Design Principles for Visual Communication

8/19/2019 Design Principles for Visual Communication

1/10

60 COMMUNICATIONS OF THE ACM | APRIL 2011 | VOL. 54 | NO. 4

contributed articles

I L L U

S T R A T I O N B Y M A R K S K I L L I C O R N

V I S UA L C O M M U N I C AT I O N V I A diagrams, sketches,charts, photographs, video, and animation isfundamental to the process of exploring conceptsand disseminating information. The most-effective visualizations capitalize on the human facility forprocessing visual information, thereby improvingcomprehension, memory, and inference. Such visualizations help analysts quickly nd patternslurking within large data sets and help audiencesquickly understand complex ideas.

Over the past two decades a number of books 10,15,18,23 have collected examples of effective visual displays.One thing is evident from inspecting them: the bestare carefully crafted by skilled human designers. Yet even with the aid of computers, hand-designingeffective visualizations is time-consuming and

requires considerable effort. More-over, the rate at which people world- wide generate new data is growingexponentially year to year. Gantz et al. 5 estimated we collectively produced161 exabytes of new information in2006, and the compound growth ratebetween 2007 and 2011 would be 60%annually. We are thus expected to pro-duce 1,800 exabytes of information in2011, 10 times more than the amount we produced in 2006. Yet acquiringand storing this data is, by itself, oflittle value. We must understand it to

produce real value and use it to makedecisions.The problem is that human design-

ers lack the time to hand-design effec-tive visualizations for this wealth ofdata. Too often, data is either poorly vi-sualized or not visualized at all. Either way, the results can be catastrophic;for example, Tufte 24 explained howMorton Thiokol engineers failedto visually communicate the risksof launching the Challenger SpaceShuttle to NASA management in 1986,

leading to the vehicle’s disasterousfailure. While Robison et al. 20 arguedthe engineers must not be blamed forthe Challenger accident, better com-munication of the risks might haveprevented the disaster.

Skilled visual designers manipu-late the perception, cognition, and

DOI:10 .1145 /1924421 .1924439

How to identify, instantiate, and evaluatedomain-specic design principles for creatingmore effective visualizations.

BY MANEESH AGRAWALA, WILMOT LI,AND FLORAINE BERTHOUZOZ

DesignPrinciplesfor VisualCommunication

key insights Design principles connect the visual

design of a visualization with theviewer’s perception and cognitionof the underlying information thevisualization is meant to convey.

Identifying and formulating gooddesign principles often requiresanalyzing the best hand-designedvisualizations, examining prior researchon the perception and cognition ofvisualizations, and, when necessary,conducting user studies into howvisual techniques affect perception andcognition.

Given a set of design rules andquantitative evaluation criteria, wecan use procedural t echniques and/or

energy optimization to build automatedvisualization-design systems.


2/10

C R E D I T T K

APRIL 2011 | VOL. 54 | NO. 4 | COMMUNICATIONS OF THE ACM 61

4

13

1

16

11

812

5

6

7

15

142b

3

2a

9

17

18


3/10


ontributed articles

F I G U R E 1

. H A R R Y B E C K ©

T F L F R O M

T H E L O N D O N T R A N S P O R T M U S E U M

C O L L E C T I O N

F I G U R E 2

. ( L E F T ) S T E P H E N B I E S T Y ©

D O R L I N G K I N D E R S L E Y ;

( R I G H T ) L I F E A R T I M A G E S

communicative intent of visualiza-tions by carefully applying principlesof good design. These principles ex-plain how visual techniques can beused to either emphasize importantinformation or de-emphasize irrel-evant details; for example, the mostimportant information in a subwaymap is the sequence of stops along

each line and the transfer stops thatallow riders to change lines. Most subway passengers do not need to knowthe true geographic path of each line.Based on this insight, map designerHarry Beck redesigned the map of theLondon Underground in 1933 usingtwo main principles: straighteningthe subway lines and evenly spacing

the stops to visually emphasize thesequence of stops and transfer points(see Figure 1).

Such design principles connect the visual design of a visualization withthe viewer’s perception and cognitionof the underlying information the vi-sualization is meant to convey. In theeld of design, there is a long-standing

debate regarding the interaction ofaesthetic and functional properties ofdesigned artifacts. We do not seek toengage in this debate here; rather, wefocus on how particular design choic-es affect the perception and cognitionof the visualization, not the aestheticstyle of the visualization. Accordingly, we use the term “design principle” as

a shorthand for guidelines that helpimprove viewers’ comprehension of visually encoded information.

Design principles are usually notstrict rules, but rules of thumb thatmight even oppose and contradictone another. For instance, Beck didnot completely straighten the subway lines; he included a few turns inthem to give viewers a sense of a line’soverall spatial layout. Skilled visualdesigners implicitly apply the relevantdesign principles and balance thetrade-offs between them in an itera-tive process of creating example de-signs, critiquing the examples, andimproving the designs based on thecritiques. Designers usually do notdirectly apply an explicitly dened setof design principles. The principles

are a form of tacit knowledge that de-signers learn by creating and studyingexamples. It is far more common forbooks on visual design to contain vi-sual examples rather than explicit de-sign principles.

Many of the analysts and end usersinundated with data and charged withcreating visualizations are not traineddesigners. Thus, our work aims toidentify domain-specic design prin-ciples, instantiating them within au-tomated visualization design systems

that enable non-designers to createeffective visual displays. While otherresearchers have considered specic ways to use cognitive design princi-ples to generate visualizations (see theonline appendix) we have been devel-oping a general, three-stage approachfor creating visualization design sys-tems:

Figure 1. Harry Beck’s map of the London Underground from 1933. Beck straightened thelines and more evenly spaced the stops to visually emphasize the sequence of stops alongeach line.

Figure 2. Hand-designed cutaway and exploded-view illustrations (left) design the cuts and explosions to emphasize the shape of the missing

geometry and spatial relationships among par ts. Our system incorporates such principles to generate interactive cutaway and exploded-view illustrations (middle, right).


4/10



F I G U R E 4

. D A V I D M A C A U L A Y

, T H E N E W

W A Y T H I N G S W O R K

Stage 1. Identify design principles. We identify domain-specic designprinciples by analyzing the best hand-designed visualizations within a par-ticular information domain. We con-nect this analysis with research onperception and cognition of visualiza-tions;

Stage 2. Instantiate design principles. We encode the design principlesinto algorithms and interfaces for cre-ating visualizations; and

Stage 3. Evaluate design principles. We measure improvements in infor-mation processing, communication,and decision making that result fromour visualizations. These evaluationsalso serve to validate the design prin-ciples.

We have used this three-stage ap-

proach to build automated visualiza-tion design systems in two domains:cartographic visualization and tech-nical illustration. In the domain ofcartographic visualizations we havedeveloped automated algorithms forcreating route maps 1,3,12 and touristmaps of cities. 8 In the domain of tech-nical illustration we have developedautomated techniques for generatingassembly instructions of furnitureand toys 2,9 and for creating interactivecutaway and exploded-view illustra-

tions of complex mechanical, mathe-matical, and biological objects. 11,13,14,19

Here, we focus on articulating thetechniques we have used to identifyand evaluate the design principlesfor each domain. These techniquesgeneralize to other domains, and ap-plying our three-stage approach willresult in a better understanding of the

strategies people use to make infer-ences from visualizations.

Stage 1. Identify Design PrinciplesDesign principles are prescriptiverules describing how visual tech-niques affect the perception and cog-nition of the information in a display.

In some cases, they are explicitly out-lined in books; for example, books onphotography techniques explain therules for composing pleasing photo-graphs (such as cropping images ofpeople just below the shoulders ornear the waist, rather than at the neckor the knees). Researchers have di-rectly applied them to build a variety

of automated photo-manipulation al-gorithms (see the online appendix forexamples).

However, our experience is that de-sign principles are rarely stated so ex-plicitly. Thus, we have developed threestrategies for extracting and formulat-ing domain-specic design principles:(1) analyze the best hand-designed vi-sualizations in the domain, (2) exam-ine prior research on the perceptionand cognition of visualizations, and, when necessary, (3) conduct new userstudies that investigate how visualtechniques affect perception and cog-nition.

Hand-designed visualizations. Wehave found that a useful rst step inidentifying design principles is toanalyze examples of the best visual-

izations in the domain. This analysisis designed to nd similarities and re-curring patterns in the kinds of infor-mation the visualizations highlight, as well as the techniques used to empha-size the information.

Consider the problem of depictingthe internal structure of complex me-chanical, mathematical, anatomical,and architectural objects. Illustratorsoften use cutaways and exploded viewsto reveal such structure. They careful-ly choose the size and shape of cuts,

as well as the placement of the partsrelative to one another, to expose andhighlight the internal structure andspatial relationships between parts. We have analyzed a large corpus of cut-aways and exploded views to identifythe principles and conventions expertillustrators commonly use to generatethese images. 11,13,14,19 Our process for

Figure 4. Hand-designed “how things work” illustrations (a) use motion arrows and frame sequences to convey the motion and interactions

of the parts within a mechanical assembly. Our system analyzes a geometric model (b) of a mechanical assembly to infer the motion andinteractions of the parts, then generates the motion arrows and frame sequences (c–d) necessary to depict how the assembly works.

Figure 3. Exploded views of complexmathematical surfaces are designedto reveal local geometric features (suchas symmetries, self-intersections, andcritical points).


5/10


ontributed articles

F I G U R E 5

. M A P P O I N T S C R E E N S H O T R E P R I N T E D W I T H P E R M I S S I O N O F M I C R O S O F T C O R P

.

F I G U R E 6

. G O O G L E M A P S

identifying these principles is basedon three main objectives:

Style independence. In order to

identify a general set of principles wecould apply to a variety of complex3D objects, we looked for visual tech-niques common across different artis-tic styles and types of objects;

Generative rules. To ensure that wecould apply the principles in a gen-erative manner to create cutaways orexploded views, we formed explicit, well-dened rules describing whenand how each principle should be ap-plied. We designed the rules to be asgeneral as possible while remaining

consistent with the evidence from theexample illustrations; and

Perceptual/cognitive rationale. Wemotivated each principle by hypoth-

esizing a perceptual or cognitive ratio-nale explaining how the conventionhelps viewers better understand the

structure of the 3D object depicted.Through this analysis, we iden-tied a set of general, perceptuallymotivated design principles for creat-ing cutaways and exploded views. Forinstance, the size and shape of cutsin a cutaway illustration are often de-signed to not only reveal internal partsbut to help viewers mentally recon-struct any occluding geometry thathas been removed. Thus, illustratorscut radially symmetric objects with wedge-shape cutaways that empha-

size the object’s cylindrical structure.Similarly, rectangular objects are cut with object-aligned cutting planes,or box cuts; skin-like covering sur-

faces are cut using window cuts; andlong tubular structures are cut usingtransverse tube cuts. Illustrations of

complex mathematical surfaces of-ten use exploded views in which eachslice is positioned to reveal local geo-metric features (such as symmetries,self-intersections, and critical points). We have also examined “how things work” illustrations designed to showthe movement and interaction ofparts within a mechanical assembly.The hand-designed illustrations oftenuse diagrammatic motion arrows andsequences of frames to help viewersunderstand the causal chains of mo-

tion that transform a driving forceinto mechanical movement. Afteridentifying the design principles, weimplemented them algorithmically within interactive systems for gener-ating cutaways, exploded views, andhow-things-work illustrations (seeFigures 2, 3, and 4).

We applied a similar approach toidentify the design principles for de-picting route maps that provide direc-tions from one location to another 1,3 and destination maps that show mul-tiple routes from all around a region toa single location (such as an airport ora popular restaurant). 12 We analyzed a variety of such hand-drawn maps andfound they are often far more usefulthan computer-generated driving di-rections (available at sites like maps.bing.com and maps.google.com) be-cause they emphasize roads, turningpoints, and local landmarks. Thesehand-designed maps signicantly dis-tort the distance, angle, and shape ofroads while eliminating many detailsthat would only serve to clutter the

Figure 6. A general-purpose computer-generated map of San Francisco (left) is notan effective destination map because it is cluttered with extraneous information andneighborhood roads disappear. Our destination map (right) includes only the relevant

highways, arterials, and residential roads required to reach a destination. The layout andrendering style further emphasize the information required to reach it.

Figure 5. A computer-generated route map rendered at a xed scale does not depict (left) all the turns necessary for navigation. A hand-designed map (middle) emphasizes the turning points by exaggerating the lengths of short roads and simplifying the shape of roads. OurLineDrive system incorporates these design principles (right) into an automated map-design algorithm.


6/10



F I G U R E 7 . U N I Q U E M E D I A

map. Tufte 23 pointed out that triptiksand subway maps similarly distortthe shape of routes and eliminateunnecessary detail. Hand-designeddestination maps include only themajor routes to a location rather thanall possible routes. These maps pro-gressively increase the level of detail,showing only the highways far fromthe destination while including arte-rial roads and nally the residentialroads near the destination. Both routeand destination maps typically usemulti-perspective rendering in whichthe roads are drawn in top-down plan view while important landmark build-ings are drawn from a side view sotheir facades are visible.

Although analyzing hand-designed visualizations is often a good initial

approach for identifying design prin-ciples, this strategy also involves limi-tations. In some cases it may be tempt-ing to form generative rules that aretoo specic and do not apply outsidethe range of analyzed examples. Inother cases the rules may be so generalit is unclear how to apply them to spe-cic examples. Such difculties oftenarise when the perceptual or cognitiverationale behind a particular visualtechnique is not clear. In the contextof route maps, for example, although

our analysis revealed that mapmakersoften distort road length, angle, andshape, it was not immediately clearhow such distortions improved the per-ception and cognition of route maps.

Similarly, we have found that oneof the challenges in analyzing hand-designed visualizations is to factor

out differences due to artistic style.Designers may choose visual attri-butes (such as font type, color palette,and line width) for aesthetic reasons whereby one font may simply looknicer than another to the designer. Al-though such aesthetic design choicesare important considerations, thegoal of our analysis is to determinehow the design choices improve theperception and cognition of the infor-mation, rather than how these choicesimprove aesthetics. The difculty isthat these design choices often affectboth the aesthetics of the display andthe perception and cognition of theinformation; how to separate the twoeffects is not always clear.

In light of these limitations andchallenges, we have found it is often

useful to connect our observationsand hypotheses from the analysis ofhand-designed examples with relevant work from perception and cognitivepsychology. These connections serveto clarify the perceptual or cognitiverationale for the design principles.

Prior work in perception and cog-nition. In some cases, prior researchin perception and cognition suggestsor formalizes the appropriate designprinciples; for example, cognitive psy-chologists have shown that people

think of routes as a sequence of turns25

and that when following a route theexact length of a road is far less impor-tant than properly executing the turns.The topology of the route is more im-portant than its absolute geometry.This insight helps explain why hand-drawn maps often distort geometry—

distance, angle, and shape of roads—to ensure that all roads and turningpoints are visible, but almost nevermodify the topology of the route.

In this case, the prior research con-rmed and formalized the perceptual/cognitive rationale for the visual tech-niques we rst noticed when analyz-ing hand-drawn route maps. Based onthe resulting design principles, we developed LineDrive (http://vis.berkeley.edu/LineDrive), a fully automated sys-tem for rendering route maps in thestyle of hand-drawn maps. 3 LineDrivehas been publicly accessible since Oc-tober 2000, and surveys have shownthat for navigation tasks users strong-ly prefer LineDrive maps to computer-generated maps drawn at a xed scale(see Figure 5).

Researchers have also found thatnavigators familiar with a geographicarea (such as cab drivers) plan routeshierarchically. 4 They rst select thehighways necessary to get close to thedestination, then the arteries, and -nally the residential streets. Such hi-erarchical planning corresponds tothe progressive increase in road detail we rst identied in hand-designeddestination maps. We recently ap-plied this level-of-detail principle inconjunction with the distortion prin-

ciples to build an automated systemfor generating destination maps. 12 Asin LineDrive, we produced a map thatlooks hand-drawn but that eliminatesclutter while preserving the infor-mation necessary for anyone in thesurrounding region to reach the des-tination. Our destination maps are

Figure 7. A hand-designed tourist map of San Francisco emphasizes semantically, visually, and structurally important landmarks, paths,

districts, nodes, and edges, using multi-perspective rendering to ensure the facades of buildings are visible (left). Our tourist-map designsystem is based on these principles and similarly emphasizes the information most important for tourists in this map of San Francisco (right).


7/10


ontributed articles

available through the Bing Maps Website (http://vis.berkeley.edu/DestMap)(see Figure 6).

We applied a similar approach toautomatically generating maps fortourists visiting a new city. 8 Prior workon mental representations of cities 16 showed that people consider ve mainelements: landmarks, paths, districts,nodes, and edges. However, a map with every instance of such elements would be cluttered with excessive de-tail. The most-effective tourist mapsinclude only those elements that aresemantically meaningful (such as thehome of a well-known writer), visuallydistinctive (such as an oddly shapedor colored building), or placed in astructurally important location (suchas a building at a prominent intersec-

tion).22

After choosing the elements toinclude in the map, mapmakers usu-ally apply a variety of cartographic-generalization techniques, includingsimplication, displacement, defor-mation, and selection. Cognitive psy-chologists and cartographers study-ing the cognition of maps have shownsuch generalizations improve claritybecause they emphasize the most im-portant map elements while preserv-ing spatial relationships betweenthese elements. 17

Our tourist-map-design system isbased on these design principles. In-put consists of a geometric model of acity, including streets, bodies of water,parks, and buildings (with textures).The system automatically determinesthe importance of map elements us-

ing top-down Web-based information-extraction techniques to computesemantic importance and bottom-up vision-based image/geometry analy-sis to compute visual and structuralimportance. It then generates a mapthat emphasizes the most importantmap elements, using a combinationof multi-perspective rendering andcartographic generalization to high-light the important landmarks, paths,districts, nodes, and edges while de-emphasizing less-important elements(see Figure 7).

Experiments on perception andcognition. In some domains, new per-ception and cognition research is re-quired to provide the rationale for thedesign principles. Working with cog-nitive psychologist Barbara Tversky,

we developed a methodology for con-ducting human-subject experimentsto understand how people think aboutand communicate the information within a domain. We rst applied thismethodology to identify the designprinciples for creating assembly in-structions for everyday objects (suchas furniture and toys). 2,9 The experi-ments are conducted in three phases:

Production. Participants create vi-sualizations for a given domain. Inthe context of assembly instructions,

they assembled a TV stand without in-structions using only a photograph ofthe assembled stand as a guide. Theythen drew a set of instructions show-ing how to assemble it;

Preference. Participants rate the ef-fectiveness of the visualizations. In

the assembly-instructions project, anew set of participants assembled theTV stand, without instructions. Theythen rated the quality of the instruc-tions created by the rst set of partici-pants, redrawn to control for clarity,legibility, and aesthetics; and

Comprehension. Participants usethe ranked visualizations, and we testfor improvements in learning, com-prehension, and decision making. Inthe assembly-instructions project, yetanother set of participants assembledthe TV stand, this time using the in-structions rated in the preferencephase. Tests showed the highly ratedinstructions were easier to use andfollow; participants spent less timeassembling the TV stand and madefewer errors.

Following these experiments, welook for commonalities in the highlyrated visualizations to identify thedesign principles. In the context ofassembly instructions, we identiedthree main principles: (1) use a step-by-step sequence of diagrams show-ing one primary action in each dia-gram; (2) use guidelines and arrows todepict the actions required to t partstogether; and (3) ensure that the partsadded in each step are visible. Our au-tomated assembly-instruction-design

system is based on these principles(see Figure 8). Tversky and Lee 25 havestudied mental representations ofmaps using a similar methodology, where subjects rst draw maps to fa-miliar locations, then other subjectsrate the effectiveness of the maps.

Figure 8. We asked subjects to assemble a TV stand and then create instructions for a novice explaining how to assemble it (left, middle).Analyzing hand-drawn instructions, we found that diagrammatic, step-by-step instructions using guidelines and arrows to indicate the

actions required for assembly and providing good visibility for the attached parts are easiest to use and follow. Our system automaticallygenerates assembly instructions (right) based on these principles.


8/10



Stage 2. InstantiateDesign PrinciplesDesigning a visualization usually re-quires choosing visual properties orattributes for each element in thedisplay; for example, to create a routemap, the designer must choose attri-butes, including position, size, andorientation for each road, landmark,and label that appears in the map.Similarly, to create a cutaway illustra-tion, the designer must choose howand where to cut each structure thatoccludes the target part. Because thereare many possible choices for each at-tribute, the design space of possible visualizations is usually quite large. Tobuild automated visualization designsystems, we treat the relevant designprinciples as guidelines for making

these design decisions. The principleshelp us navigate through the designspace and obtain an effective design.

Most design principles are statedas qualitative guidelines, rather thanas procedures we can directly instan-tiate in an automated design algo-rithm. The challenge is to transformsuch high-level principles into imple-mentable algorithms.

Design principles generally fallinto two categories: design rules andevaluation criteria. Design rules sepa-

rate the design space into regions con-taining effective designs and thosecontaining inviable designs. They areessentially hard constraints in the de-sign space. In creating route maps, forexample, designers commonly adjustthe turn angle to emphasize the ori-entation of the turn, to the left or tothe right. However, adjusting the turnangle so much that a left turn appearsto be a right turn or vice versa is unac-ceptable. This design rule puts a hardconstraint on how much designers areable to adjust the turn angle.

Evaluation criteria quantify theeffectiveness of some aspect of the visualization. We can assess the over-all effectiveness of a visualization byconsidering a set of evaluation crite-ria covering all major aspects of the visual design. In creating an exploded view, for instance, designers must bal-ance two such criteria: part separationand compactness. A good exploded view separates the parts so all of themare visible, yet the visualization mustalso remain compact and maintain a

roughly square aspect ratio to makethe best use of available screen space.To quantify the overall effectivenessof an exploded view we measure the visibility of each part, as well as thecompactness of the overall visualiza-tion. Similarly, in designing routemaps, designers must ensure that allroads are visible. To quantify this cri-terion, we compute the length of eachroad in the map and check the lengthis greater than some minimum vis-ibility threshold. The number of roadslonger than the threshold length is aquantitative measure of the effective-ness of the map with respect to thiscriterion.

Given a set of design rules andquantitative evaluation criteria, wecan use procedural techniques to

build an automated visualization de-sign system; for example, our systemfor designing cutaways and exploded views is driven exclusively by proce-dural techniques. In this case, we en-code the design rules as a decisiontree describing how to cut or explodeaway occluding parts based on theirgeometry. Another approach is to con-sider visualization design as an ener-gy-minimizing optimization problem.In this case, we treat the design rulesas hard constraints that dene the

boundaries of the design space andthe evaluation criteria as soft con-straints that guide the system to theoptimal visualization. While this op-timization-based approach is general, we have found it essential to develop aset of design rules and evaluation cri-teria that sufciently limit the designspace so it is feasible to complete theoptimization. Both LineDrive and ourassembly-instruction design systemuse such an energy-minimizing opti-mization.

Stage 3. Evaluate Design PrinciplesThe nal stage of our approach is tomeasure the usefulness of the visual-izations produced by our automateddesign systems. We consider severalsuch measures, including feedbackfrom users in the form of qualitativeinterviews and quantitative usage sta-tistics. In some cases, we have alsoconducted more-formal user studiesto check how well the visualizationsimprove information processing, com-munication, and decision making.

These principlesexplain how

visual techniquescan be used toeither emphasizeimportantinformation orde-emphasizeirrelevant detailsin the display.


9/10


ontributed articles

User feedback. We nd it is criticalto involve users early on and conductqualitative interviews and surveys tocheck their overall impressions of the visualizations produced by our sys-tems. Such feedback is essential foridentifying problems and ensuringour design principles and the visual-izations converge on effective designs.The interviews and surveys providehigh-level checks of the effectivenessof our design principles and allow usto tweak the principles when not quiteright; for example, early on buildingLineDrive, we asked users to rate hand-crafted prototype route-map designs,nding that 79 out of 90 respondentspreferred the distorted LineDrive pro-totypes to maps drawn to scale 1 andconrming that users thought the dis-

torted maps were useful. Continualfeedback and evaluation yields more-effective algorithms and tools.

Another approach is to release the visualization on the Web, then checkusage statistics; for example, at itspeak, LineDrive was serving more than750,000 maps per day and became thedefault route-mapping style for Map-Blast, an early Web-based provider ofroute maps. Such public feedback isa strong test of effectiveness, as inef-fective solutions are quickly rejected.

We also recognize that usage statisticsare at best an indirect measure of ef-fectiveness. Many excellent solutionsremain little-used due to a variety ofexternal forces that have little to do with the usefulness or effectiveness ofa visualization.

User studies. To quantitatively as-sess the effectiveness of a visualiza-tion, we conduct user studies com-paring visualizations created with ourdesign algorithms to the best hand-designed visualizations in the do-main; for example, we have comparedour computer-designed instructionsto factory-produced instructions andhand-drawn instructions for assem-bling a TV stand, nding that userscompleted the assembly task about35% faster and made 50% fewer errorsusing our instructions. In additionto completion time and error rate, itis also possible to use eye-tracking todetermine how a visualization affectsthe way people scan and process in-formation. 6,21 Such eye-tracking stud-ies help us evaluate the effectiveness

of low-level design choices in creating visualizations. Rigorous user studiesare especially important because theyalso serve to validate the effectivenessof the design principles on which the visualizations are based.

However, how to design such quan-titative studies is not always clear.How should one visualization be com-pared against another visualization?For example, in the domain of ana-tomical illustrations it is not clear howto compare our cutaway illustrationsagainst hand-designed illustrations. What task should we ask users to per-form using the two illustrations? Oneapproach might be to measure howquickly and accurately viewers locatea particular organ of the body. How-ever, if the task is to learn the location

of the organ, then both illustrations would label the organ, and with la-bels, speed and accuracy are unlikelyto differ signicantly. Our cutawaysand exploded views are also designedto convey the layering relationshipbetween parts. So, an alternative taskmight be to ask viewers to indicate thelayering relationships between parts.But how can we ask them to completethis task without leading them toan answer? For many domains, likeanatomical illustrations, developing

a new methodology is necessary forevaluating the effectiveness of visual-izations and validating underlying de-sign principles.

ConclusionThe approach we’ve outlined for iden-tifying, instantiating, and evaluatingdesign principles for visual commu-nication is a general methodology forcombining ndings about human per-ception and cognition with automateddesign algorithms. The systems we’vebuilt for generating route maps, tour-ist maps, and technical illustrationsdemonstrate this methodology can beused to develop effective automated visualization-design systems. Howev-er, there is much room for extendingour proposed approach, and we hoperesearchers improve on the methods we have described. Future work cantake several directions:

Many other information domainscould benet from a deeper under-standing of the ways visual-displaytechniques affect the perception and

Many otherinformation

domainscould benetfrom a deeperunderstandingof the waysvisual-displaytechniques affectthe perception

and cognitionof information.


10/10



cognition of information. We com-monly encounter a variety of differenttypes of information, including cook-ing recipes, budgets and nancialdata, dance steps, tutorials on usingsoftware, explanations of strategiesand plays in sports, and political poll-ing numbers. Effective visualizationsof such everyday information couldempower citizens to make better deci-sions.

We have focused our work on iden-tifying domain-specic design prin-ciples. An open challenge is to gener-alize them across multiple domains.One approach might be to rst identifydomain-specic design principles in very different domains, then look forcommonalities between the domain-specic principles; for example, we

recently developed an automated sys-tem for generating tutorials explain-ing how to manipulate photographsusing Photoshop and GIMP. 7 The de-sign principles for photo-manipula-tion tutorials are similar to those weidentied for assembly instructionsand include step-by-step sequences ofscreenshots and highlighting actionsthrough arrows and other diagram-matic elements. Finding such simi-larities in design principles acrossmultiple domains may indicate more

general principles are at work.Though we presented three strate-

gies for identifying design principles,other strategies may be possible as well. The strategies we presentedall require signicant human effortto identify commonalities in hand-designed visualizations, synthesizethe relevant prior studies in percep-tion and cognition, and conduct suchstudies. Moreover, the Internet makesa great deal of visual content publiclyavailable, often with thousands ofexample visualizations within an in-dividual information domain. Thus,a viable alternative strategy for iden-tifying design principles may be tolearn them from a large collection ofexamples using statistical machine-learning techniques. We have takenan initial step in this direction, witha project designed to learn how to la-bel diagrams from a few examples. 26 One advantage of this approach is thatskilled designers often nd it easier tocreate example visualizations than ex-plicitly describe design principles.

by demonstration. ACM Transactions on Graphics 27 , 3(Aug. 2009), 66:1–66:9.

8. Grabler, F., Agrawala, M., Sumner, R.W., andPauly, M. Automatic generation of tourist maps.ACM Transactions on Graphics 27 , 3 (Aug. 2008),100:1–100:11.

9. Heiser, J., Phan, D., Agrawala, M., Tversky, B., andHanrahan, P. Identication and validation of cognitivedesign principles for automated generation ofassembly instructions. In Proceedings of AdvancedVisual Interfaces (Gallipoli, Italy, May 25–28). ACMPress, New York, 2004, 311–319.

10. Hodges, E. The Guild Handbook of ScienticIllustration. Van Nostrand Reinhold, New York, 1989.

11. Karpenko, O., Li, W., Mitra, N., and Agrawala, M.Exploded-view diagrams of mathematical surfaces.IEEE Transactions on Visualization and ComputerGraphics 16 , 6 (Oct. 2010), 1311–1318.

12. Kopf, J., Agrawala, M., Salesin, D., Bargeron, D., andCohen, M.F. Automatic generation of destination maps.ACM Transactions on Graphics 29 , 6 (Dec. 2010),158:1–158:12.

13. Li, W., Agrawala, M., Curless, B., and Salesin, D.Automated generation of interactive exploded-viewdiagrams. ACM Transactions on Graphics 27 , 3 (Aug.2008), 101:1–101:11.

14. Li, W., Ritter, L., Agrawala, M., Curless, B., and Salesin,D. Interactive cutaway illustrations of complex 3Dmodels. ACM Transactions on Graphics 26 , 3 (July2007), 31:1–31:11.

15. London, B. and Upton, J. Photography. Longman

Publishing Group, New York, 1997.16. Lynch, K. The Image of the City. The MIT Press,Cambridge, MA, 1960.

17. MacEachren, A.M. How Maps Work. The GuilfordPress, New York, 1995.

18. Mijksenaar, P. and Westendorp, P. Open Here: The Artof Instructional Design. Joost Elffers Books, NewYork, 1999.

19. Mitra, N.J., Yang, Y., Yan, D., Li, W., and Agrawala,M. Illustrating how mechanical assemblies work.ACM Transactions on Graphics 29 , 4 (July 2010),29:58:1–58:12.

20. Robison, W., Boisjoly, R., Hoeker, D., and Young, S.Representation and misrepresentation: Tufte and theMorton Thiokol engineers on the Challenger. Scienceand Engineering Ethics 8 , 1 (Jan. 2002), 59–81.

21. Santella, A., Agrawala, M., DeCarlo, D., Salesin, D.,and Cohen, M. Gaze-based interaction for semi-automatic photo cropping. In Proceedings of the

SIGCHI Conference on Human Factors in ComputingSystems (Montréal, Apr. 24–27). ACM Press, NewYork, 771–780.

22. Sorrows, M. and Hirtle, S. The nature of landmarksfor real and electronic spaces. In Proceedings of theInternational Conference on Spatial InformationTheory (Stade, Germany, Aug. 25–29). Springer-Verlag,London, U.K., 1999, 37–50.

23. Tufte, E. Visual Explanations. Graphics Press, Cheshire,CT, 1997.

24. Tufte, E. Envisioning Information. Graphics Press,Cheshire, CT, 1990.

25. Tversky, B. and Lee, P. Pictorial and verbal tools forconveying routes. In Proceedings of the InternationalConference on Spatial Information Theory (Stade,Germany, Aug. 25–29). Springer-Verlag, London, U.K.,1999, 51–64.

26. Vollick, I, Vogel, D., Agrawala, M., and Hertzmann, A.Specifying label layout by example. In Proceedings ofthe ACM Symposium on User Interface Software andTechnology (Newport, RI, Oct. 7–10). ACM Press, NewYork, 2007, 221–230.

Maneesh Agrawala ([email protected]) isan associate professor in the Electrical Engineering andComputer Sciences Department of the University ofCalifornia, Berkeley.

Wilmot Li ([email protected]) is a research scientist inthe Creative Technologies Lab of Adobe Systems Inc., SanFrancisco, CA.

Floraine Berthouzoz ([email protected]) is a Ph.D. candidate in the Electrical Engineeringand Computer Sciences Department of the University ofCalifornia, Berkeley.

© 2011 ACM 0001-0782/11/04 $10.00

Techniques for evaluating the ef-fectiveness of visualizations and vali-dating the design principles couldalso be improved. Design principlesare essentially models that predicthow visual techniques affect percep-tion and cognition. However, as wenoted, it is not always clear how tocheck the effectiveness of a visualiza-tion. More sophisticated evaluationmethodology could provide strongerevidence for these models and there-by experimentally validate the designprinciples.

Acknowledgments We would like to thank DavidBargeron, Michael Cohen, BrianCurless, Pat Hanrahan, John Hay-maker, Julie Heiser, Olga Karpenko,

Jeff Klingner, Johannes Kopf, NiloyMitra, Mark Pauly, Doantam Phan,Lincoln Ritter, David Salesin, ChrisStolte, Robert Sumner, Barbara Tver-sky, Dong-Ming Yan, and Yong-LiangYang for their contributions to thisresearch. Jeff Heer, Takeo Igarashi,and Tamara Munzner provided ex-cellent suggestions and feedback onearly drafts of this article. Figure 4ais from The New Way Things Work byDavid Macaulay: compilation copy-right © 1988, 1998 Dorling Kindersley,

Ltd., London; illustrations copyright© 1988, 1998 David Macaulay. Used bypermission of Houghton Mifin Har-court Publishing Company. All rightsreserved.

References1. Agrawala, M. Visualizing Route Maps. Ph.D. Thesis,

Stanford University, Stanford, CA, 2002; http://graphics.stanford.edu/papers/maneesh_thesis/

2. Agrawala, M., Phan, D., Heiser, J., Haymaker, J., andKlingner, J. Designing effective step-by-step assemblyinstructions. In Proceedings of SIGGRAPH (San Diego,July 27–31). ACM Press, New York, 2003, 828–837.

3. Agrawala, M. and Stolte, C. Rendering effective routemaps: Improving usability through generalization. InProceedings of SIGGRAPH (Los Angeles, Aug. 12–17).ACM Press, New York, 2001, 241–250.

4. Chase, W.G. Spatial representations of taxi drivers.In Acquisition of Symbolic Skills , D.R. Rogers andJ.A. Sloboda, Eds. Plenum Press, New York, 1983,391–405.

5. Gantz, J., Chute, C., Manfrediz, A., Minton, S., Reinsel,D., Schlichting, W., and Toncheva, A. The Diverse andExploding Digital Universe: An Updated Forecast ofWorldwide Information Growth Through 2011. IDCWhite Paper, Mar. 2008; http://www.emc.com/about/destination/digital universe/

6. Goldberg, J., Stimson, M., Lewenstein, M., Scott,N., and Wichansky, A. Eye tracking in Web searchtasks: Design implications. In Proceedings of the EyeTracking Research and Applications Symposium (NewOrleans, Mar. 25–27). ACM Press, New York, 2002,51–58.

7. Grabler, F., Agrawala, M., Li, W., Dontcheva, M., andIgarashi, T. Generating photo manipulation tutorials

Design Principles for Visual Communication

Documents

Transcript of Design Principles for Visual Communication