service of journalism versus other contexts (like art) is that there is always an underlying, ultimate goal: to be useful. Not just beautiful or efficient – although something can (and should!) be all of those things. But journalism presents a certain set of constraints. A journalist has to always ask the question: How can I make this more useful? How can what I am creating help someone, teach someone, show someone something new?’ Lena Groeger, Science Journalist, Designer and Developer at ProPublica

As you saw listed at the start of this section, the selected, related design principles from Dieter Rams’ list collectively include the aim of ensuring our work is useful, unobtrusive and understandable. Thinking about what not to do – focusing on the likely causes of failure across these aims – is, in this case, more instructive.

Your Solution is Useless

You have failed to focus on relevant content. It is not deep enough. You might have provided a summary- level/aggregated view of the data when the audience wanted further angles of analysis and greater depth in the details provided. A complex subject was oversimplified. It is not fit for the setting. You created work that required too much time to make sense of, when immediate understanding and rapid insights were needed.

Your Solution is Obtrusive

It is visually inaccessible. There is no appreciation of potential impairments like colour blindness and the display includes clumsily ineffective interactive features. Its format is misjudged. You were supposed to create work fit for a small-sized screen, but the solution created was too fine-detailed and could not be easily read. It has too many functions. You failed to focus and instead provided too many interactive options when the audience had no desire to put

76

in a lot of effort interrogating and manipulating the display.

You Solution is not Understandable

Complex subject or complex analysis. Not explained clearly enough – assumed domain expertise, such as too many acronyms, abbreviations and technical language. Used a complex chart type. Not enough explanation of how to read the graphic or failure to consider if the audience would be capable of understanding this particular choice of chart type. Absent annotations. Insufficient details like scales, units, descriptions, etc.

Principle 3: Good Data Visualisation is Elegant Elegance in design is the final principle of good visualisation design. This relates closely to the essence of three more of Dieter Rams’ general principles of good design:

3 Good design is aesthetic. 8 Good design is thorough down to the last detail. 10 Good design is as little design as possible.

What is Elegant Design?

Elegant design is about seeking to achieve a visual quality that will attract your audience and sustain that sentiment throughout the experience, far beyond just the initial moments of engagement. This is presented as the third principle for good reason. Any choices you make towards achieving ‘elegance’ must not undermine the accomplishment of trustworthiness and accessibility in your design. Indeed, in pursuing the achievement of the other principles, elegance may have already arrived as a by-product of trustworthy and accessible design thinking. Conversely, the visual ‘look and feel’ of your work will be the first thing viewers encounter before experiencing the consequences of your other principle-led thinking. It therefore stands that optimising the perceived appeal of your work will have a great impact on your viewers.

The pursuit of elegance is elusive, as is its definition: what gives

77

something an elegant quality? As we know, beauty is in the eye of the beholder, but how do we really recognise elegance when we are confronted by it?

When thinking about what the pursuit of elegance of means, the kind of words that surface in my mind are adjectives like stylish, dignified, effortless and graceful. For me, they capture the timelessness of elegance, certainly more so than fancy, cool or trendy, which seem more momentary. Elegance is perhaps appreciated more when it is absent from or not entirely accomplished in a design. If something feels cumbersome, inconsistent and lacking a sense of harmony across its composition and use of colour, it is missing that key ingredient of elegance.

‘When working on a problem, I never think about beauty. I think only how to solve the problem. But when I have finished, if the solution is not beautiful, I know it is wrong.’ Richard Buckminster Fuller, celebrated inventor and visionary

‘Complete is when something looks seamless, as if it took little effort to produce.’ Sarah Slobin, Visual Journalist

When it feels like style over substance has been at the heart of decision- making, no apparent beauty can outweigh the negatives of an obstructed or absent functional experience. While I’m loathe to dwell on forcing a separation in concern between form and function, as a beginner working through the design stages and considering all your options, functional judgements will generally need to be of primary concern. However, it is imperative that you also find room for appropriate aesthetic expression. In due course your experience will lead you to fuse the two perspectives together more instinctively.

In his book The Shape of Design, designer Frank Chimero references a Shaker proverb: ‘Do not make something unless it is both necessary and useful; but if it is both, do not hesitate to make it beautiful.’ In serving the principles of trustworthy and accessible design, you will have hopefully covered both the necessary and useful. As Chimero suggests, if we have served the mind, our heart is telling us that now is the time to think about beauty.

78

How Do You Achieve Elegance in Design?

There are several components of design thinking that I believe directly contribute to achieving an essence of elegance.

‘“Everything must have a reason”… A principle that I learned as a graphic designer that still applies to data visualisation. In essence, everything needs to be rationalised and have a logic to why it’s in the design/visualisation, or it’s out.’ Stefanie Posavec, Information Designer

Eliminate the arbitrary: As with any creative endeavour or communication activity, editing is perhaps the most influential skill, and indeed attitude. Every single design decision you make – every dot, every pixel – should be justifiable. Nothing that remains in your work should be considered arbitrary. Even if there isn’t necessarily a scientific or theoretical basis for your choices, you should still be able to offer reasons for every thing that is included and also excluded. The reasons you can offer for design options being rejected or removed are just as important in evidence of your developing eye for visualisation design. Often you will find yourself working alone on a data visualisation project and will therefore need to demonstrate the discipline and competence to challenge yourself. Avoid going through the motions and don’t get complacent. Why present data on a map if there is nothing spatially relevant about the regional patterns? Why include slick interactive features if they really add no value to the experience? It is easy to celebrate the brilliance of your amazing ideas and become consumed by work that you have invested deeply in – both your time and emotional energy. Just don’t be stubborn or precious. If something is not working, learn to recognise when to not pursue it any further and then kill it. Thoroughness: A dedicated visualiser should be prepared to agonise over the smallest details and want to resolve even the smallest pixel- width inaccuracies. The desire to treat your work with this level of attention demonstrates respect for your audience: you want them to be able to work with quality so pride yourself on precision. Do not neglect checking, do not cut corners, do not avoid the non-sexy duties, and never stop wanting to do better.

79

Style: This is another hard thing to pin down, especially as the word itself can have different meanings for people, and especially when it has been somewhat ‘damaged’ by the age-old complaints around something demonstrating style over substance. Developing a style – or signature, as Thomas Clever suggests – is in many ways a manifestation of elegant design. The decisions around colour selection, typography and composition are all matters that influence your style. The development of a style preserves the consistency of your strongest design values, leaving room to respond flexibly to the nuances of each different task you face. It is something that develops in time through the choices you make and the good habits you acquire.

‘You don’t get there [beauty] with cosmetics, you get there by taking care of the details, by polishing and refining what you have. This is ultimately a matter of trained taste, or what German speakers call fingerspitzengefühl (“finger-tip-feeling”)’. Oliver Reichenstein, founder of Information Architects (iA)

Many news and media organisations seek to devise their own style guides to help visualisers, graphics editors and developers navigate through the choppy waters of design thinking. This is a conscious attempt to foster consistency in approach as well as create efficiency. In these industries, the perpetual pressure of tight timescales from the relentless demands of the news cycle means that creating efficiency is of enormous value. By taking away the burden of having always to think from scratch about their choices, the visualisers in such organisations are left with more room to concern themselves with the fundamental challenge of what to show and not just get consumed by how to show it. The best styles will stand out as instantly recognisable: there is a reason why you can instantly pick out the work of the New York Times, National Geographic, Bloomberg, the Guardian, the Washington Post, the Financial Times, Reuters and the South China Morning Post. Decoration should be additive, not negative: The decorative arts are historically considered to be an intersection of that which is useful and beauty, yet the term decoration when applied to data can often suggest a negative connotation of dressing it up using superfluous devices to attract people, but without any real substance. Visual embellishments are, in moderation and when discernibly deployed,

80

effective devices for securing visual appeal and preserving communicated value. This is especially the case when they carry a certain congruence with the subject matter or key message, such as with the use of the different ground textures in the treemap displayed in Figure 1.13. In this graphic, Vienna is reduced to an illustrative 100m2 apartment and the floor plan presents the proportional composition of the different types of space and land in the city. This is acceptable gratuitousness because the design choices are additive, not negatively obstructive or distracting.

‘I suppose one could say our work has a certain “signature”. “Style” – to me – has a negative connotation of “slapped on” to prettify something without much meaning. We don’t make it our goal to have a recognisable (visual) signature, instead to create work that truly matters and is unique. Pretty much all our projects are bespoke and have a different end result. That is one of the reasons why we are more concerned with working according to values and principles that transcend individual projects and I believe that is what makes our work recognisable.’ Thomas Clever, Co-founder CLEVER°FRANKE, a data driven experiences studio

Figure 1.13 If Vienna Would be an Apartment

Any design choices you make with the aim of enhancing appeal through

81

novelty or fun need to support, not distract from, the core aim of facilitating understanding. Be led by your data and your audience, not your ideas. There should, though, always be room to explore ways of seeking that elusive blend of being fun, engaging and informative. The bar chart in Figure 1.14 reflects this: using Kit Kat-style fingers of chocolate for each bar and a foil wrapper background, it offers an elegant and appealing presentation that is congruent with its subject.

Figure 1.14 Asia Loses Its Sweet Tooth for Chocolate

82

Allow your personality to express itself in the times and places where such flair is supportive of the aims of facilitating understanding. After all, a

83

singularity of style is a dull existence. As Groove Armada once sang: ‘If everybody looked the same, we’d get tired of looking at each other.’

Not about minimalism: As expressed by Rams’ principle ‘Good design is as little design as possible’, elegant design achieves a certain invisibility: as a viewer you should not see design, you should see content. This is not to be confused with the pursuit of minimalism, which is a brutal approach that strips away the arbitrary but then cuts deeper. In the context of visualisation, minimalism can be an unnecessarily savage and austere act that may be incongruous with some of the design options you may need to include in your work.

‘I’ve come to believe that pure beautiful visual works are somehow relevant in everyday life, because they can become a trigger to get people curious to explore the contents these visuals convey. I like the idea of making people say “oh that’s beautiful! I want to know what this is about!” I think that probably (or, at least, lots of people pointed that out to us) being Italians plays its role on this idea of “making things not only functional but beautiful”.’ Giorgia Lupi, Co-founder and Design Director at Accurat

In ‘De architectura’, a thesis on architecture written around 15 BC by Marcus Vitruvius Pollio, a Roman architect, the author declares how the essence of quality in architecture is framed by the social relevance of the work, not the eventual form or workmanship towards that form. What he is stating here is that good architecture can only be measured according to the value it brings to the people who use it. In a 1624 translation of the work, Sir Henry Wooton offers a paraphrased version of one of Vitruvius’s most enduring notions that a ‘well building hath three conditions: firmness, commodity, and delight’, of which a further interpretation for today might be read as ‘sturdy, useful, and beautiful’. One can easily translate these further to fit with these principles of good visualisation design. Trustworthy is sturdy – it is robust, reliable, and has integrity. Useful is accessible – it can be used without undue obstruction. Beautiful is elegant – it appeals and retains attraction.

1.3 Distinctions and Glossary As in any text, consistency in the meaning of terms or language used

84

around data visualisation is important to preserve clarity for readers. I began this chapter with a detailed breakdown of a proposed definition for the subject. There are likely to be many other terms that you either are familiar with or have heard being used. Indeed, there are significant overlaps and commonalities of thought between data visualisation and pursuits like, for example, infographic design.

As tools and creative techniques have advanced over the past decade, the traditional boundaries between such fields begin to blur. Consequently, the practical value of preserving dogmatic distinctions reduces accordingly. Ultimately, the visualiser tasked with creating a visual portrayal of data is probably less concerned about whether their creation will be filed under ‘data visualisation’ or ‘infographic’ as long as it achieves the aim of helping the audience achieve understanding.

Better people than me attach different labels to different works interchangeably, perhaps reflecting the fact that these dynamic groups of activities are all pursuing similar aims and using the same raw material – data – to achieve them. Across this book you will see plenty of references to and examples of works that might not be considered data visualisation design work in the purest sense. You will certainly see plenty of examples of infographics.

The traditional subject distinctions still deserve to be recognised and respected. People are rightfully proud of identifying with a discipline they have expertise or mastery in. And so, before you step into the design workflow chapters, it is worthwhile to spend a little time establishing clarifications and definitions for some of the related fields and activities so all readers are on the same page of understanding. Additionally, there is a glossary of the terms used that will help you more immediately understand the content of later chapters. It makes sense to position those clarifications in this chapter as well.

Distinctions Data vis: Just to start with one clarification. While the abbreviated term of data visualisation might be commonly seen as ‘data vis’ (or ‘data viz’; don’t get me started on the ‘z’ issue), and this is probably how all the cool kids on the street and those running out of characters on Twitter refer to it, I am sticking with the full Sunday name of ‘data

85

visualisation’ or at the very least the shortened term ‘visualisation’. Information visualisation: There are many who describe data visualisation as information visualisation and vice versa, myself included, without a great deal of thought for the possible differences. The general distinction, if there is any, tends to be down to one’s emphasis on the input material (data) or the nature of the output form (information). It is also common that information visualisation is used as the term to define work that is primarily concerned with visualising abstract data structures such as trees or graphs (networks) as well as other qualitative data (therefore focusing more on relationships rather than quantities). Infographics: The classic distinction between infographics and data visualisation concerns the format and the content. Infographics were traditionally created for print consumption, in newspapers or magazines, for example. The best infographics explain things graphically –systems, events, stories – and could reasonably be termed explanation graphics. They contain charts (visualisation elements) but may also include illustrations, photo-imagery, diagrams and text. These days, the art of infographic design continues to be produced in static form, irrespective of how and where they are published. Over the past few years there has been an explosion in different forms of infographics. From a purist perspective, this new wave of work is generally viewed as being an inferior form of infographic design and may be better suited to terms like info-posters or tower graphics (these commonly exist with a fixed-width dimension in order to be embedded into websites and social media platforms). Often these works will be driven by marketing intent through a desire to get hits/viewers, generally with the compromising of any real valuable delivery of understanding. It is important not to dismiss entirely the evident – if superficial – value of this type of work, as demonstrated by the occasionally incredible numbers for hits received. If your motive is ‘bums on seats’ then this approach will serve you well. However, I would question the legitimacy of attaching the term infographic to these designs and I sense the popular interest in these forms is beginning to wane. Visual analytics: Some people use this term to relate to analytical- style visualisation work, such as dashboards, that serve the role of operational decision support systems or provide instruments of business intelligence. Additionally, the term visual analytics is often

86

used to describe the analytical reasoning and exploration of data facilitated by interactive tools. This aligns with the pursuit of exploratory data analysis that I will be touching on in Chapter 5. Data art: Aside from the disputes over the merits of certain infographic work, data art is arguably the other discipline related to visualisation that stirs up the most debate. Those creating data art are often pursuing a different motive to pure data visualisation, but its sheer existence still manages to wind up many who perhaps reside in the more ‘purest’ visualisation camps. For data artists the raw material is still data but their goal is not driven by facilitating the kind of understanding that a data visualisation would offer. Data art is more about pursuing a form of self-expression or aesthetic exhibition using data as the paint and algorithms as the brush. As a viewer, whether you find meaning in displays of data art is entirely down to your personal experience and receptiveness to the open interpretation it invites. Information design: Information design is a design practice concerned with the presentation of information. It is often associated with the activities of data visualisation, as it shares the underlying motive of facilitating understanding. However, in my view, information design has a much broader application concerned with the design of many different forms of visual communication, such as way-finding devices like hospital building maps or in the design of utility bills. Data science: As a field, data science is hard to define, so it is easier to consider this through the ingredients of the role of data scientists. They possess a broad repertoire of capabilities covering the gathering, handling and analysing of data. Typically this data is of a large size and complexity and originates from multiple sources. Data scientists will have strong mathematical, statistical and computer science skills, not to mention astute business experience and many notable ‘softer’ skills like problem solving, communication and presentation. If you find somebody with all these skills, tie them to a desk (legally) and never ever let them leave your organisation. Data journalism: Also known as data-driven journalism (DDJ), this concerns the increasingly recognised importance of having numerical, data and computer skills in the journalism field. In a sense it is an adaption of data visualisation but with unquestionably deeper roots in the responsibilities of the reporter/journalist. Scientific visualisation: This is another form of a term used by many

87

people for different applications. Some give exploratory data analysis the label scientific visualisation (drawing out the scientific methods for analysing and reasoning about data). Others relate it to the use of visualisation for conceiving highly complex and multivariate datasets specifically concerning matters with a scientific bent (such as the modelling functions of the brain or molecular structures).

Glossary The precision and consistency of language in this field can get caught up in a little too much semantic debate at times, but it is important to establish early on some clarity about its usage and intent in this book at least.

Roles and Terminology

Project: For the purpose of this book, you should consider any data visualisation creation activity to be consistent with the idea of a project. Even if what you are working on is only seen as the smallest of visualisation tasks that hardly even registers on the bullet points of a to-do list, you should consider it a project that requires the same rigorous workflow process approach. Visualiser: This is the role I am assigning to you – the person making the visualisation. It could be more realistic to use a term like researcher, analyst, creator, practitioner, developer, storyteller or, to be a little pretentious, visualist. Designer would be particularly appropriate but I want to broaden the scope of the role beyond just the design thinking to cover all aspects of this discipline. Viewer: This is the role assigned to the recipient, the person who is viewing and/or using your visualisation product. It offers a broader and better fit than alternatives such as consumer, reader, recipient or customer. Audience: This concerns the collective group of people to whom you are intending to serve your work. Within the audience there will be cohorts of different viewer types that you might characterise through distinct personas to help your thinking about serving the needs of target viewers. Consuming: This will be the general act of the viewer, to consume. I will use more active descriptions like ‘reading’ and ‘using’ when consuming becomes too passive and vague, and when distinctions are

88

needed between reading text and using interactive features. Creating: This will be the act of the visualiser, to create. This term will be mainly used in contrast with consuming to separate the focus between the act of the visualiser and the act of the viewer.

Data Terminology

Data is: I’m sorry ‘data are’ fans, but that’s just not how normal people speak. In this book, it’s going to be ‘data is’ all the way. Unless my editor disagrees, in which case you won’t even see this passage. Raw data: Also known as primary data, this is data that has not been subjected to statistical treatment or any other transformation to prepare it for usage. Some people have a problem with the implied ‘rawness’ this term claims, given that data will have already lost its purity having been recorded by some measurement instrument, stored, retrieved and maybe cleaned already. I understand this view, but am going to use the term regardless because I think most people will understand its intent. Dataset: A dataset is a collection of data values upon which a visualisation is based. It is useful to think of a dataset as taking the form of a table with rows and columns, usually existing in a spreadsheet or database. Tabulation: A table of data is based on rows and columns. The rows are the records – instances of things – and the columns are the variables – details about the things. Datasets are visualised in order to ‘see’ the size, patterns and relationships that are otherwise hard to observe. For the purpose of this book, I distinguish between types of datasets that are ‘normalised’ and others that are ‘cross-tabulated’. This distinction will be explained in context during Chapter 5. Variables: Variables are related items of data held in a dataset that describe a characteristic of those records. It might be the names, dates of birth, genders and salaries of a department of employees. Think of variables as the different columns of values in a table, with the variable name being the descriptive label on the header row. There are different types of variables including, at a general level, quantitative (e.g. salary) and categorical (e.g. gender). A chart plots the relationship between different variables. For example, a bar chart might show the number of staff (with the size of bar showing the quantity) across different departments (one bar for each department or

89

category). Series: A series of values is essentially a row (or column, depending on table layout) of related values in a table. An example of a series of values would be all the highest temperatures in a city for each month of the year. Plotting this on a chart, like a line chart, would produce a line for that city’s values across the year. Another line could be added to compare temperatures for another city thus presenting a further series of values. Data source: This is the term used to describe the origin of data or information used to construct the analysis presented. This is an important feature of annotation that can help gain trust from viewers by showing them all they need to know about the source of the data. Big Data: Big Data is characterised by the 3Vs – high volume (millions of rows of data), high variety (hundreds of different variables/columns) and high velocity (new data that is created rapidly and frequently, every millisecond). A database of bank transactions or an extract from a social media platform would be typical of Big Data. It is necessary to take out some of the hot air spouted about Big Data in its relationship with data visualisation. The ‘Bigness’ (one always feels obliged to include a capitalised B) of data does not fundamentally change the tasks one faces when creating a data visualisation, it just makes it a more significant prospect to work through. It broadens the range of possibilities, it requires stronger and more advanced technology resources, and it amplifies the pressures on time and resources. With more options the discipline of choice becomes of even greater significance.

Visualisation

Chart type: Charts are individual, visual representations of data. There are many ways of representing your data, using different combinations of marks, attributes, layouts and apparatus: these combinations form archetypes of charts, commonly reduced to simply chart types. There are some charts you might already be familiar with, such as the bar chart, pie chart or line chart, while others may be new to you, like the Sankey diagram, treemap or choropleth map. Graphs, charts, plots, diagrams and maps: Traditionally the term graph has been used to describe visualisations that display network relationships and chart would be commonly used to label common devices like the bar or pie chart. Plots and diagrams are more