Download to read offline
Uploaded byKaarin Hoff
Listening to Data
The document discusses the importance of data-driven strategies for effective visualizations in business contexts, emphasizing clear, useful, ethical, and credible communication of data. It highlights core principles of visualizations as rhetorical tools that aim to engage audiences in meaningful conversations about data. Examples are provided to illustrate how visualizations can effectively present complex information while maintaining integrity and relevance.
Listening to Data
- 1. Listening to Data Conversationbased, Data driven strategy for better visualizations Daniel O’Neil, Business Analyst Kaarin Hoff, Information Architect The Understanding Group (TUG) @phoenix1189 @kaarinh @undrstndng @undrstndng
- 2. The Understanding Group isan Information Architecture practice dedicated to making things be good. We work to Understand the goals of your Business and Users, then we architect your information to achieve those goals. @undrstndng
- 3. Kaarin Hoff Information Architect,TUG ter Chap Source: http://www.artble.com/artists/johannes_vermeer, http://ffffound.com/home/vvva/found/ ter 1 Chap 2 EdR Log o 8 H om 1 r le d xamp icing elit, sena is ag ne y E ip m ou tur ad dolore tation mer J amet, consectebore et rud exercit. Duis ut la t nost nsequa lor si cididunt quis llum o co m do am, se ci r in m ipsu tempo inim veni ea commod e velit es caecat Lore ad m lit nt oc luptat usmod p ex do ei Ut enim si ut aliqui derit in vo epteur si serunt mol . xc ni E ia de ehen aliqua laboris riatur. a qui offic in repr co lp lla pa ullam re dolor iru at nu , sunt in cu gi ex ea aute eu fu oident iquip in ut al derit dolore tat non pr m. nisi ehen r. ru boris in repr lla pariatu a cupida est labo co la re dolor id lp nu ullam iru anim in cu giat tation Duis aute re eu fu dent, sunt erci ud ex nsequat. llum dolo t non proi . ci Nostr odo co m ta esse at cupida est laboru comm e velit caec im id at volupt teur sint ocnt mollit an Excep ia deseru ex ea fic si iquip qui of ut al laboris ni pens ris nisi lamco bo g hap ethin lamco la citation ul Som er ul ud ex ter 1: citation er Nostr nsequat. Chap ea uat. ud ex p ex in Nostr odo conseq mmodo co aliqui rit co ens comm p ex ea happ ris nisi ut reprehende tur. ui else ria ut aliq hing mco labo dolor in nulla pa lpa re cu omet la r 2: S citation ul s aute iru eu fugiat , sunt in te nt ui er Chap dolore n proide uat. D ud ex Nostr odo conseq se cillum datat no borum. es cupi est la comm e velit caecat im id at volupt teur sint ocnt mollit an Excep ia deseru fic qui of Ser v 10 al Extern tX Clien 3 9 Custo e Prop rty #3 2 5 3 6 5 7 e Prop rty #2 4 2 4 erty Prop #1 1 e> ice Ser vic e ice s s Em plo yme s Intr nt | odu New cto ava ry c s|C ilab onte onta le, elit. nt a ct U Duis exclus bou s | In ively t th laore e va ves •S thro et e riety ed tor ugh gest non Rela of s as n EdR •P just NYS erv hase tion equ o od Tru ices E-E e, q llus st. io. • In DR uis ava Lore eleife teg : vive ilab m ip er b nd vu rra le a Sed 1.03 land sum ipsu nd lputa adip it te % dolo the m ve Abo te te isci mpu mag r sit cus hicu ng co llus s vo ut E na, ame tom la a sed nse lutp vel izati t, co pulv c. dR qua ven at. vari on nse inar ena t lob us. tha ctetu nec, tis n ortis t is r ad equ gra . Do ipis vida e le nec Fin cing ctus id n a nc ultric unc. in e ing es, ros. era Nam t vita Dev e po vita e ph elop sue are re p me tra Fin orttit nt dui. a nc or, fe Acq Etia ing lis ve m te uisit lit te llus ions mpu velit s , ma Pro ttis pert a D Pro 11 eve per ties yM lo Abb pm revi en t Cas ated each eS con tud serv ten 12 ies ipsu ice t sp offer m do ecifi ing Acq lor c to adip sit am . Lo uisit isci rem Abb ng et, ions eges elit. revi cons Dui tas ated ecte each s la nequ vehi tur con oree serv e, qu cula ten t ipsu ice t sp is vi ac. offer m do ecifi verr ing lor a ip Pro c to Lear adip sit am . Lo sum pert nm isci rem Abb ng ore et, yM eges elit. cons revi abou an a Dui tas ated ecte t Fin each gem s la nequ vehi tur con anci oree serv ent e, qu cula ten ng t ipsu ice t sp is vi ac. offer m do ecifi verr Lear ing lor a ip c to adip nm sit am . Lo sum isci ore rem Abb ng et, abou eges elit. revi cons t Dev Dui tas ated ecte each s la elop nequ vehi tur con oree serv men e, qu cula ten t ipsu t ice t sp is vi ac. offer m do ecifi verr Lear ing lor a ip c to adip nm sit am . Lo sum isci ore rem ng et, abou eges elit. cons t Acq Dui tas ecte s la nequ uisi vehi tur oree tion e, cu @undrstndng ana ge m en t
- 4. Daniel O’Neil Business Analyst,TUG Source: http://www-personal.umich.edu/~phyl/baboon.html @undrstndng Source: http://scan.oxfordjournals.org/content/2/4/323/F2.expansion
- 5. Visualizations are partof our everyday life @undrstndng
- 6. PURPOSE: To provide acore set of principles that transcend best practices @undrstndng
- 7.
- 8.
- 9.
- 10.
- 11.
- 12. Show information inlayers More 2 More 3 Main point More @undrstndng
- 13.
- 14.
- 15. Uhh, we’re nothere to talk about military history…. True: but the lessons of this visualization persist @undrstndng
- 16. The Common DNAof Visualizations All visualizations: • Are rhetorical acts Ask deep value questions – what matters? What do we really care about? How are we going to describe our world? • Are abstractions e.g. Histogram buckets • Work on multiple dimensions Visual, cognitive, emotional, analytical @undrstndng
- 17.
- 18.
- 19. From book, The Artof Clear Up @undrstndng
- 20.
- 21.
- 22.
- 23. Core Principle ofPresenting Data Conversation Based Your data has a point of view, and wishes to start a conversation Source: https://www.earlymoments.com/dr-seuss/How-to-Use-DrSeuss-Book-Clubs/Advanced-Reader-Books/ @undrstndng
- 24. Realizing the ConversationPrinciple Visualizations should be: • Clear • Useful • Ethical • Credible @undrstndng
- 25. Clear Can someone describewhat your chart is trying to do in 2 sentences using simple words? Better yet, can two people look at the chart and give the same basic explanation? Pick a model based on your information, not vice-versa. @undrstndng
- 26. Let’s look atsome examples: Worldwide Nuclear Weapon Detonations A Real-Time Map of Births and Deaths @undrstndng
- 27. Useful Key data pointsare visible without relying on interaction Data is downloadable as a table (assuming interactive data) Is applicable/appropriate to your audience and your goals @undrstndng
- 28. Let’s look atthose examples again: Worldwide Nuclear Weapons Detonations A Real-Time Map of Births and Deaths @undrstndng
- 29. Ethical While designing thechart, write down what you are leaving out and review it. Identify what narrative you are trying to tell and determine if what you are leaving out undermines that narrative. If a story is too complex to tell in a chart, it may not be true. Change scale, proportions, etc on the chart to identify possible distortions of the visual data. @undrstndng
- 30. Source: U.S. CensusBureau, 2011 @undrstndng
- 31. Credible A lot ofthis is an outcome of doing other things right, but there are some things you can do make sure you don’t lose credibility: • Aesthetic decisions • Citations • Proper professional and cultural vernacular for audience (e.g. physicists vs. engineers, dollar vs. euro, children vs. adults) @undrstndng
- 32. U.S. Crime RateTrends Finding: Between 1990 and 2008 there was a forty-five percent decline in violent crime Many theories about this: • Community policing • Improved economic situation • “Tough on Crime” and prisons @undrstndng
- 33. Credibility problems Community policinghappened after initial decline Crime continued to drop even in a bad economy Prison population growth largely made up of nonviolent offenders @undrstndng
- 34. The “Pb” Theory Source:http://science.howstuffworks.com/lead.htm @undrstndng
- 35. Correlate Related Measures Source:http://www.motherjones.com/environment/2013/01/lead-crime-link-gasoline @undrstndng
- 36.
- 37.
- 38.
- 39. Resources • Data VisualizationBest Practices by Jen Underwood • http://www.slideshare.net/idigdata/data-visualization-bestpractices-2013 • More on Abstraction by Kaarin • 20 minute version from IA Summit: http://understandinggroup.com/2013/04/abstraction-forclarity/ • 5 minute version from A2 Ignite UX: http://understandinggroup.com/2013/10/ignite-ux-annarbor-abstraction-talk/ @undrstndng
- 40. Comments? Thoughts? We’d loveto hear from you Daniel O’Neil, @phoenix1189 Kaarin Hoff, @kaarinh www. understandinggroup.com @undrstndng
Editor's Notes
- #2 Welcome to our talk, Listening to Data. We’re very happy to be part of an event like A2 Data Drive – and talking about such a great topic. Today we will be discussing how to make better visualizations by focusing on being conversation based and data driven.
- #4 Hello, my name is Kaarin Hoff – I’m an information architect, or IA for short, at TUG. My background is also quite varied – but visualizations and UM have remained consistent. In undergrad I studied History and Art History, spending a lot of time in archives piecing together information from images and text to get as clear a story as possible. After that, I worked at UM planning orientations from end-to-end -> promotion, registration, ect. That is when I really began to understand the how crucial visualizations are to our everyday life. So many of my meets were disasters – getting so far off track bc a column’s sum wasn’t correct or a website’s purpose wasn’t agreed on because the whole meeting was spent discussing how an image wasn’t aligned correctly. One day someone told me the IA of the site I managed was bad. So I Google “IA” – and shortly there after I applied to grad school bc it sounded so cool that’s what I wanted to do all the time! So I when to SI where I met Dan Klyn by taking his IA course and now I work at his company making useful things out of information and loving it.
- #5 B.A., Cognitive ScienceM.S., Evolutionary PsychologyBusiness analyst, online marketing and IA
- #6 Everyone here has a different story of how they came to be in this room with their current data need – but the great news is we’ve ALL been working with visualizations our whole lives. Who here has made a bar chart? Knows that a pie chart has nothing to do with Pillsbury dough? Exactly. So whether you know 2 charts or 50 charts, the principles we’re talking about here today applyBecause we live in a world full of data questions. We have all this information and now we need to figure out what to do with it. Visualizations are a tool for putting that data to work for you – in convincing your stakeholders, in explaining the problem, even in suggesting solutions.
- #7 The purpose of today’s talk is to provide a core set of principles that transcends best practices. There are no shortage of examples of good and bad visualizations. The trick is choosing the BEST way to visualize your data – so let’s discuss that principles that will allow you to get there.Before we dive in let’s look at some visualizations to think about what they have in common and what they can accomplish
- #8 This visualization was made by Minard in 1869. It is the aspirational, unachievable, Mona Lisa of visualizations. Analyzing the best of the best, something that is still regarded as the best 130+ years later can show us some of the core principles of visualizations. Tufte analyzed this visualization in his book, Beautiful Evidence. Documented
- #9 Tufte analyzed this visualization in his book, Beautiful Evidence.
- #10 Answer the question: Compared to what?That’s 1 out of 42 survival rate
- #11 Show explanationWarmest day was 32 degreesMinard doesn’t say “the cold killed them” – but the evidence being presented together like this declares the relation. We humans make associations when things are placed next to each other like this. Minard used this wisely, since he had many legitimate first- hand accounts of soldiers freezing to death.
- #12 6 variables here : size of army, geographical location, direction of movement (invading & retreating), temperature, timeIf you evidence is numerical and geographical, that’s ok. Combine them to tell the story. Don’t be feel limited to a certain type of visualization, like a line chart, and miss out on telling your whole story. Be true to your data
- #13 Visual can be deep to add evidence – text, images Make your main point right away, but you can give you audience more to digest in they lingerLike a painting…
- #14 Minard’s visualization has stood the test of time and is so revered because of the rigorous documentation attached to it. Everything is proved that he represents here. So no you don’t need to go to the library archives, but do include or refer to where your data is from. Do include a ledged if necessary.
- #15 Minard was a soldier telling soldier stories, exposing the cost of war. Tufte said it best so I’ll read this “That the word ‘Napoleon’ does not appear on the map of Napoleon’s march indicates here at least full attention is to be given to memorializing the dead soldiers rather than celebrating the surviving celebrity.”
- #18 subway - Rhetoric - Sequence trumps location Abstraction - location eliminated Multi-D punch - yin/yang, the power of a loop or line
- #19 What the system would look like with geography…
- #21 Army - Rhetoric - War is suffering? don’t invade Russia? Abstraction - Described in the narrative Multi-D punch - Temperature, location, time, death
- #22 Bell curve - Rhetoric - THIS IS HOW THE WORLD WORKS Abstraction - sample size, histogram bucket size, “p<.05” Multi-D punch - pattern matching, storytelling, profiling----- Meeting Notes (11/8/13 14:00) -----Don't mention histogram
- #23 ----- Meeting Notes (11/8/13 14:00) -----We've looked at some reasons that visualiztions work and we've described the common DNA of visualizations.
- #24 Visualizations as conversation centerpiece - sometimes with people in the room, sometimes with people you will never see - must consider your data, your goal, your audience to pick the right visualizationClearUsefulEthicalCredible
- #26 Pick a model based on your information, not viceversaCan someone describe what your chart is trying to do in about two sentences using simple words.Better yet, can two people look at the chart and give the same basic explanation.
- #27 US Population simulatorhttps://googledrive.com/host/0B2GQktu-wcTiZlAyTTFEaFVuOUk/Too granularAnnoying popup elementsHard to really get sense of time and space because of way data is displayed (scroll on left doesn’t really match scroll on the right)History of Nuclear bomb explosions, by nationhttp://www.youtube.com/watch?v=I9lquok4Pdkright level of granularitydoesn’t overwhelm the viewer with the passage of timeUse of sound, color, and space
- #33 A lot of this is an outcome of doing other things right, but there are some things you can do make sure you don’t lose credibility:aesthetic decisionscitationsproper professional and cultural vernacular for audience (ex, physicists vs. engineers, dollar vs. euro, children vs. adults)aggregate of other good best practices abovehttp://www.motherjones.com/environment/2013/01/lead-crime-link-gasoline
- #34 A lot of this is an outcome of doing other things right, but there are some things you can do make sure you don’t lose credibility:aesthetic decisionscitationsproper professional and cultural vernacular for audience (ex, physicists vs. engineers, dollar vs. euro, children vs. adults)aggregate of other good best practices abovehttp://www.motherjones.com/environment/2013/01/lead-crime-link-gasoline
- #35 A lot of this is an outcome of doing other things right, but there are some things you can do make sure you don’t lose credibility:aesthetic decisionscitationsproper professional and cultural vernacular for audience (ex, physicists vs. engineers, dollar vs. euro, children vs. adults)aggregate of other good best practices abovehttp://www.motherjones.com/environment/2013/01/lead-crime-link-gasoline
- #36 Four charts:drop in crime and leaddrop in pregnancies and lead“compared to what”“Show explanation”“show information in layers”“Credible”“Details matter”http://www.motherjones.com/environment/2013/01/lead-crime-link-gasoline
- #37 Four charts:drop in crime and leaddrop in pregnancies and lead“compared to what”“Show explanation”“show information in layers”“Credible”“Details matter”http://www.motherjones.com/environment/2013/01/lead-crime-link-gasoline
- #38 Four charts:drop in crime and leaddrop in pregnancies and lead“compared to what”“Show explanation”“show information in layers”“Credible”“Details matter”http://www.motherjones.com/environment/2013/01/lead-crime-link-gasoline
- #39 Four charts:drop in crime and leaddrop in pregnancies and lead“compared to what”“Show explanation”“show information in layers”“Credible”“Details matter”http://www.motherjones.com/environment/2013/01/lead-crime-link-gasoline