adapting user-centered design methods to design for ......user-centered design (ucd) methods such as...

51

Cynthia PutnamDoctoral CandidateDept. of Human CenteredDesign & Engineering423 Sieg Hall, Box 352315University of WashingtonSeattle, WA 98195-2315 USA(206) 543-2567

Emma RoseDoctoral CandidateDept. of Human CenteredDesign & Engineering423 Sieg Hall, Box 352315University of WashingtonSeattle, WA 98195-2315 USA(206) 543-2567

Erica J. JohnsonPost-Doctoral Fellow in CentralAsian StudiesCenter for Eurasian, Russianand East European StudiesEdmund A. Walsh School ofForeign ServiceGeorgetown UniversityBox 571031, ICC 111Washington, DC 20057-1031USA(202) 687-6080

Beth KolkoAssociate ProfessorDept. of Human CenteredDesign & Engineering423 Sieg Hall, Box 352315University of WashingtonSeattle, WA 98195-2315 USA(206) 543-2567

Adapting User-Centered Design Methods to Design for Diverse Populations PUTNAM, ROSE, JOHNSON, KOLKO

Research Article

Adapting User-Centered DesignMethods to Design for DiversePopulationsAbstract

User-centered design (UCD) methods such as personas and scenarios are use-ful tools for communicating extensive user research to design teams. However,the intensive data collection and onsite ªeldwork required to create personasand scenarios can present barriers for user researchers and design teamsworking on technology innovations for diverse populations, including those indeveloping regions. This article presents a model that demonstrates how userresearchers and design teams can use data collected for other purposes tosource personas and scenarios and to substantively adopt user-centered designapproaches. We demonstrate our methods based on research conducted in acase study in Kyrgyzstan, a country in Central Asia.

IntroductionAs designers and researchers who have worked in settings in both theUnited States and in developing regions, we have come to recognize thattraditional approaches to user-centered design (UCD) methods are difªcultto deploy in developing regions (Kam, Bhagwani, Kuman, Lal, Mathur etal., 2007; Maunder, Marsden, Gruijters, & Blake, 2007). These difªcultiesare not conªned to developing regions, but also apply to potential usersfrom a substantively different cultural context (i.e., diverse populations)than that in which technology design occurs. UCD is part of themultidisciplinary ªeld of human-computer interaction (HCI). HCI embracesmany disciplines, including (but not limited to) computer science, cogni-tive psychology, anthropology, graphic design, and industrial design. WhileHCI broadly explores how humans interact with technology, UCD is adesign philosophy that emphasizes the importance of considering the enduser’s needs, goals, and desires when creating products or services. Thisarticle reports on our attempt to adapt two traditional UCD methods (per-sonas and scenarios) so that they are more easily adoptable for research-ers who are working with diverse audiences and those who want tocommunicate their research to design teams.1 This work also recognizesthat it is difªcult and resource intensive to conduct user research in devel-oping or culturally distant regions, and our goal is to demonstrate waysthat researchers can use unconventional data sources to communicatewith design teams about end users.

A persona is an archetypal character that represents a group of users

1. In this article, the term design team includes professionals responsible for interaction design, deªned by Cooper,Reimann, and Cronin (2007) as “the practice of designing interactive digital products, environments, systems and ser-vices.” Actual job titles could include graphic designers, developers, interaction designers, information architects, tech-nical writers, interface designers, product designers, etc.

© 2009 USC Annenberg School for Communication & Journalism. Published under Creative Commons Attribution-NonCommercial-Share Alike 3.0 Unported

license. All rights not granted thereunder to the public are reserved to the publisher and may not be exercised without its express written permission.

Volume 5, Number 4, Winter 2009, 51–73

who share common goals, attitudes, and behaviorswhen interacting with a particular product or service(Cooper, Reimann, & Cronin, 2007; Mulder & Yaar,2007; Pruitt & Adlin, 2006). A scenario in a UCDcontext is a story that describes how a characterperforms an activity in relation to a product (Carroll,2000b; Go & Carroll, 2004; Quesenbery, 2006;Rosson & Carroll, 2003). Although there is debatewithin the UCD community about the persona-and-scenario approach (see Chapman & Milham, 2006;Portigal, 2008), this article starts with the assump-tion that personas and scenarios are useful in certaincircumstances, and we address how to make thesedesign approaches more viable in diverse design set-tings. Traditional research employed to create perso-nas and scenarios has focused on users inrelationship to a speciªc product or service. Ourresearch, however, did not ªt the traditional mold asit was not associated with a speciªc product. Rather,it was culled from existing general social research.

As members of the Central Asia � Informationand Communication Technology project (CAICT), amultiyear study of the use of information and tech-nologies in Central Asia, we have collected thou-sands of data points pertaining to attitudes,behaviors, and goals in relation to technologies usedthroughout the region. CAICT’s ªeld research, then,was not about the design of a speciªc product orservice. Rather, it was general research collectedover several years, some of which had a designfocus, and some of which was geared to largersocial issues. However, as the research includedtechnology usage and trends, we realized that thedata we collected could be leveraged to build perso-nas and scenarios to communicate about CentralAsian end users. Summarizing user research intousable and actionable documentation for designteams is always challenging and even more so whenrelating information about a culture or group withwhich the design team is unfamiliar. Our analysisand discussion here focuses on the developingworld, but this approach can be used when design-ing for diversity in general, whether for economicallydisadvantaged users or for other resource-constrained communities.

Because personas and scenarios must be createdin relation to a product or service, we ªrst neededto ideate a product concept. What emerged fromour data, targeted at mobile phone users, was amobile social software (MoSoSo) directory thatenabled users to access listings for local businesses,

rate local businesses, view ratings of localbusinesses, and create password-protected shareddirectories that could contain business listings, rat-ings, and message boards. We focused our initialprototype design efforts on SMS/text access to thedirectory, but we also envisioned voice accessthrough telephony phone trees in later iterations.

The persona-and-scenario approach resonatedwith our general sense of how to communicate ourdata to design teams, but after looking at how mostof the literature discusses personas and scenarios, itbecame clear that adapting those methods wouldbe necessary. We had very rich data that were notcollected with the intent of creating personas andscenarios, but that nevertheless allowed us to usethese traditional UCD approaches. We believe theability to leverage existing data sources to createpersonas and scenarios expands the utility of UCDmethods. Because many existing data sources (e.g.,the World Values Survey or the World Internet Proj-ect) are readily available, we argue that this casestudy can be generalized to other design projects fordeveloping regions or for resource-constrained ordiverse communities. Additionally, the methods dis-cussed in this article might help user researchers anddesigners repurpose qualitative work from anthro-pologists, sociologists, and other researchers to cre-ate personas and scenarios.

BackgroundThis section discusses personas and scenarios andthe claims regarding their utility in the design pro-cess. In the next section, we brieºy describeKyrgyzstan. This is followed by a detailed discussionof the MoSoSo product concept that inspired thiscase study. Finally, we outline our research approachfor leveraging existing data to create Kyrgyz perso-nas and scenarios.

PersonasProponents claim that personas are effective fordescribing users and user requirements because theyincrease empathy, focus, and communication, andhelp design teams avoid stereotypes (Cooper, et al.,2007; Kuniavsky, 2003; Mulder & Yaar, 2007; Pruitt& Adlin, 2006). Each of these claims is subsequentlydiscussed in more detail.

Increased empathy. Grudin (2006) argues thatsome of the power of personas lies in our ability toempathize with ªctional characters much as we do

52 Information Technologies and International Development

ADAPTING USER-CENTERED DESIGN METHODS TO DESIGN FOR DIVERSE POPULATIONS

with those in movies and books. He speculates thatthe more a designer engages in the persona ªction,the more aligned his mental models will be withthat of the user and, therefore, the more alignedthe user interface and interaction will be with realuser goals and needs.

Tighter focus. Focus emancipates designers fromproblems that might arise when considering a fullspectrum of users. It allows designers to concentrateon the highest priority goals and needs of a speciªcuser or subset of users. In fact, Cooper et al. (2007)argue that it is better to design speciªcally for oneperson than vaguely for everyone.

Better communication. Proponents contend thatpersonas provide a communication channel for con-veying a wide range of quantitative and qualitativedata (Pruitt & Grudin, 2003). Clear communicationin turn facilitates consensus and efªciency in teamdecision-making (Mulder & Yaar, 2007). Making theassumptions about users explicit can also provide aclear benchmark to measure a design’s effectivenessand provide guidance for choosing participants inlater usability studies. In other words, personas thatmake the end user explicit can put all design teammembers on the same page.

Avoid stereotypes. Lacking user research, design-ers have only their assumptions and intuitions toguide their work. Goodwin argues that “the wholepoint in creating personas is to get past our per-sonal opinions and presuppositions” (Goodwin,2002). Personas, therefore, can replace the naïve as-sumptions about and stereotyping of end users bythe design team.

If personas achieve these beneªts, proponentsassert, design teams will have a useful image of theend user. Empathy will lead to the creation of aninteraction experience that is more aligned with theuser’s needs and goals. A clear focus will lead to aconsistent interaction experience with salient cues toguide the high-priority users. Good communicationwill lead to better decision-making among designteam members. Avoiding stereotypes forces design-ers to dismiss their assumptions and consider realuser goals and needs. If all these beneªts areachieved, an improved user experience is expected.This dynamic is precisely what can lead to moreeffective designs for diverse populations, includingin the information and communication technologyfor development (ICTD) space.

ScenariosQuesenberry (2006) suggests that stories are animportant augmentation to personas because storiesare effective at communicating culture and transmit-ting persona information into a memorable format.Whereas personas describe the individuals who willinteract with a product, scenarios in HCI describethe content and context of the interaction. Propo-nents suggest that scenarios in HCI are effectivebecause people are prewired to receive complexinformation through storytelling (Grudin, 2006).Advocates of scenarios stress the inclusion of typicaland signiªcant user activities that reºect user goalswhen interacting with a speciªc product or service(Carroll, 2000a; Go & Carroll, 2004).

Depending on the discipline and granularity ofdetail the scenario is designed to depict, the termscenario can have many meanings. Other disciplinesthat use scenarios include strategic planning forbusiness, requirements analysis and engineering,and object-oriented design (Go & Carroll, 2004). Instrategic planning for business, companies use“what if” scenarios to prompt decision-makingabout alternative approaches to possible eventsyears in the future (Kahn, 1962). HCI scenarios usedfor requirements analysis and engineering tell thestory of a computer interaction from a system’s per-spective and are concerned with moment-to-moment interactions at the task level. Seminal workin this ªeld includes the “Inquiry Cycle Model”(Potts, 1995). HCI scenarios used in object-orienteddesign approaches (often termed “use cases”)describe the possible path a user might takethrough a system focusing on the task level. Object-oriented design approaches using these types ofscenarios include “Responsibility-Driven Design” andautomated system modeling (Jacobson, 1995;Koskimies, Systä, Tuomi, & Männistö, 1998; Wirfs-Brock, 1993). The differences in these types of sce-nario use are (a) the perspective from which the sce-nario is written, and (b) how far into the future thereader is asked to imagine possibilities.

For this article, we created HCI scenarios thatCooper et al. (2007) refer to as “context” or as“a-day-in-the-life” scenarios in their Goal-DirectedMethod™ design approach. This type of scenario istold from the user (persona) perspective and focuseson a speciªc user goals while that user is interactingwith a product or service. Context scenarios are cre-ated at an early phase of development (compared toother HCI scenarios) and can be used to “explore, at

Volume 5, Number 4, Winter 2009 53

PUTNAM, ROSE, JOHNSON, KOLKO

a high level, how the product can best serve theneeds of the personas,” and “help create initial userrequirements before the speciªcs of the product areunderstood” (Cooper et al., 2007).2

In summation, personas and scenarios helpdesign team members think about the user’s per-spective. Whereas personas help design teamsunderstand who the users are, scenarios help themunderstand what users want to do.

KyrgyzstanKyrgyzstan is a post-Soviet country transitioning toa capitalist economy. The population, estimated at5.5 million, is concentrated in the capital city ofBishkek in the north and in the heavily agriculturalFerghana Valley in the south. The country’s popula-tion is very literate, with a literacy rate of 99%, andyoung, with a median age of 24.2 years (by compar-ison, the U.S. median age is 36.7). The ethnicity ofthe population is mostly Kyrgyz (65%), followed byUzbek (14%), and Russian (13%). Most Kyrgyz(about 75%) consider themselves Muslims (CIAWorld Factbook, 2008).

Face-to-face social networks are critically impor-tant in the region (Kuehnast & Dudwick, 2004).Social networks serve as avenues for gathering andsharing information, assistance, and goods. Theyalso offset the lack of reliable and trusted informa-tion from public institutions, such as the govern-

ment, police, and court system(Kolko, Johnson, & Rose, 2007).We have also found that technol-ogy is positively associated withan individual’s use and trust oftheir face-to-face social network(family, friends, and neighbors)for multiple types of informationseeking (Putnam & Kolko, 2009).

In addition, technology growthin Central Asia has not followedthe same pattern as in the West.As shown in Figure 1, Internetand computer use remain rela-tively low, whereas mobile phoneuse has increased at a rate ofabout 25% per year.3 Combinedwith the importance of social net-

works, rapid mobile phone adoption providesopportunities for designing appropriate technologiesthat leverage existing patterns. This goal inspiredour initial product concept of the MoSoSo directory.

MoSoSo Directory ConceptThe responses of people in Kyrgyzstan in focusgroups and interviews, as well as to surveys,prompted the idea for a mobile software servicethat would act like a Yellow Pages delivered viamobile phone and that would also leverage socialnetworks. When Kyrgyzstan was part of the SovietUnion (prior to 1991), information directories wereavailable, but no longer exist today. Some businessdirectories can be purchased in some bookstores,but these publications do not include listings ofeveryday services, and they are primarily geared forthe foreigner and NGO community. Moreover,although two free information lines operate in thecountry (similar to 411), respondents complainedthat it could take all day to reach an operator, and itis expensive to call the landline-based informationservice with a mobile phone. Throughout our work,the theme of scant public information resources hasbeen noted. Thus, a technological solution thatleverages the possibility of user-generated contentseemed a likely grassroots solution to the informa-tion scarcity problem. Speciªcally, we identiªed aneed for a recommendations tool/service to ªll a



2. In the approach of Cooper et al. (2007), context scenarios are followed by “key path scenarios” and “validation sce-narios,” both of which are analogous to the more detailed scenarios used in other HCI approaches.3. However, we found that increased mobile phone usage has not resulted in high levels of mobile Internet usage.

Figure 1. Technology trends in Kyrgyzstan, based on survey data collectedby the CAICT project.

niche, similar to what Internet sites like “Angie’sList” and online retailers like Amazon.com do in theUnited States. With low Internet penetration, wereasoned that Internet-based solutions would notattract the necessary volume of users to make adirectory based on user recommendations useful ortrustworthy. An appropriate solution to the informa-tion problem would, instead, leverage mobilephones.

In summation, three key factors inspired theshape of the service: (a) no standard phone directo-ries are available in the country; (b) Kyrgyz highlyvalue close social networks for information, assis-tance, and goods (Kuehnast & Dudwick, 2004); and(c) there is a strong upward trend in mobile phoneuse and ownership in Kyrgyzstan whereas thegrowth of computer and Internet use is relativelyºat. Ideally, after we identiªed the MoSoSo directoryas a tenable concept, we would then have con-ducted ªeld studies to elicit speciªc user require-ments and collected data speciªcally for personasand design scenarios. However, due to timing andeconomic constraints, we were unable to collectsuch data; instead, we leveraged data already avail-able to us.

Research for Personas and ScenariosThe creation of personas and scenarios commonlyuses face-to-face interviews and observation studiesto examine current and possible future uses of aspeciªc product or service (Cooper et al., 2007).When designing products for the developing worldor other diverse communities, however, these com-mon research methods are not always viable formany reasons: Budgets are limited, the ability toperform ªrsthand onsite research can be challengingwithout extensive local knowledge, and productcycle time frames can limit the feasibility of ªeldresearch. Each of these factors might prevent the in-depth study of a target market that would be typicalfor traditional approaches to persona and scenariocreation.

One solution to these challenges is to use datathat are otherwise available. Many countries, marketresearch ªrms, and academics conduct social surveywork and ethnographies. Global projects, such asthe World Values Survey or the World InternetProject, provide overviews of populations, theirattitudes, and patterns of media and technologyusage. Anthropologists produce ethnographiesabout societies around the world, and sociologists

conduct in-depth studies that designers couldrepurpose if they had a methodology to do so.

In this article, we present personas and scenariosof mobile phone users in Kyrgyzstan. They were cre-ated using data from two previous CAICT studies:(a) a 2007 survey of 1,000 respondents in Kyrgyz-stan, and (b) 16 interviews, with 12 participants.Neither study focused on user requirements or thecreation of personas and scenarios for a speciªcproduct or service. The survey was conducted toidentify attitudes and behaviors associated withtechnology use, and the focus groups and inter-views were conducted to identify how mobilephones support existing and new social networks.

Using statistical analyses of the survey data andinformation from the interviews, we created threepersonas, each with an accompanying context sce-nario to communicate user requirements for theproposed MoSoSo directory application. This casestudy demonstrates that researchers can useaccepted UCD methods (personas and scenarios) tohelp designers make appropriate technology designdecisions, even when they lack the resources to con-duct ªrsthand research on a given product or ser-vice. Our effort thus acknowledges the constraintsfacing many researchers and design teams who lackinstitutional support and/or budgets that allow forresearch studies focused on product design.

Data Collection Methods (ExistingData)As we did not have the luxury of conducting pri-mary research focused on the MoSoSo directory, weused two sets of previously collected data to createthe personas and scenarios. Neither data collectionwas focused on user requirements for a speciªcproduct or service. The characteristics of eachdataset are described below.

Survey DataThe ªrst set of data was from an April–May 2007large-scale social survey of 1,000 respondents, aged15 and older, that was administered in urban andrural areas in several regions in Kyrgyzstan. The sur-vey sample was based on government census infor-mation on age, gender, ethnicity, and geographiclocation. The sample included 50 sampling loca-tions, with 12–29 respondents interviewed in eachlocation. The survey instrument was designed by ateam of researchers from the University of Washing-



ton. The BRiF Research Group, located in Kazakh-stan, administered the survey. Households wereselected by a random walk procedure. Then eachhousehold respondent was chosen using the KishGrid method, a common technique to ensure a ran-dom selection of household members (Kish, 1949).

Focus Group and Interview DataThe second set of data used to inform the personasand scenarios was from focus groups and interviewsconducted by three University of Washingtonresearchers in two Kyrgyzstan sites: the capital cityof Bishkek and Kara Balta, a smaller suburban city.Researchers conducted two sessions in each loca-tion; each session involved a focus group and indi-vidual interview of three individuals, for a total of12 adult participants. This data identiªed mobilephone usage patterns that inform the maintenanceof social networks.

A Model for Using Existing DataThe following sections discuss how we modiªed tra-ditional user requirement research approaches forcreating personas and scenarios as they are dis-cussed in the literature. While there is no one recipefor creating personas and scenarios, the literatureagrees on three basic steps: (a) collect data aboutusers, (b) segment the users, and (c) create a per-sona for each user segment and develop contextscenarios for each persona (Cooper et al., 2007;Pruitt & Adlin, 2006). Context scenarios should illus-trate the persona interacting with the product andreºect a usage goal that persona has for the prod-uct or service (Cooper et al., 2007). As we wereusing existing data, the ªrst step was complete.

Because the data collection step was completed,the other two steps became more complicated andrequired modiªcation from common approaches dis-cussed in the literature. The next section comparesthe traditional methods for persona and scenariocreation with our approach. In so doing, we providea roadmap for others who might want to build user

requirements with limited data sources and presentthem via personas and scenarios.

Segment the Users (Step 2): What theLiterature SuggestsWhile marketing segments are typically deªned tohone a message for increasing sales aimed at theconsumer (buyer), segments for personas focus onthe person(s) who will ultimately use a product(user). Demographics, psychographics (psychologicalfactors that drive behaviors), and job roles are com-mon approaches for marketing segmentation(Barlow-Busch, 2006). These approaches are alsovalid in persona segmentation, but the literaturestresses focusing on differences in uses of the prod-uct for persona segments.

Mulder and Yaar (2007) argue that the bestdimension with which to segment users is accordingto the goals they have when using the product,whereas Cooper et al. (2007) focus more on atti-tudes and behaviors that end users have toward theyet-to-be-designed product. Cooper et al. (2007)also advocate the use of interviews for data collec-tion, using a clustering method to put each inter-view subject on spectrums of opposing attitudes orbehaviors concerning the speciªc product. Forexample, when segmenting users for a softwareproduct, they used attitudinal extremes of necessityversus entertainment, as shown in Figure 2. Inter-view subjects are placed along various attitudinaland behavioral spectrum extremes, and then clustersof users are identiªed.

Segment the Users: What We DidAs discussed, in two key areas, our data sourceswere much different than those typically used:(a) Rarely have other publications addressed a pro-cess that used data not collected in relationship to aspeciªc product, and (b) one dataset was from asurvey that led us to use several quantitative meth-ods. (Using quantitative methods in the creation ofpersonas is rarely discussed in the dominant litera-ture.4)



4. The few examples of publications that discuss quantitative persona segmentation include Chapman, Love, and Al-ford (2008) and Mulder and Yaar (2007).

Figure 2. Example of a Cooper et al. (2007) attitude spectrum in relation to a product.

Persona Segmentation Using Survey DataIn our survey results, 460 respondents ownedmobile phones. We segmented respondents via sur-vey questions that most closely pertained to goals,as suggested by Mulder and Yaar (2007). Thesequestions asked respondents whether they agreedor disagreed with 12 motivations (reasons) for whythey acquired a mobile phone. Respondents couldagree to multiple motivations, as shown in Table 1.This was our ªrst modiªcation to cited methods.Although research would typically inform goals inrelationship to a speciªc product, we focused onmotivations related to the technology that woulddeliver the service.

Using Phi correlations, we discovered signiªcantpositive relationships that broke into three logicalgroupings: (a) the motivation to replace a homephone; (b) practical motivations, including a need tomake calls and mobile phone affordability; and(c) social motivations, which included a desire toreceive calls and a need for a mobile phone becausefriends had them.

Final GroupingsWe next placed the respondents in one of threegroups based on their responses. We found that354 of the 460 respondents who owned mobilephones claimed motivations that fell into one of thethree ªnal motivation groups without overlap. Ourªnal three groups are discussed below.

Replacement group. In the ªnal segmentationmodel, there were 46 (13%) individuals in the re-placement motivation group. A large majority (84%)of this group reported not having a phone at home,7% said their home phone line was bad quality, and9% thought that home phones took too long to in-stall. Therefore, the primary driving replacement mo-tivation of this group was a lack of a home phone.

Practical group. There were 195 (55%) individualsin the practical motivation group. Almost all mem-bers of this group (99%) indicated a need to makecalls when away from home or work as the motiva-tion for acquiring a mobile phone. Some 2% werealso motivated by mobile phones being cheaper topurchase than landlines and another 2% by gettinga good price for the phone. The primary driving mo-tivation of this group was a desire to make outgoingcalls when they were away from landlines at homeor work.

Social group. There were 113 (32%) individuals inthe social motivation group. A majority (85%)wanted people to reach them at all times, 19% gottheir mobile phone because their friends had them,and 4% wanted to receive voice mail. Therefore, theprimary driving motivation of this group was a de-sire to receive incoming calls at all times.

Note that the primary motivation of each groupwas signiªcantly negatively associated with theother primary motivation statements. These negativeassociations conªrmed the logic of our initial per-sona segmentation groupings. In sum, we ªrstlooked for an appropriate survey question that cap-tured how users would use the MoSoSo directoryand that approximated user goals when using theproduct. After choosing “user motivation for acqui-sition of their mobile phone,” we determined ifthere were, indeed, distinct groups throughsigniªcant Phi correlations. Once satisªed that thesegroups were potentially different, we determinedthat the three groups differed in other areas asdetailed in the next sections.

Persona Group DifferencesWe analyzed the survey data through the lens of ourproposed three groups to look for important groupdifferences. Using multivariate analyses of variance(MANOVAs), we conducted omnibus tests on threeconstructs relevant to our personas: (a) demograph-ics; (b) mobile phone attitudes, behaviors, and expe-rience; and (c) other technology.

Construct One: Demographics.There were 10 variables in this construct, includingage, years of schooling, living in an urban versusrural location, household size, gender, maritalstatus, number of children, self-reported socio-economic status (SES), employment status, and stu-dent status.

Construct Two: Mobile phone attitudes,behaviors, and experience.There were 22 variables in this construct. Ten vari-ables asked respondents to rate several attitudesabout mobile phones on a scale of 1–4. Eight vari-ables asked respondents to report on possiblemobile phone usage behaviors. One variablefocused on the respondent’s actual mobile phoneusage patterns. Another variable explored howrespondents acquired their ªrst phones,5 and stillanother asked respondents how much they would



5. This variable was dummy coded (0, 1): Received as a gift (the most common answer) was coded as 1.



Tabl

e1.

PhiC

orre

latio

nsfo

rM

otiv

atio

nsto

Acq

uire

aM

obile

Phon

e.

n1

.2

.3

.4

.5

.6

.7

.8

.9

.1

0.

11

.1

2.

1.

Id

on

ot

hav

ea

ho

me

ph

on

e4

60

—

2.

My

ho

me

ph

on

eis

bad

qu

alit

y4

60

�.0

8—

3.

Itta

kes

too

lon

gto

get

ah

om

ep

ho

ne

46

0�

.20

**

�.1

9*

*—

4.

Tom

ake

calls

wh

enaw

ayfr

om

ho

me

46

0�

.25

**

�.0

4�

.06

—

5.

Itis

chea

per

than

ah

om

ep

ho

ne

46

0�

.03

�.0

3�

.03

�.0

6—

6.

Ig

ot

ag

oo

dp

rice

46

0�

.04

�.0

4�

.04

�.1

9*

*�

.36

**

—

7.

My

frie

nd

sal

lh

ave

mo

bile

ph

on

es4

60

�.0

2*

�.0

1�

.03

�.0

2�

.35

**

�.3

2*

*—

8.

Ilik

eto

rece

ive

voic

emai

l4

60

�.0

2�

.19

**

�.0

2�

.06

�.0

3�

.09

�.2

6*

*—

9.

Ilik

ep

eop

leto

reac

hm

eat

all

tim

es4

60

�.1

4*

*�

.03

�.0

2�

.32

**

�.0

9�

.17

**

�.1

3*

*�

.05

—

10

.Ea

sier

tom

ake

inte

rnat

ion

alca

lls4

60

�.0

1�

.03

�.0

5�

.19

**

�.0

4�

.05

�.0

4�

.03

�.0

9*

—

11

.I

nee

dit

for

wo

rk4

60

�.2

6�

.01

�.1

0�

.01

�.0

1�

.02

�.0

3�

.01

�.0

5�

.01

—

12

.C

on

ven

ien

ce4

60

�.0

8�

.03

�.0

3�

.03

�.0

4�

.05

�.0

4�

.04

�.0

4�

.04

—

No

tes:

Posi

tive

sig

niª

can

tco

rrel

atio

ns

are

inb

old

face

.St

atem

ents

are

reo

rder

edh

ere

toem

ph

asiz

eg

rou

pin

gs;

the

last

thre

est

atem

ents

wer

en

ot

sig

niª

can

tly

po

siti

vely

asso

ciat

edw

ith

any

of

the

thre

eg

rou

pin

gs,

soth

eyw

ere

om

itte

dfr

om

eval

uat

ion

.

miss their phones if they no longer had them.6 Thelast variable in this construct examined the fre-quency of mobile phone use.7

Construct Three: Other TechnologyThere were seven variables in this construct, includ-ing computer ownership, computer and Internetuse, frequency of computer and Internet use, land-line ownership, and cable or satellite TV availabilityin the home.

Several steps were taken to assure assumptionsfor MANOVA tests were met. Homogeneity of vari-ance/covariance was tested through Box’s Test ofEquality (Tabachnick & Fidell, 2007). Whereas theªrst two constructs met the Box test requirements,Construct 3 did not and therefore could not be ana-lyzed through a MANOVA. Consequently, items forConstruct 3 were analyzed using univariate ANOVAand Chi-square tests, with a Bonferroni adjustedalpha.

Construct One: DemographicsThe groups differed on the best linear combinationof the 10 demographic outcomes, Wilks’� � 0.868,F (20, 684) � 2.52, p � 0.001, �2 � 0.069, meaningthat approximately 7% of the variance is accountedfor by demographic group differences. Follow-up,pair-wise comparisons, using a Bonferroni adjust-ment, showed that groups differed signiªcantly onthree of the 10 variables in the construct: years ofschooling, living in an urban versus rural location,and household size. In Figure 3, we also show ªnd-ings that were important to the persona descrip-tions, but non-signiªcant in the MANOVA model,

including average age, employment status, andstudent status.

It is important to note that although we areusing inferential statistics to analyze the quantitativedata (and are careful to control type-one error inºa-tion for each construct), signiªcance for a variable isa fairly high bar to clear. An alpha of 0.05 meansthat group differences need to be so extreme thatthey would only happen by chance in fewer than5% of cases. Additionally, using a Bonferroni adjust-ment means that, in the univariate follow-up test,the 5% alpha is distributed among all the variablesin the construct. Whereas this type of rigor is impor-tant in evaluating experimental data, personas areultimately descriptive tools. Consequently, we feelthat descriptive data describing group differencesthat do not reach the bar of statistical signiªcanceshould, nevertheless, be included in the personas toincrease the overall breadth of their descriptions;especially data that follow clear, but non-signiªcanttrends. For example, see Employment Status inFigure 3.

Construct Two: Mobile Phone Attitudes,Behaviors, and ExperiencesThe groups differed on the best linear combinationof the 22 mobile use outcomes, Wilks’� � 0.665,F (44, 378) � 2.95, p � 0.05, �2 � 0.185, meaningthat approximately 19% of the variance isaccounted for by mobile phone attitudes, behavior,and experience group differences. The variables inthe mobile phone construct accounted for almostthree times the variance in group differences when



6. Most (52%) said they would miss their phone “a lot,” skewing the data; therefore, the variable was dummy codedso that “a lot” was coded as 1 and all other answers coded as 0.7. Most (73%) used their phone several times a day, which introduced skewed data; therefore, the variable wasdummy coded so that “several times a day” was coded as 1 and all other answers were coded as 0.

Figure 3. Construct One: Notable Differences in Demographics.

** � signiªcant differences found in the MANOVA omnibus test

compared to the demographic construct and werean important validation of the segmentation model.In other words, the groups differed more on theirmobile phone attitudes and usage than they differedon demographics. Follow-up pair-wise comparisons,using a Bonferroni adjustment, showed that groupsdiffered signiªcantly on 5 of the 22 variables in Con-struct Two, as shown in Figure 4.

Construct Three: Other TechnologyWe created the technology construct because it iscommonly advised to include computer, Internet,and other technology usage in persona develop-ment, if the product is technologically based(Mulder & Yaar, 2007). Items for the last constructwere analyzed using univariate ANOVA and Chi-square tests, with a Bonferroni-adjusted alpha foreach variable in the construct (0.05/7 variables �

adjusted alpha of 0.007). Computer ownership, hav-ing a landline phone, and cable/satellite TV were allsigniªcantly different among the three personagroups. Figure 5 shows computer and Internet usedetails that were not signiªcantly different, but thatwere incorporated in the ªnal personas. Many ofthese details show interesting trends among thethree user groups, such as Internet use and lengthof Internet use.

In sum, the segmentation model provided a lensthrough which we evaluated the multiple variablesin the survey and provided an interesting depictionof mobile users in Kyrgyzstan. Our steps were to:(a) ªnd a meaningful segmentation model thatwould reºect how users would utilize the MoSoSodirectory, namely their motivation (goal) to acquire amobile phone; and (b) investigate whether our initial



Figure 4. Construct Two: Notable Differences on Attitudes, Behaviors, and Experience.

** � signiªcant differences found in the MANOVA omnibus test

Figure 5. Construct Three: Notable Differences in Other Technology Use.

** � signiªcant differences found in Chi-square tests

segments demonstrated signiªcant differences inthree constructs that were important for personacreation that might affect how, when, or if individ-ual respondents would use the MoSoSo directory.To encapsulate this data into a set of memorablepersonas, we needed more than simple statisticaldifferences. We needed realistic back stories andscenarios based on people’s experiences that wouldillustrate user requirements. To create these stories,we turned to focus group and interview data.

Persona Segmentation of the InterviewParticipantsWe segmented the 12 interview participants intothe three established user groups, based on theirdescriptions of: (a) telephone use, both land linesand mobile phone, and (b) how they used theirmobile phone and other technology in their lives.See Table 2 for interview identiªcations and descrip-tions. In the next section, we describe how theinterview data were integrated and used for theªnal persona and scenario creation.

Create the Personas and Scenarios(Step 3): What the Literature SuggestsThe literature suggests a few basic guidelines onpersona content. The guidelines include the use of anon-posed photo (Mulder & Yaar, 2007; Pruitt &Adlin, 2006), a personal name beginning with theªrst letter of the segmentation group name, a bio-graphical proªle, and personal information thataffects usage patterns in relation to the product(Pruitt & Adlin, 2006). Additionally, it is commonlyadvised to include computer, Internet, and othertechnology usage, if the product is technologicallybased (Mulder & Yaar, 2007). Finally, user goals forthe product are considered critical (Cooper et al.,2007).

Personas are presented through many mediums,including posters and 81

2” � 11” sheets of paper(Pruitt & Adlin, 2006). Others have experimentedwith more interesting mediums, such as employingaction ªgures (Nieters, Ivaturi, & Ahmed, 2007), cre-ating living spaces in which the design team inter-acts in the same types of physical spaces as the



Table 2. Interview Participants by Motivation Group.

Interview ID Location Participants Roles Primary motivation

KG_UF1 Bishkek Urban Family Father(KG_UF1_F1)

Practical

Youngest son(KG_UF1_S2)

Social

KG_UY1 Bishkek Urban Youth Female Friend 1(KG_UY1_F1)

Social

Female Friend 2(KG_UY1_F2)

Social

Male Friend 1(KG_UY1_F3)

Social

KG_RF1 Kara Balta Rural Family Father(KG_RF1_1)

Practical

Mother(KG_RF1_2)

Did not own a mobile phone

Youngerbrother(KG_RF1_3)

Practical/Social

KG_RY1 Kara Balta Rural Youth Male Friend 1(KG_RY1_1)

Social

Male Friend 2(KG_RY1_2)

Practical/Social

Female Friend 1(KG_RY1_3)

Social

personas (Goodwin, 2006), and placing a summarypersona on reference cards (Pruitt & Adlin, 2006).Pruitt and Grudin (2003) suggest creating a “foun-dation document” that explicitly links the ªnal per-sonas to the supporting data that emphasize theirconnection to the underlying empirical research.

Recall that context scenarios are intended as asimulation of the persona interacting with the prod-uct that focuses on usage patterns and personagoals in a speciªc context. Carroll (2000b) suggeststhat scenarios need three basic elements: (a) the set-ting where the action takes place, (b) one or moreactors [i.e., personas], and (c) a plot that details asequence of events and actions. Scenario represen-tations can be presented as storyboards, prototypes,videos, or in a sequenced narrative (Cooper et al.,2007).

Kyrgyz Case Study: What We DidWe followed the guidelines above as closely as pos-sible to create our personas and scenarios. We usedphotos from interview participants and gave ourpersonas Kyrgyz names, with the ªrst letter match-ing the ªrst letter of the segment group. The cultur-ally appropriate names also clearly identify thepersonas as representatives of a non-Western audi-ence. We used our interview data for the biographi-cal and personal content and the survey data toidentify technology usage patterns. Identifying usergoals, however, was more difªcult. Because wewere unable to travel to Kyrgyzstan to re-interviewparticipants about their goals for a MoSoSo service,we inferred goals from our existing data.

To make these inferences, we examined respon-dents’ answers to focus group and interview ques-tions. One question, in particular, asked participantsto describe a recent task that was hard to complete.The responses provided data that elaborated onchallenges participants had experienced completingeveryday tasks. On the basis of these stories, weimagined how the MoSoSo directory could havebeen part of a solution to their problems. For ourªnal scenarios, we also included details from individ-ual participant interviews in which participantsdescribed their experiences from the previous day.

We designed the persona presentation for an81

2 � 11” sheet of paper. Additionally, we created adetail sheet for each persona that tied each datapoint for the persona to the speciªc data fromwhich it was extracted. This detail sheet was a

simpliªed stand-in for the “foundation document”discussed above (Pruitt & Grudin, 2003). We presentour context scenarios here as numbered sequences(Cooper et al., 2007).

Findings: User Requirements, Personas,and ScenariosOur resulting three personas were: Parxat, the prac-tical user, Shirin, the social user, and Roza, the re-placement user. Each of the one-page personapresentations included: (a) a photo that looks like areal person, not a model (Mulder & Yaar, 2007;Pruitt & Adlin, 2006); (b) key differentiators (Mulder& Yarr, 2007), (c) personal information formatted asa bullet list (Mulder & Yarr, 2007), (d) computer andInternet usage that characterizes the technical abili-ties of each persona and illustrates signiªcantbehavior patterns (Mulder & Yarr, 2007; Cooper &Reimann, 2003), (e) a personal proªle written inprose (Mulder & Yarr, 2007; Pruitt & Adlin, 2006),and (f) user goals for the product that emphasizeend goals (Mulder & Yarr, 2007; Pruitt & Adlin,2006). We also encapsulated several critical userrequirements for the MoSoSo directory in the con-text scenarios.

User Requirements Reºected in the ContextScenariosWe identiªed six key user requirements for theMoSoSo directory that allowed individuals to:(a) add/edit business entries in a public directory viaSMS/text, (b) create and join private group directo-ries that are shared by family or friends to supportsocial networks via SMS/text, (c) broadcast (push) orpost (pull) information within their private sharedgroup directories via SMS/text, (d) contribute andrate services in both the public directory and in pri-vate shared directories via SMS/text, (e) retrieve rec-ommendations or information from the publicdirectory and from shared private directories viaSMS/text, and (f) retrieve information from the ser-vice without using SMS/text. Each scenario high-lights speciªc user requirements as suggested in theliterature (Cooper et al., 2007).

Practical ParxatParxat’s background story of being a computergame club owner is largely based on that of inter-view participant KG_RY1_2 who, at the time of theinterview, was a 25-year-old computer game clubowner. Since the persona segment represented 55%of the survey respondents, we identiªed him as the



primary persona whose goals and needs are themost important for the MoSoSo directory to meet.See Figures 6 and 7.

We identiªed three deªning characteristics forthe Parxat persona: (a) he uses his phone for workcalls, reºecting the fact that 41% of practical usersin the survey data use their phone for work, a ratehigher than that of any other group; (2) he pur-chased his phone, reºecting the practical group’sgreater likelihood—more so than our other twogroups—to have bought their phones; and (c) he ismore tech savvy than the other two personas, basedon the higher likelihood of his owning a computer.We made Parxat a little bit older than the mean agefor the practical group (35.9 years) to emphasize thedifference with the social group. All other personaldata were based on survey data.

Parxat’s technical information and mobile phoneuse was mostly based on survey ªndings. We found,however, that our interview participants were morelikely to use text messaging than the 27% rate ofuse extrapolated from the survey. Since the MoSoSoservice needs some text interfacing, we exaggeratedParxat’s texting ability to align it with the reportedtext use of interview participant (KG_UF1_S1),whom we had situated in the practical segment.

The quotes attributed to Parxat’s persona arelifted directly from the interview participants andemphasize his strong reliance on friends. Parxat’sgoals for the MoSoSo directory, as a small businessowner, are inferred from the interviews.

Parxat’s Context ScenarioFor Parxat’s context scenario, we focused on twospeciªc user requirements: (a) add business entries,and (b) contribute and rate services in both the pub-

lic directory and in private shared directories viaSMS/text. See Table 3.

Social ShirinThe context scenario for the social user came frommale participant KG_UF1_S2, who conveyed hisexperience in a student group where he functionedas both a member and organizer. Shirin was alsoidentiªed as a primary persona because her segmentrepresents 32% of the survey respondents. See Fig-ures 8 and 9.

We identiªed three key characteristics for Shirin:(a) she uses her phone to primarily call friends,emphasizing the fact that 93% of social groupmembers use their phones for social calls; (b) she isthe least likely to feel that mobile access is tooexpensive, reºecting the fact that only 50% of thesocial group felt that mobile phone use was tooexpensive (signiªcantly less than the other twogroups); and (c) she is somewhat tech savvy,reºecting our ªnding that the social group had thesecond highest level of experience with computersand the Internet. We made Shirin a little youngerthan the mean age for the social group (33.6 years)to emphasize the group’s difference from the practi-cal group. All other personal data were based onsurvey ªndings.

Like Parxat, we also exaggerated Shirin’s textingenthusiasm and based her texting rate on interviewparticipant KG_RY1_1, whom we had placed in thesocial segment. The quotes in Shirin’s persona aredirectly from interview participants who we hadlocated in the social group and focus on socialaspects of her life. Shirin’s goals for the MoSoSodirectory, like those of Parxat, are inferred from theinterviews.



Table 3. Parxat’s Primary Context Scenario.

1 When Parxat arrives at his small computer club in the morning, he sees a ºyer advertising the MoSoSo di-rectory. The ºyer explains that as a small business owner he can advertise his shop in the public informa-tion space where users can vote to recommend shops and services.

2 Parxat calls the service and discovers that listing a service requires using text.

3 Navigating the service through a phone tree system, he decides to locate his shop under the public infor-mation space heading of “Computer cafe/club.” The entry system allows Parxat to set up two types of ad-vertisements, one for users who use text and one for users who do not use text.

4 Later, Parxat asks one of his clients, Ilzat, to call the service and submit a positive vote. Ilzat already has afriends group that he belongs to in the MoSoSo directory and offers to not only give Parxat’s shop a goodvote in the public information space, but to also add it as a recommended computer club in his group’s in-formation space.



Figure 6. Parxat Persona.



Figure 7. Parxat Persona Data Detail.



Figure 8. Shirin Persona.



Figure 9. Shirin Persona Data Detail.

Shirin’s Context ScenarioFor Shirin’s context scenario, we focused on twouser requirements: (a) create and join private sharedgroup directories to support social networks viaSMS/text, and (b) broadcast (push) or post (pull)information within their private shared group direc-tories via SMS/text. See Table 4.

Replacement RozaRoza’s biographical background was derived primar-ily from an interview with suburban family members(KG_RF1). The father’s brother lived in a villagewhere the one landline was housed in a communitybuilding that closed at 5:00 P.M. See Figures 10 and11. Because Roza’s persona segment only represents13% of survey respondents, we identiªed her as asecondary persona whose goals and needs are lessof a driving force in the design of the MoSoSo direc-tory than those of the other two personas.

We identiªed three key characteristics for Roza:(a) she is the least likely to use her phone for work;(2) she lives in a rural area, reºecting survey results

that indicate 82% of the replacement group lived ina rural area; and (3) she is not tech savvy, whichreºects our ªnding that the replacement group wasthe least tech savvy. Roza’s remaining personal datawere based on survey ªndings for her segment.

As with the previous two participants, Roza’stechnical information and mobile phone use werebased on survey ªndings. Since none of our inter-view participants ªt in the replacement group,Roza’s quotes are from interview participantKG_UF1_F1, who emphasized the importance ofsocial networks (rather than technologies) whenªnding services.

Roza’s Context ScenarioRoza’s context scenario came from urban fatherparticipant KG_UF1_F1, who described his recentdifªculty in ªnding a mechanic. Roza’s scenariofocuses on one speciªc user requirement: allowusers to retrieve information from the service with-out using SMS/text.



Table 4. Shirin’s Primary Context Scenario.

1 As a cofounder of an unregistered student organization, Shirin is always trying to ªnd ways to recruit newmembers on campus and let existing members know about upcoming events and meetings.

2 After hearing about the MoSoSo directory, she decides to set up a group for her student organization.

3 She also wants to set up the group so users can get notices in several different ways. They can (1) call infor new information, (2) ask for a text alert that there is new information and then call in, or (3) sign upfor text broadcasts.

4 A few days later, once the group has been set up, she sends out a message about an upcoming meetingfor group members.

Table 5. Roza’s Primary Context Scenario.

1 When Roza’s husband, Ermek, left for work this morning, he discovered that their car would not start. Hecalled a co-worker who luckily was able to pick up Ermek on his way to the local farm where they bothwork. Ermek asked Roza to ªnd a mechanic to ªx the car as soon as possible.

2 Trusting her social network for answers, Roza began to call friends and family to see if anyone knew of agood mechanic with expertise ªxing their make and model of car.

3 Her sister-in-law suggested Roza call the general information line. This suggestion frustrated Roza becauseit can take several hours to get through the information line and then she will have no idea if the me-chanic will be good.

4 Rather than calling the information line right away, Roza calls a neighbor who tells her about the MoSoSodirectory and gives Roza a password for the neighborhood group.

5 Since Roza does not use text messaging, Roza is pleased to ªnd she can interact using only her keypad byanswering questions that guide her to a list of locally recommended mechanics.

6 After navigating her way through the phone tree, Roza ªnds a mechanic that the neighborhood grouprecommended.



Figure 10. Roza Persona



Figure 11. Roza Persona Data Detail.

Each of these context scenarios illustrates particu-lar user requirements of the MoSoSo directory,based on real stories about difªculties users encoun-tered in their lives, the importance of social net-works, and how the MoSoSo service might supportthose existing networks and solve a problem. Thepersonas amalgamate multiple sources of data toprovide an idea of who the users are, and the sce-narios describe a particular interaction with the pro-posed service. Together, the personas and scenariosroll up complex data from multiple sources intocommunication conduits that design teams can useto understand end users.

DiscussionThe case study presented here discusses how weused existing qualitative and quantitative data tocreate personas and scenarios that identify userrequirements and communicate user needs. Thedata were not collected in relation to a speciªcproduct or service. However, after the product ideaof a MoSoSo directory was generated, we revisitedthe existing data to help create UCD deliverables.We began by segmenting the audience through sta-tistical analysis of survey questions and identiªedthree groups that were differentiated by their moti-vation to acquire a mobile phone: practical, social,and replacement. We then analyzed the survey datafor additional differences among the three groups.The differences reºected in the ªnal personas werebased on three constructs: (a) demographics,(b) attitudes about mobile phones and mobilephone usage, and (c) other technology experience.Whereas quantitative datasets were useful for usersegmentation, the richer qualitative data wereneeded to create personal proªles and scenarios.

Once each interview participant was placed inthe appropriate segment, the interview informationwas used to ªll in biographical stories. The resultingpersona information was amalgamated into a singlesheet for each persona to provide a memorableimage of end users. The scenarios presented herewere based on real-life stories in response to inter-view questions about difªculties the participants hadencountered. For each context scenario, we inferredhow the MoSoSo directory could help the partici-pants navigate those difªculties, and we focused onspeciªc user requirements.

Because design research for diverse populations,

including developing regions, is expensive anddemands signiªcant resources and expertise, weargue that using existing data sources can reducecosts while still representing the user needs of thesepopulations. Designing for diverse populations iscrucial for both development and economic reasons.Whereas on-the-ground studies are irreplaceable asdata sources, other approaches are needed if designis to address increasingly diverse users. Userresearchers and designers can make signiªcantheadway toward creating appropriate designs byextracting user requirements from existing datasources.

Additionally, we demonstrate that personas andscenarios make compelling communication summa-tions of user research that help deªne user require-ments. Because there is little information about howto create personas and scenarios using existing data,the methods presented here will help researchers(a) create effective conduits of user research (perso-nas and scenarios) that capture and communicateuser requirements to design teams, and (b) allow abroader range of designers, including those inworkplaces that lack ªnancial resources and/or cul-tural expertise, to make use of such approaches todevelop creative and appropriate technologies fordiverse users.

Future WorkWe plan to use the work presented here in two verydifferent studies. The ªrst is part of an investigationin which we hope to identify the variables thataffect how useful, actionable, and usable personasand scenarios are from the perspective of designteam members. In part of that study, several profes-sional design team members will be asked to inter-act with the personas and scenarios presented herein an experimental lab setting. Additionally, they willbe asked to reºect on previous use (if relevant) ofpersonas and scenarios and whether interactingwith personas/scenarios that represent a diverseaudience has an effect on their utility. We hope toidentify which claims made of personas and scenar-ios in the literature are met, and what circumstancesaffect their usefulness.

We are also in the process of developing a sec-ond version of the MoSoSo prototype for usabilitytesting. Version one was created by design teammembers who used the personas and scenarios pre-



sented here. In late March 2009, we conductedeight usability tests in Bishkek, Kyrgyzstan, using arough prototype of version one created in AdobeFlash. This early prototype lacked voice functionality;all interaction was through text with a simulatedphone displayed on a computer. Our initial resultspresented rich data with respect to the functionalityas well as the user interface. The basic concept wasenthusiastically received by six of our eight partici-pants. Our future work will include continual itera-tion and testing of the MoSoSo directory prototypesto see if our development process that relied sig-niªcantly on existing data did indeed generate ausable and useful product. ■

AcknowledgmentsThis material is based on work supported by theNational Science Foundation (NSF) under grants#0326101 and #0219350. Any opinions, ªndings,and conclusions or recommendations expressed inthis material are those of the author(s) and do notnecessarily reºect the views of the NSF. Additionalthanks to our survey respondents, interview partici-pants, and local research assistants.

ReferencesBarlow-Busch, B. (2006). Marketing versus design

personas. In J. Pruitt & T. Adlin (Eds.), The Per-sona Lifecycle (pp. 602–641). San Francisco:Elsevier.

Carroll, J. M. (2000a). Five reasons for scenario-based design. Interacting with Computers, 13,43–60.

Carroll, J. M. (2000b). Making use: Scenario-baseddesign of human computer interactions. Cam-bridge, MA: MIT Press.

Chapman, C. N., Love, E., & Alford, J. L. (2008, Jan-uary 7–10). Quantitative early-phase user re-search methods: Hard data for initial productdesign. Paper presented at the Proceedings ofthe 41st Hawaii International Conference on Sys-tem Sciences, Hawaii, 37–48.

Chapman, C. N., & Milham, R. P. (2006, October16–20). The persona’s new clothes: Methodolog-ical and practical arguments against a popularmethod. Paper presented at the Proceedings of

the Human Factors and Ergonomics Society 50thAnnual Meeting, Santa Monica, CA, 634 –636.

Central Intelligence Agency (CIA). (2008). CIA WorldFactbook. Retrieved September 22, 2008, fromhttps://www.cia.gov/library/publications/the-world-factbook/

Cooper, A., Reimann, R., & Cronin, D. (2007). Aboutface 3: The essentials of interaction design. India-napolis, IN: Wiley Publishing.

Go, K., & Carroll, J. M. (2004). The blind men andthe elephant: Views of scenario-based system de-sign. interactions, 11(6), 45–53.

Goodwin, K. (2002). Getting from research to perso-nas: Harnessing the power of data. RetrievedApril 23, 2008, from http://www.cooper.com/journal/2002/11/getting_from_research_to_perso.html

Goodwin, K. (2006). Taking personas too far.Retrieved April 28, 2008, from http://www.cooper.com/journal/2006/12/taking_personas_too_far.html

Grudin, J. (2006). Why personas work: The psycho-logical evidence. In J. Pruitt & T. Adlin (Eds.), ThePersona Lifecycle (pp. 642–663). San Francisco:Elsevier.

Jacobson, I. (1995). The use-case construct in ob-ject-oriented software engineering. In J. M.Carroll (Ed.), Scenario-based design: Envisioningwork and technology in system development,(pp. 309–336). New York, NY: John Wiley &Sons, Inc.

Kahn, H. (1962). Thinking about the unthinkable.New York: Horizon Press.

Kam, M., Bhagwani, S., Kumar, A., Lal, S., Mathur,A., Tewari, A., et al. (2007, December 15–16).The social complexities of user-centered design inICTD: Experiences from four schools in India’s vil-lages and slums. Paper presented at the Proceed-ings of the 2nd IEEE/ACM InternationalConference on Information and CommunicationTechnologies and Development, Bangalore, India,294–299.

Kish, L. (1949). The procedure for objective respon-dent selection within the household. Journal ofAmerican Statistical Association, 44, 380–387.



Kolko, B., Johnson, E., & Rose, E. (2007). Mobile so-cial software for the developing world. In D.Schuler (Ed.), Online Communities and SocialComputing (pp. 385–394). Berlin/Heidelberg:Springer.

Koskimies, K., Systä, T., Tuomi, J., & Männistö, T.(1998). Automated support for modeling OOsoftware. IEEE Software, 15(1), 87–94.

Kuehnast, K., & Dudwick, N. (2004). Better a hun-dred friends than a hundred rubles? Social net-works in transition—The Kyrgyz Republic.Washington DC:World Bank Working Papers.

Kuniavsky, M. (2003). Observing the user experi-ence: A practitioner’s guide to user research(pp. 129–159). San Francisco: Morgan KaufmanPublishers.

Maunder, A., Marsden, G., Gruijters, D., & Blake, E.(2007). Designing interactive systems for the de-veloping world: Reºections on user-centred de-sign. Paper presented at the Proceedings of the2nd IEEE/ACM International Conference on Infor-mation and Communication Technologies andDevelopment, Bangalore, India, 321–328.

Mulder, S., & Yaar, Z. (2007). The user is alwaysright: A practical guide to creating and using per-sonas for the Web. Berkeley, CA: New Riders.

Nieters, J. E., Ivaturi, S., & Ahmed, I. (2007, April28–May 3). Making personas memorable. Paperpresented at the Proceedings of CHI ’07, SanJose, CA, 1817–1823.

Portigal, S. (2008). True tales: Persona non grata. in-teractions, 15(1), 72–73.

Potts, C. (1995). Using schematic scenarios to un-derstand user needs. Paper presented at the DIS’95, Ann Arbor, MI.

Pruitt, J., & Adlin, T. (2006). The Persona Lifecycle.San Fransisco, CA: Elsevier.

Pruitt, J., & Grudin, J. (2003, June 6–7). Personas:Practice and theory. Paper presented at the Pro-ceedings of the 2003 Conference on DesigningFor User Experiences, DUX ’03, San Francisco,CA, 1–15.

Putnam, C., & Kolko, B. (2009, July 20–22). Gettingonline but still living ofºine: The complex rela-tionship of technology adoption and in-personsocial networks. Paper presented at the Proceed-ings of Advances in Social Networks and Mining,ASONAM ’09, Athens, 33–40.

Quesenbery, W. (2006). Storytelling and narrative. InJ. Pruitt & T. Adlin (Eds.), The Persona Lifecycle(pp. 520–555). San Fransisco, CA: Elsevier.

Rosson, M. B., & Carroll, J. M. (2003). Scenario-based design. In J. A. Jacko & A. Sears (Eds.), Thehuman-computer interaction handbook: Funda-mentals, evolving technologies and emerging ap-plications (pp. 1032–1050). Mahwah, NJ:Lawrence Erlbaum Associates.

Tabachnick, B. G., & Fidell, L. S. (2007). UsingMultivariate Statistics (5th ed.). Boston: PearsonEducation.

Wirfs-Brock, R. J. (1993). Designing scenarios:Making the case for a use case framework. Re-printed from the Nov/Dec 1993 issue of TheSmalltalk Report 3(3).



adapting user-centered design methods to design for ......user-centered design (ucd) methods such as...

Documents