a design science primer
DESCRIPTION
An introduction to design science, in particular for IT and information systemsTRANSCRIPT
A Design Science Primer
Paul Johannesson
Erik Perjons
Published under the
Creative Commons Attribution‐NonCommercial‐ShareAlike 3.0
Unported License
A Design Science Primer, 1st edition
by Paul Johannesson and Erik Perjons
First published 2012
ISBN‐13: 978‐1477593943
ISBN‐10: 1477593942
Printed by CreateSpace
Creative Commons Attribution‐NonCommercial‐ShareAlike 3.0
Unported License:
Users are free to use, copy, share, distribute, display, and reference this book
under the following conditions:
ATTRIBUTION: Whole or partial use of this book should be attributed (refer‐
enced or cited) according to standard academic practices.
NON‐COMMERCIAL: This book may not be used for commercial purposes.
SHARE ALIKE: Users may alter, transform, or build upon this book, but must
distribute the resulting work under the same or similar license as this one.
For any reuse or distribution, the license terms of this work must be clearly
specified.
Your fair use and other rights are in no way affected by the above.
Copyright © 2012 by Paul Johannesson and Erik Perjons
i
Preface
Thisbookisanintroductiontodesignscience.Itisintendedtosup‐portbothresearchersandstudents instructuring,undertakingandpresentingdesign sciencework.The level of text is basic anddoesnotpresumeanypriorknowledgeofdesignscience.
Chapter1providesanoverviewofdesignscienceandoutlinesitsrelationshipswithempiricalresearch.Chapter2discusses thevari‐oustypesandformsofknowledgethatcanbeusedandproducedbydesignscience.Chapter3givesabriefoverviewofcommonempiri‐calresearchstrategiesandmethods.Chapter4introducesamethodthataimsatsupportingresearchers indoingdesignscienceaswellas presenting design science results. The method consists of fiveactivities,whicharedescribed indetail inChapters5 to9.Chapter10discusseshowtocommunicatedesignscienceresults.Chapter11compares the proposed design science method with methods forsystemsdevelopmentandshowshowtheycanbecombined.Finally,Chapter 12 discusses how design science relates to research para‐digms, inparticularpositivismand interpretivism.There is also anappendix that suggests adesign science structure forbachelor andmastertheses.Themaintextisinterspersedwithanumberofboxesthatreflectonvariousthemesindesignscience;initiallythereadermaywishtoskiptheseandrefertothemlateronsubsequentread‐ing.
ii
Thebookoffersnovelinstrumentsforvisualizingdesignscienceresults,both in the formofprocessdiagramsandthroughacanvasformat.
The authors would like to thank Birger Andersson, Ilia Bider,Shengnan Han, Martin Henkel and Benkt Wangler for discussionsandconstructivefeedbackonearlierversionsofthistext.
Thepdfversionof the text isdesigned foreye‐friendlyviewingon tablets aswell as eco‐friendly printing onpaper, forwhich it issuggestedtoprinttwopagespersheet.Filesfordownloadandotherdocumentationcanbefoundat:http://designscienceprimer.wordpress.com/
Paul Johannesson, [email protected]
Erik Perjons, [email protected]
iii
Table of Contents
1 Introduction ....................................................................................... 1
2 Knowledge Types and Forms ........................................................... 12
3 Research Strategies and Methods ................................................... 25
4 A Design Science Method................................................................. 33
5 Explicate Problem ............................................................................ 51
6 Outline Artefact and Define Requirements .................................... 63
7 Design and Develop Artefact ........................................................... 77
8 Demonstrate Artefact ...................................................................... 85
9 Evaluate Artefact .............................................................................. 89
10 Communicate Artefact Knowledge .............................................. 100
11 The Design Science Method and Systems Development ............ 105
12 Research Paradigms ..................................................................... 115
References ........................................................................................ 126
Appendix ........................................................................................... 131
1
1 Introduction
Empiricalresearchaimsatdescribing,explainingandpredictingtheworld.Forexample,theLinnaeantaxonomydescribesandclassifiesthe kingdoms of animals and plants into classes, orders, families,genera,andspecies.Newton’slawsareabletoexplainthemotionofplanets, the trajectoriesofmissilesandthereasons for tides.Mete‐orologypredictsrainfall,stormsandotherweatherphenomena.Thegoalofempiricalresearch,atleastinthenaturalsciences,istofaith‐fullydescribeandexplaintheworldasitexistsregardlessofhumaninterestsandbiases.Theworldisoutthere,andcanbeexplainedbysciencesothatpeoplehaveacommonunderstandingofit,irrespec‐tiveoftheirbackgrounds,traditionsandvalues.
Incontrasttoempiricalresearch,designresearchisnotcontentonly to describe, explain and predict. It also wants to change theworld,toimproveitandtocreatenewworlds.Designresearchdoesthis by developing artefacts that can help people fulfil their needs,overcome their problems, and graspnewopportunities. In this en‐deavour, design research not only creates novel artefacts but alsoknowledgeaboutthem,theiruse,andtheirenvironment.
Thisbookisaboutaspecialstrandofdesignresearch,calledde‐sign science,which has its origins in the areas of information sys‐temsandIT.Designscienceintheseareasaimstocreateinnovationintheformofideas,models,methodsandsystemsthatsupportpeo‐ple indeveloping,usingandmaintainingITsolutions.Thisbookfo‐cuses on design science as applied to information systems and IT,
2
butitalsoincludesexamplesfromandoutlooksoverotherfieldsofhumanpractice.Asdesign science aims to create artefacts that ad‐dressproblemsexperiencedbypeople, the restof this chapterwillintroduceandrelate thenotionsofpeople,practices,problemsandartefacts.
1.1 People, Practices and Problems
Apractice isasetofhumanactivitiesperformedregularlyandseenasmeaningfullyrelatedtoeachotherbythepeopleparticipatinginthem.Anexampleisthepracticeofdentists,whoengageincleaningteeth,drillingteeth,pullingoutteeth,takingX‐rays,andmanyotheractivities.Whenpeopleengage inpractices, theywill typicallyneedtohandlenaturalaswellasman‐madeobjects.Forexample,dentistsanddentalnurseswillrepairteethandmakeuseofpliers,drills,X‐raymachines, etc.Another exampleof apractice is cooking,wherepeople cut fruit, frymeat,boil vegetables,marinate fish, andsoon,while using stoves, refrigerators, pans, graters, and other kitchenutensils.
Practices can be more or less structured or formalized. Somepracticestakeplacewithinorganisations,e.g.theproductionofcarsin factoriesor themanagementof customercomplaints in call cen‐ters. Other practices occur in informal settings, for example, kidsplayingballsinabackyardorpeoplehavingdinnertogether.Thereare also practices in which people can engage as individuals, e.g.brushingtheirteethortyingtheirshoelaces.
Whenpeopleengageinpractices,theymayexperienceproblems,practical problems. A practical problem is an undesirable state ofaffairs,ormoreprecisely,agapbetweenthecurrentstateandade‐sirablestate,asperceivedbytheparticipants inapractice.Thede‐sirable state is seen as better than the current one, because itwillallowpeopletobemoresuccessfulwhenengaginginthepractice.Anexampleofapracticalprobleminthepracticeofdentistryisthat,forsomepeople,theirdentalfillingsmayfalloutaftersometime.Thereisa current state inwhichsomedental fillings tend to fallout, and
3
there is amore desirable state inwhich dental fillings always stayput.Thepracticalproblemperceivedbydentistsisthegapbetweenthesetwostates.
Aproblemisnotalwaysanobstacletoovercome,butcanalsobeapuzzlingquestionor anunexpected circumstance that couldpro‐videanopportunity for improvement.Thus, thereare twokindsofproblems. First, there are problems in which the current state isviewedas trulyunsatisfyingand thedesirablestate isseenasneu‐tral,e.g.,havinga toothacheora flat tire.Secondly, thereareprob‐lems where the current state is seen as neutral and the desirablestateisregardedasapotentiallyhugeandsurprisingimprovement.Oftensuchproblemsarenotperceiveduntilsomeinnovationarisesand captures people’s imagination, and they realise that their cur‐rentpracticecanbeimproved.AnexamplecouldbetheinventionofX‐rays,which gave doctors themeans to overcome the problemofnotbeingable toview the insideof thehumanbody.Summarising,theterm“problem”isusedheretodenotetroublesomesituationsaswellaspromisingopportunities.(Itcouldbearguedthat“challenge”and “practical challenge” would be better choices of word than“problem” and “practical problem”, but this book sticks with thewell‐establishedlatterterms).
1.2 Artefacts as Solutions to Problems in Practices
Whenfacingapracticalproblem,peoplemayreactto it indifferentways.Oneoptionistoadoptastoicattitudeandjustaccepttheprob‐lemasafactoflifewithouttryingtodoanythingaboutit.Theotherextremewouldbetoviewtheproblemasbeingsoseriousthattheentirepracticeshouldbeabandoned,oratleastpartofit.Anexam‐plecouldbebloodletting,whichwasanestablishedpracticeinmedi‐cine forcenturiesbutceasedwhenevidencemountedregarding itsadverse health effects. However, the most common reaction to apracticalproblemistotrytofindsome,oftenpartial,solutiontoit.
Inmanycases,practicalproblemscanbesolvedbymeansofar‐tefacts.Anartefactisdefinedhereasanobjectmadebyhumanswith
4
the intention to be used for addressing a practical problem. Someartefacts are physical objects, like hammers, cars and hip replace‐ments.Otherartefacts take the formofdrawingsorblueprints, e.g.anarchitect’splanforabuilding.Methodsandguidelinescanalsobeartefacts,e.g.amethodfordesigningdatabases.Commontoalltheseartefactsisthattheysupportpeoplewhentheyencounterproblemsinsomepractice.
ThereexistsaplethoraofartefactsintheITandinformationsys‐temsarea,rangingfromalgorithms, logicprogramsandformalsys‐tems over software architectures, information models and designguidelinestodemonstrators,prototypesandproductionsystems.IntheearlyyearsofIT,mostartefactsweredevelopedformilitaryandbusinesspractices.However,inrecenttimessomeofthemostinno‐vative IT artefacts have been designed for everyday practices, likekeeping in touch with friends, sharing and organising photos, orplayinggames.
Therelationshipsbetweenpeople,practices,problemsandarte‐facts are summarized in figure 1.1. People engage in practices inwhichtheymayperceiveproblemsthatcanbeaddressedbymeansof artefacts.Thus, artefactsdonot exist in isolationbut are alwaysembeddedinalargercontext.
Figure 1.1 People, Practices, Problems and Artefacts
5
1.3 The Anatomy of Artefacts
Everyartefacthasaninside,anoutside,andaninterfacebetweentheinsideandoutside.Moreprecisely,everyartefacthasan innercon‐struction,issituatedinanenvironment,andoffersafunctionforitsintendedpractice.Theintendedpracticeisheredefinedastheprac‐ticethatholdsthepracticalproblem,whichtheartefactistoaddress.
The constructionof an artefact is about its innerworkings, thecomponents it consists of, how theseare related, andhow they in‐teractwitheachother.Anartefactisalwaysconstructedfromsmall‐erpartsthatareassembledinsuchawaythattheycaninteractwitheach other and produce some behaviour. An example could be aclockconstructedfromcogwheels,watch‐hands,andothermechani‐calparts.Anotherexampleisatruck,whichismadeofachassis,anengine,wheelsandotherparts.
Theenvironment of an artefact is about the external surround‐ingsandconditionsinwhichitwilloperate.Theenvironmentofanartefact always includes its intendedpractice aswell as thepeopleand other objects participating in that practice. The environmentmayalso includeotherpractices thatareaffectedby theuseof theartefact.Furthermore,theenvironmentmaycontainvariousobjectsthatarenotrelatedtoanyspecificpractice.Asanexample,theenvi‐ronment of a truck includes the goods transportation practice inwhich it is used. If the truck passes through areaswhere kids areplaying, thepracticeof childrenplayingalsobecomesapart of thetruck’senvironment.Finally, theenvironmentcontainsthephysicalsurroundingsofthetruck,includingstreetsandair.
Thefunctionofanartefactistheintendedresultofusingitinitsintendedpractice,i.e.thosebenefitstheartefactisexpectedtobringtoitsusers.Forexample,thefunctionofaclockistotellthetime,thefunctionofalawn‐moveristocutgrass,andthefunctionofatruckistomovegoods.
Whenanartefactisusedinapractice,itwillhavecertaineffectsonitsenvironment, i.e. itwillchangeit inintendedaswellasunin‐tendedways.Theintendedchangescorrespondtothefunctionofthe
6
artefact. For example, the intended effect of using a truck is thatsomegoodsaremovedfromoneplacetoanother,whichisthesameas the function of the truck. Using an artefactmay also have unin‐tendedeffects,oftencalledsideeffects.Theseeffectsmayconcernnotonly the intended practice of the artefact but also other practices,sometimes with adverse consequences for them. For example, atruckpassingthroughanareawherechildrenplaymayposesuchasafety hazard that the play has to stop. Side effects may also beharmful for other valuable resources even if these are not directlyused in any specific practice. Emissions from truck driving pollutetheair,whichindirectlymayharmmanypractices.
Figure 1.2 illustrates howan artefact is situated in an environ‐ment, which may include several practices. The artefact offers itsfunctiontotheintendedpractice,butitmayhavesideeffectsforthisaswell as other practices. Thus, an artefactmay havemany stake‐holders,i.e.peopleinpracticesthatareaffectedbyit.
Figure 1.2 Construction, Function and Environment
7
A common guiding principle in artefact design is to hide the con‐structionofanartefactfromitsfutureusersandinsteadfocusonitsfunction.Ausershouldnotneedtocareabouttheinternalstructureof the artefact but only about its function, i.e. how it can serve theuser.Ideally,theusershouldnotevenbeawareoftheconstruction.An example is a clock, which someone can use without knowingwhetherit isconstructedusingmechanicalpartsorelectroniccom‐ponents. In the history of IT, the idea of hiding the internals of anartefacthasbeenappliedrepeatedlywith labels likeencapsulation,object‐orientation,informationhiding,andserviceorientedarchitec‐tures.
Whendesigninganartefact,adesigneroftenstartsbycreatingaspecification that defines the functional requirements for the arte‐fact,i.e.whichfunctionstheartefactshouldoffer.Tworequirementsfor awatch are that it should be usable as a stopwatch and as analarmclock.Typically,requirementsaregatheredfromandvalidat‐edbypeoplewithintheintendedpractice.Therequirementscanbeexpressedasalistoffunctionsoftheartefactwithnoreferencetoitsconstruction.Instead,theconstructioncanbedevelopedlaterwhenthe designer has a more complete understanding of the require‐ments. However, it is also common that function and constructionareelaboratedinaniterativeway.
Thedistinctionbetweenconstruction,function,andenvironmentis sometimes reflected in the professional roles of designers. Forexample,inthehouseconstructionindustry,aconstructionengineerwill focuson the internalstructureofbuildings, includingselectionofbuildingmaterials, layoutofplumbing, strengthcalculations,etc.Anarchitect,on theotherhand,will focuson theenvironmentandfunctionofbuildingsinordertocaterforexternalconstraintsaswellastheneedsandrequirementsoftheusers.Similarly, intheITandinformationsystemsindustry,enterprisearchitectsaddressbusinessrequirementsaswellaslegal,culturalandotherenvironmentalfac‐tors,whileprogrammers and software engineers focuson the con‐structionofsoftwarewithinthesystemstobebuilt.
8
1.4 Design Science – the Study of Artefacts
Artefacts are studied indifferent fields of science, including formalsciences, behavioural sciences and social sciences. For example, astudy in theoretical computerscience (formal science)coulddeter‐minethecomplexitypropertiesofanewalgorithmfor traversingasocial graph.A study inpsychology (behavioural science) could in‐vestigate how photo sharing on social networks influences stresslevels.Astudy inbusinessadministration(socialscience)couldex‐aminehow theadoptionofERPsystems in companies affects theirinternalcommunication.
Artefactsarealsostudiedwithindesignscience,wheretheyareinvestigated as solutions to practical problems that people experi‐enceinpractices.Moreprecisely:
Design science is the scientific study and creation of arte‐
facts as they are developed and used by people with the
goal of solving practical problems of general interest
The starting point for a design researcher is that something isnotquiterightwiththeworld,andithastobechanged.Anewarte‐factshouldbeintroducedintotheworldtomakeitdifferent,tomakeit better. The design researcher does not only think and theorizeabout theexistingworld.Shemodels,makesandbuilds inorder tocreate new worlds. She produces both a novel artefact andknowledgeaboutitanditseffectsontheenvironment.Inparticular,sheneeds to formulateproblemstatements,determinestakeholdergoals and requirements, and evaluate proposed artefacts. In otherwords, artefacts aswell as contextual knowledge are researchout‐comesfordesignscience.
Inthisbook,designsciencewillbeviewedmainlyfromanITandinformation systems perspective. However, the principles underly‐ingdesignscienceareapplicabletomanyotherareas,seetheboxonmedicalsciencebelow.
9
Medical Science and Design Science
Encyclopaedia Britannica defines medicine as “the practice concerned with the maintenance of health and the prevention, alleviation, or cure of disease”. Early medical practices incorporated plants, animal parts and minerals as instruments for healing. They were often used in magi‐cal rituals overseen by priests or shamans. Medicine thereby became closely related to spiritual systems like animism, shamanism, spiritual‐ism and divination. Today, these relationships have largely been broken, and medical practices are instead supported by medical science.
Medical science is in many ways akin to design science. There is a practice, the medical practice that aims at healing people. There are practical problems that have to do with the effectiveness, safety and cost of engaging in this practice. There are artefacts that address practi‐cal problems and support the practice, such as pharmaceutical drugs, medical devices and therapies. A large part of medical science is devot‐ed to studying, in a scientific way, how such artefacts can help solve practical problems in medical practice. Thus, many of the notions and principles behind design science are also relevant for medical science.
1.5 Design and Design Science
Designsciencemayappearhighlysimilartodesign,asbothfocusonthe development of artefacts. Both of themalso aim at novelty, i.e.theyaretoproduceorinvestigateoriginalartefactsthatdifferfromexistingones.However,theirpurposesaredifferentwithrespecttogeneralizabilityandknowledgecontribution.Whiledesign isapro‐cess for developing a working solution to a problem that may berelevantonlyforasingleactor,designscienceaimsatproducingandcommunicating knowledge that is of general interest. Results fromdesignwork are sometimes relevant only for a local practice, i.e. apractice in which just one single individual, organization or groupengages. Incontrast,designscienceshouldproduceresultsthatarerelevantforaglobalpractice,i.e.acommunityoflocalpractices.
Thedifferentpurposesofdesignanddesignsciencegiverisetothree additional requirements on design science work. Firstly, the
10
purposeofcreatingnewknowledgeofgeneralinterestrequiresde‐signscienceprojectstomakeuseofrigorousresearchmethods.Sec‐ondly,theknowledgeproducedhastoberelatedtoanalreadyexist‐ing knowledge base in order to ensure that proposed results arebothwell founded and original. Thirdly, the new results should becommunicatedtobothpractitionersandresearchers.
As an example of the specific requirements on design science,consideraprojectfordesigninganewelectronichealthrecordsys‐tem. In order to count as design science, the project first has tochoose an overall research strategy for investigating the problemsituation and eliciting stakeholder requirements. This strategywillincluderesearchmethodsfordatageneration,e.g.questionnairesforlarge groupsofhealth careprofessionals anddeep interviewswithselectedphysiciansandhealthcaremanagers.Thestrategywillalsoincludemethodsforanalysingthegenerateddata.Furthermore,theproject has to relate the produced results to existing knowledgewithin various subareas of health informatics and information sys‐tems.Thisknowledgewillincludenotonlyestablishedtheoriesandmodels but also relevant artefacts, in particular other electronichealthrecordsystems.Onlybyrelatingtheprojectresultstoexistingknowledge, does it become possible to assess their originality andvalidity.Theprojectmustalsoevaluatetheartefactproducedusingadequateresearchstrategiesandmethods.Finally,theprojecthastodisseminateitsresultstobothhealthcareprofessionalsandacadem‐ics through publications in journals and conferences as well aspresentationsathealthcarefairs,professionalconferences,andoth‐ersimilarevents.
Must an artefact in design science aim at changing the real world?
Short answer: Yes
Long answer: In design science, an artefact is an object made by humans with the intention to be used for addressing a practical problem. In oth‐er words, some stakeholders must have formulated a goal of the arte‐
11
fact, and this goal has to be related to a practical problem. The stake‐holders want to address, or ideally solve, the problem by using the arte‐fact. The stakeholders are not disinterested observers who desire im‐proved knowledge for its own sake, but people engaged in a practice that has encountered a problem. They want to employ the artefact in order to change the world so that the problem is solved or at least miti‐gated. Thus, theories or descriptive models do not count as design sci‐ence artefacts, as they aim at explanation and description but not change and problem solving.
Further Reading
Oneoftheearliestandmostinfluentialtextsontherelationshipbe‐tweendesignandscienceisTheSciencesoftheArtificialbyHerbertSimon(Simon,1996).Anearlypaperondesignsciencewaswrittenby March and Smith (1995). Another highly influential paper waswrittenbyHevneretal.(2004).Österleetal.(2010)discusstheim‐portance and relevance of design science. Vaishnavi and Kuechler(2004)havedesigned aweb site fordesign science that includes alarge bibliography. The notion of practice is discussed in depth byCetinaetal.(2001).PracticeresearchanditsrelationshipstodesignscienceareinvestigatedbyGoldkuhl(2012).
12
2 Knowledge Types and Forms
Scientificresearchproducesdifferentkindsofknowledgeandtherehave been numerous attempts to classify these; one of the earliestbeingKant’sdistinctionsbetweenapriorivs.aposterioriandanalyticvs.syntheticknowledge.Inthischapter,knowledgeisclassifiedintosix types depending on its contents and uses. The classificationbuildsonabroadviewofknowledgeasexpressed in the followingdefinitionfromWikipedia;“Knowledgeisafamiliaritywithsomeoneor something, which can includefacts,information,descriptions,orskillsacquiredthroughexperienceoreducation.Itcanrefertothetheoreticalorpracticalunderstandingofasubject.Itcanbeimplicit(aswithpracticalskillorexpertise)orexplicit(aswiththetheoreti‐calunderstandingofasubject);itcanbemoreorlessformalorsys‐tematic”(Wikipediacontributors,“Knowledge”,2012).
Anotheraspectofknowledgeistheforminwhichitcanbemate‐rialized. This is particularly relevant for design science, as theknowledgeitproducesisnotalwaysexplicitbutsometimesembed‐dedinartefacts.AfterhavingintroducedknowledgetypesinSection2.1andknowledgeformsinSection2.2,thechapterconcludeswithaclassificationofartefacttypes.
13
2.1 KnowledgeTypes
Definitional Knowledge
Definitional knowledge consists of concepts, constructs, terminolo‐gies, definitions, vocabularies, classifications, taxonomies andotherkindsofconceptualknowledge.Itmaybeformalandprecise,suchasthebasicconceptsofrelationaldatabasetheory,whichincludesrela‐tions,attributes,functionaldependencies,andmulti‐valueddepend‐encies.Definitionalknowledgemayalsoincludevagueandinformalconcepts like theHCInotionsofusability, affordance, andsituated‐ness.Definitionalknowledgedoesnotincludestatementsaboutreal‐ity that are claimed to be true. Instead, it is used as a basis for allothertypesofknowledgeinthesensethatitprovidesbasicconceptsrequired toexpress thatknowledge. J.G.Bennett (1985)succinctlyphrasedthisas;“Wedonotknowstructures,butweknowbecauseofstructures”.
Examplesofdefinitionalknowledge:
Arighttriangleisatriangleinwhichoneangleisarightan‐gle
A unicorn is amythical creature depicted as awhite horsewithalarge,pointed,spiralinghornprojectingfromitsfore‐head
Ablackholeisaregionofspacefromwhichnothingcanes‐cape
A relation in relational database theory is in third normalform if all of its attributes are dependent on the key, thewholekeyandnothingbutthekey
Thesestatementsonlydefinecertainconceptsbutdonotclaimthatright triangles, unicorns, black holes, or third normal forms reallyexist;suchclaimsaremadebydescriptiveknowledge.
14
Descriptive Knowledge
Descriptive knowledge describes and analyses an existing or pastreality.Thistypeofknowledgetypicallydescribes,summarises,gen‐eralises and classifies observations of phenomena or events. Forexample,descriptiveknowledgecanclaimthatamajorityoftheERPusersinacompanyaredissatisfied.Inaddition,itmaydescribeindi‐vidual entities and events, for example, that the company EricssonhasinvestedinanewglobalERPsystem.Itmayalsodescriberela‐tionshipsamongentities,therebyprovidingananalysisofthese,forexample, themodularcompositionofanERPsystem. Incontrast todefinitional knowledge, descriptive knowledge does include state‐ments that are claimed to be true. In other words, descriptiveknowledgeclaimstodescribetheworld“asis”.
Examplesofdescriptiveknowledge:
TheheightoftheEiffeltoweris300meters Allswansarewhite AlllargecompaniesinEuropeuseERPsystems
Descriptive knowledge does not claim to provide explanations orpredictions;itonlydescribes.
Explanatory Knowledge
Explanatory knowledge provides answers to how and why ques‐tions,explaininghowobjectsbehaveandwhyeventsoccur.Explana‐tory knowledge goes beyond descriptive knowledge, as it not onlydescribes and analyses but also explains in order to offer under‐standing. Explanations often take the form of cause‐effect chains,showinghoweventsandoutcomesarecausallyrelatedtounderlyingmechanisms and factors. For example, the increased acceptance ofInternetbankingamongthepubliccanpartiallybeexplainedbyim‐provementsinsecurityinfrastructures.
High‐level explanatoryknowledgeaims to showhow theworldcanbeviewed ina certainway, sometimeswith the intent to chal‐lenge conventional assumptions andbring about an altered under‐
15
standingbasedonnoveltheoreticalperspectives.Sometheoriesthathave been used for this purpose are structuration theory, action‐networktheory,andcriticaltheory.Forexample,criticaltheorycanbeusedtoexplainwhyinformationsystemsintroducedforimprov‐ing productivity, often evolve into instruments formonitoring andcontrol.
Explanatory knowledge on a lower level provides explanationsforparticularsituations,e.g.throughcasestudies,whereitexplainswhyandhowspecificeventsunfold.Forexample,theintroductionofarewardsysteminanorganizationcanbeamajorcausefortheem‐ployees’acceptanceofanewERPsystem.
Examplesofexplanatoryknowledge:
Humans are descended from other animals through a pro‐cessofnaturalevolution
The earthquake in Japan 2011 was a consequence of dis‐placementsoftectonicplates
The failure to introduce social software into the companywascausedbyitsauthoritarianorganisationalculture
Explanatory knowledge only explains events after they have hap‐penedandcannotbeusedforpredictingfutureevents.
Predictive Knowledge
Predictive knowledge offers black‐box predictions, i.e. it predictsoutcomesbasedonunderlyingfactorsbutwithoutexplainingcausalor other relationships between them. The goal of predictiveknowledgeisaccurateprediction,notunderstanding.
High‐level predictive knowledgewith a broad generality is notcommonintheITandinformationsystemsarea;oneofthefewex‐amples is theCOCOMOmodel, (Boehm,1981),whichprovides costestimation for software development. Just as for explanatoryknowledge, predictive knowledge can also exist on a lower levelwhereitpredictsoutcomesinspecificsituations.
16
Examplesofpredictiveknowledge:
Physicianswithunwashedhandstreatingpatientsmayresultindiseases
Heavyexposuretosunlightmayresultinskindiseases
These examples only predict certain outcomes and do not explainthem.Today thereare explanations for thesepredictionsbut thesewerenotknownwhenthisknowledgewasdiscovered.
Explanatory and Predictive Knowledge
Explanatoryandpredictiveknowledgeoffersbothexplanationsandpredictions.Itpredictsoutcomesandexplainshowthesearerelated,often through causal relationships, to underlying mechanisms andfactors. This kindof knowledge is considered themost common innaturalscience.ItismuchlesscommonintheITarea,butthereareexamples such as the Technology Acceptance Model (TAM), (Ven‐kateshandDavis,2000),andDeLoneandMcLean’smodelof infor‐mation success, (DeLone andMcLean, 2003). Explanatory andpre‐dictive knowledge can also be incorporated intomore comprehen‐sive theories, like general systems theory and the soft systems ap‐proach(Checkland,1999).
Examplesofexplanatoryandpredictiveknowledge:
Microorganisms can survive in dirt and infect people thatcomeintocontactwiththem
Exposuretotheultravioletradiationofsunlightmaycauseaburnoftheskin,whichmayresultinskindiseases
Thesestatementsbothpredictpossibleoutcomesandexplainthembydescribingtheunderlyingmechanismscausingthem.
Prescriptive Knowledge
Prescriptiveknowledgeconsistsofmethodsandprescriptivemodelsthathelpsolvepracticalproblems.Prescriptivemodelscanbeseenasblueprintsfordevelopingartefacts,whilemethodsareguidelinesandprocedures thathelppeople towork in systematicwayswhen
17
solvingproblems.Methods often, but not always, prescribe how toconstructartefacts.
Methods and prescriptivemodels can be viewed as comprisingtwoparts.Thefirstisthemodelormethoditself,andthesecondisastatementaboutsomedesirableeffectofusingthemodelormethod.Thisstatementisnotasubjectivejudgementthatsomeeffectisde‐sirableinanabsoluteway.Instead,itisapredictivestatementimply‐ingthatifamethodorprescriptivemodelisusedinacertainprac‐tice, thiswillcontribute toeffectsdesiredbysomestakeholders. Inthissense,prescriptioncanbeseenasaspecialcaseofprediction.
Typical examples of methods are systems and software devel‐opmentmethodslikeRUP(KrollandKruchten,2003),oragilemeth‐odslikeXPandSCRUM(Cohn,2009).Examplesofprescriptivemod‐elsareconceptualreferencemodelslikeSCOR(BolstorffandRosen‐baum,2007),orarchitecturalmodelslikeOSI(DayandZimmerman,1983).
Examplesofprescriptiveknowledge:
Washyourhandsbeforegoingtosurgery! Applysunlotionbeforesunbathing!
Is there a difference between models in social science and design science?
Short answer: Yes
Long answer: Social science produces models that describe and predict phenomena and behaviour in a domain, i.e. descriptive and predictive models. In contrast, design science produces models that can be used as blueprints for developing artefacts, i.e. prescriptive models.
2.2 Knowledge Forms
While knowledge types describe how knowledge can be used,knowledge forms specify how it can be materialized, i.e. where itexists and inwhich shape. Knowledge forms are important for de‐sign sciencework, as it createsnot only knowledge explicitly codi‐
18
fiedindocumentsbutalsoembeddedinartefacts.Thefollowingsub‐sections will introduce the three main knowledge forms and theirrolesindesignscience.
Explicit Knowledge
Explicitknowledge is articulated, expressedand recorded inmediasuchastext,numbers,codes,formulas,musicalnotations,andvideotracks.Typicalcontainersofexplicitknowledgeareencyclopaedias,textbooks, and manuals. A main strength of explicit knowledge isthatitcanbeeasilytransferredbetweenindividuals,asitisstoredinmediathatcanbemovedanddecodedinaconvenientway.Howev‐er,notallknowledgecanbearticulatedinanexplicitway,e.g.practi‐cal skills, intuitions and experiences. Reading amanual on how torideabicyclewillnothelpmuchinlearningthatskill.
Embodied Knowledge
Embodiedknowledgeissituatedinthemindsofpeopleandistypi‐callydifficulttoformulateinanexplicitway.Forexample,althoughalmosteveryone is able to read facial expressions correctlyandef‐fortlessly,noonewouldbeabletoarticulatefullyhowthis isdone.Otherexamplesofembodiedknowledgearetyingashoelace,speak‐ingEnglish,orusingavideorecorder.
Sometimes,effortsaremadetotransformembodiedknowledgeinto explicit knowledge, for example, by companies who want toretaintheknowledgeofemployeeswhoaregoingtoquit.Thegoalofsuch a transformation is that people without expert backgroundshould be able to replicate the performance of thosewho possessrelevant, embodied knowledge. However, the person usingknowledge in an explicit formwill typically lack the ability of theskilledpractitioner to innovateandadapt tonewcircumstances. Inthissense,theextracted,explicitknowledgeisinferiortotheembod‐iedknowledge.
19
Embedded Knowledge
Embedded knowledge resides not in humans but in entities, likephysical objects, processes, routines, or structures. As an example,consider a bread maker, i.e. a home appliance for making bread.When loadedwith the right ingredients, such amachine is able toproduce a loaf of bread bymixing those ingredients, kneading theresultingdough,andthenbakingit;someonestudyinghowthema‐chineworkscanlearnaboutbreadmaking.Thus,themachinecanbeseen as embedding the prescriptive knowledge of how to makebread; it embeds a recipe, amethod. Furthermore, by studying theconstructionofthemachine,someonecanlearnaboutthearchitec‐tureofbreadmakingmachinesingeneral.Itmayevenbepossibletocreatea(prescriptive)modelforbuildingnewbreadmakers.Inthisway,themachinealsoembedsamodel.
Ingeneral,anobjectcanembedbothmodelsandmethods. Justas for embodied knowledge, such embedded models and methodscansometimesbemadeexplicit.However, this requiresa transfor‐mationthattypicallylosessomeoftheembeddedknowledge,i.e.theextractedmodelsormethodsarepoorerthantheonesembeddedintheobject.(Theprocessofdiscoveringtheprinciplesbehindanarte‐factthroughananalysisofitsfunctionalityandconstructioniscom‐monlyknownasreverseengineering,whichhasalongtraditionwith‐inmilitaryaswellasindustrialpractices.)
2.3 Knowledge in Craft and Science
Intraditionalcrafts,representingandtransferringdesignknowledgeismostcommonlydonebyembedding itwithinobjects.This isnotdueto lazinessof thedesigner,butbecauseobjectsareable toem‐beddesignknowledgewithall its richnessandnuance.Anappren‐ticecarefullystudyinganartefactdesignedbyhermasterwillgainadeepandintuitiveunderstandingoftheprinciplesusedincreatingit.Inthisway,thedesignknowledgeembodiedinthemasterbecomesembodied in the apprentice, using the artefact with its embeddedknowledgeasanintermediary.
20
Representingandtransferringknowledgebyembeddingitinar‐tefacts can offer substantial benefits with respect to richness andintuition, but there are also several disadvantages. First, acquiringknowledge by directly studying artefacts can be time‐consuming.Thisisevidencedbythemanyyearsittypicallytakesacraftappren‐ticetobecomeamaster.Secondly,theknowledgesomeoneactuallyacquiresbystudyinganartefactishighlydependentonthatperson’sbackground,interestsandcapabilities.Somepeoplemaylearneffec‐tively from examining an artefact, while others may fail to do so.Thirdly, it is difficult to analyse, criticize and evaluate knowledgeembeddedinanobject.Analysis,criticismandevaluationareusuallymoreeffectiveandmeaningfulwhentheknowledgeisexplicit.
Oneofthemainpurposesofdesignscienceistosupportdesign‐ers and researchers in making design knowledge explicit, therebymovingdesign fromcraft toscience.Designscienceoffers thissup‐port by clarifyingprinciplesbehinddesign andbyprovidingmeth‐odsforcreatingartefactsaswellasexplicitknowledgeabout them.However,itshouldbeacknowledgedthatnotalltheknowledgeem‐bedded inanartefactcanbemadeexplicitand, therefore, thearte‐factitselfisavitalresultofanydesignscienceproject.AsstatedbyRobinMathew,“Designiswherescienceandartbreakeven”.
2.4 Artefact Types
Basedontheknowledgetypesandformsintroduced,itispossibletodistinguish between different artefact types.Within the IT area, ithasbecomecommontoidentifyfoursuchtypes:constructs,models,methods,andinstantiations.
Constructsareterms,notations,definitions,andconceptsthatareneeded for formulatingproblemsand theirpossiblesolutions.Con‐structsdonotmakeanystatementsabouttheworld,buttheymakeit possible to speak about it, so it canbe understood and changed.Thus,constructsaredefinitionalknowledge.Someexamplesaretheconceptsof class inUML,method in Java, functionaldependency inrelationaldatabasetheory,andaffordanceinHCI.Constructsarethe
21
smallestconceptualatomswithwhichtounderstandandcommuni‐cateaboutvariousphenomena.
Modelsareusedtodepictorrepresentotherobjects.Amodelcanrepresentanexistingsituation,whichcanbeusedfordescribingandanalyzingproblemsituations.Suchadescriptivemodelmayworkasapedagogicaltoolforrepresentingacurrentsituationandexplainingwhyitischallenging.However,modelscanalsobeusedtodescribepotential solutions to practical problems, e.g. a drawing for a newtypeofvehicleoraproposalforanarchitectureofamobileoperat‐ingsystem.Suchprescriptivemodelsworkasdescriptionsofpossiblefuture solutions andhelp tobuild artefacts that can solvepracticalproblems.Therearealsopredictivemodelsthatcanbeusedtofore‐cast thebehaviorofobjectsandsystems.Thus,modelscanexpressdescriptiveaswellaspredictiveandprescriptiveknowledge.Inde‐sign science, the focus is on prescriptivemodels. Typical examplesarebusinessprocessmodels,systemsarchitectures,domainontolo‐gies,andusermodels.
Is a model a theory?
Short answer: No
Long answer: A theory aims to explain by showing how phenomena and behaviour in a domain are related to each other, typically through cause and effect relationships. In contrast, a model aims at representation, as it represents all or part of a system. It is possible to distinguish between three kinds of models: descriptive, predictive, and prescriptive (or nor‐mative) models. A descriptive model describes some existing system but without offering any explanations. A predictive model allows forecasts of events, e.g. weather forecasts. A prescriptive model, e.g. a drawing for a new building, can be used for guiding actions, thereby helping people to fulfil their goals.
Caveat: Even if models are not theories in themselves, descriptive and predictive models can still be a part of theories. For example, a concep‐tual model in the context of agency theory could include and relate con‐structs like principal, agent, risk, goal, contract, and monitoring (Eisen‐
22
hardt, 1989). Such a model provides a conceptual basis for agency theo‐ry but does not count as a theory in itself. To do so, it has to be com‐plemented with explanatory propositions like “As the principal and the agent have different goals, the agent does not always act in the interest of the principal” or predictive statements like “Monitoring may improve the efficacy of behaviour‐based contracts”. Simplifying, theory = de‐scriptive model + explanations + predictions.
Methods express prescriptive knowledge by defining guidelines
andprocesses forhowtosolveproblemsandachievegoals. Inpar‐ticular, they canprescribehow to create artefacts.Methods canbehighlyformalizedlikealgorithms,buttheycanalsobeinformalsuchasrulesofthumborbestpractices.Someexamplesaremethodsfordatabasedesign,changemanagementinitiatives,orwebservicede‐velopment.
Instantiationsareworkingsystemsthatcanbeusedinapractice.Instantiationscanalwaysembedknowledge,e.g.adatabasecanem‐bed a databasemodel. Some examples of instantiations are a Javaprogramrealisingasearchalgorithm,adatabaseforelectronicmed‐icalrecords,oranewplanetinthecomputergameEntropia.
What do instantiations instantiate?
Short answer: A model of a working system
Long answer: In this book, an instantiation has been defined as a work‐ing system that can be used in a practice. However, the term “instantia‐tion”, which has become well established in the design science commu‐nity, indicates that the system should be an instantiation of some other artefact. So, the question arises ‐ what kind of artefact is instantiated? The candidates are the three types of abstract design science artefacts: construct, model and method. Constructs are out, because instantiating a construct will give rise to something that is much too small to be a working system. Models, such as blueprints and architectures can clear‐ly be the basis for a working system, so models can be instantiated. The remaining question is whether methods can also be instantiated. Apply‐ing a method for developing a system should obviously result in a work‐
23
ing system. However, this system is not an instantiation of the method. Rather, the instantiation of the method consists of the actual actions and events that occurred during the system’s development. Thus, a method instantiation would not count as an instantiation when the lat‐ter is defined as a working system.
It could now be argued that our definition of instantiation is too nar‐row and that method applications should also be included in it. One argument in favour of this view is that methods are so complex and situation dependent that they cannot be fully specified in words. In‐stead, some of the knowledge has to be embedded in particular method applications. Such knowledge can then be communicated, for example, through video recordings. Although in principle this argument has much merit, it is in practice quite difficult to transfer knowledge through method applications, because in the IT area, these typically span over extended periods of time. Watching videos of many hours of meetings and discussions is not an attractive option. Thus, our definition of in‐stantiations as working systems is not very restrictive in practice. How‐ever, it could be speculated that in other areas, such as physiotherapy, method applications would be useful design artefacts. Watching a one‐minute video on an innovative physiotherapy exercise could be a highly effective instrument for knowledge transfer.
Must an instantiation always be an IT system?
Short answer: No
Long answer: An instantiation has been defined as a working system that can be used in practice. Clearly, such a system can be an IT system, like a hardware device, a computer program, or a combination thereof. However, there is no reason to make a restriction to only IT systems. An instantiation could also include other components, such as an IT system in an organisation together with new work processes and business rules. Another example could be a digital art installation that consists of sculptures, people and IT artefacts. In general, an instantiation can con‐tain any physical, informational, biological or social components as long as they comprise a working system.
24
Caveat: Although an instantiation can be more than IT, much work on instantiations within design science has actually focused on IT systems. One probable reason for this is that it is complex to construct and eval‐uate artefacts that contain not only IT but also physical, human or social components.
Further Reading
The knowledge types introduced in this paper builds on theworkperformed by Gregor (2006). Knowledge forms are discussed byMadhavan andGrover (1998). The artefact types of design scienceare introduced and described byMarch and Smith (1995) andHe‐vneretal.(2004).
25
3 Research Strategies and Methods
Thepurposeof research is to create reliable andusefulknowledgebased on empirical evidence aswell as logical arguments. The evi‐denceandtheargumentsneedtobepresentedinaclearwaytooth‐erresearchers,sothattheycanreviewthemanddeterminewhethertheyholduptothestandardsofacademicresearch.Inordertosup‐portresearchersincreating,structuringandpresentingtheirresults,manyscientificcommunitieshaveevolvedandestablishedanumberofresearchstrategiesandmethods.Thischapteroffersanoverviewofanumberofwell‐establishedresearchstrategiesandmethodsforempirical research, particularly within the social sciences. Thesestrategiesandmethodsareusefulalsowhendoingdesignscience,inparticularwhen investigating practical problems, defining require‐mentsandevaluatingartefacts.
3.1 Data Generation and Analysis Methods
An early activity in any empirical research study is to collect dataabout the phenomenon under investigation. For this purpose,datagenerationmethods are used. The data collected may be numeric(oftencalledquantitativedata),e.g.,numberoflinesofcode,number
26
ofsoftwareinstallations,andthevisitorfrequencyofawebsite.Oth‐erkindsofdataincludetext,sound,images,video,etc.(oftencalledqualitative data). Five of the most widely used data generationmethodsare:interviews,groupdiscussions,questionnaires,observa‐tionstudies,anddocumentstudies.
Interviews
Aninterviewisacommunicationsessionbetweenaresearcherandarespondent, where the researcher controls the agenda by askingquestionsof therespondent.An interviewcanbestructuredmean‐ingthatitstrictlyfollowsapredefinedprotocol.Alternatively,itcanbesemi‐structuredorunstructured,providingopportunities fordi‐gressingfromaprotocolandallowingtherespondenttotakeinitia‐tives.An exampleof an interview is a researcher evaluating a newMMORPGbyaskingaplayerofthegameaboutherviewofitsgraph‐icalinterface.
Group discussions
Agroupdiscussionisacommunicationsessioninwhicharesearcherand a group of respondents interact under the guidance of the re‐searcherandtheparticipantsinfluenceeachotherinordertogener‐ateinformation.Anexampleofagroupdiscussionisagroupofbusi‐ness intelligenceexpertstogetherwitharesearcher,definingmajorsteps in a method for identifying key performance indicators formeasuringorganisationalperformance.
Questionnaires
A questionnaire is awritten document consisting of questions dis‐tributed to a number of respondents. A questionnaire can includequestionswithpredefinedanswers,aswellasopenquestionswherethe respondent cananswermore freely.Anexampleof aquestion‐naireisdifferentstakeholdersinanorganisationfillinginaformonhowtheyprioritiseamonganumberofrequirementsforaplannedCRMsystem.
27
Observation studies
Inanobservationstudy,objectsareobserved in theirnaturalenvi‐ronments.Theresearcheroftentriestobeasinvisibleaspossibleinordernottointerferewiththenaturaldynamicsofthesituationun‐derinvestigation.Anexampleofanobservationstudyisaresearchergatheringinformationonthelearnabilityofamobileappbyactuallyobservingpeopleusingit.
Document studies
Adocumentstudyisaspecialkindofobservationstudywhereexist‐ingdocumentsareexamined,includingpolicydocuments,userman‐uals, video recordings, system specifications, system logs andweb‐sites. An example of a document study is a researcher identifyingproblems experienced during a systems development project byreadingrelevantmeetingnotes.
Whendatahavebeengenerated, theyneedtobeanalysed.Forthispurpose, there exist bothquantitative andqualitativedata analysismethods. Quantitative data analysis focuses on numeric (quantita‐tive)dataandusesmathematicalapproaches,inparticularstatisticalones,inordertoinvestigate,interpretandstructuredata.Qualitativedataanalysisisaboutinterpretingdataofanyform(bothqualitativeand quantitative) by discovering themes and patterns in them,therebyusingthepersonalskillsandunderstandingoftheresearch‐er who undertakes the analysis. Examples of qualitative analysismethodsarecontentanalysisandnarrativeanalysis.
3.2 Research Strategies
Datagenerationandanalysismethodsareneverusedinisolationbutalwaysinthecontextofaresearchstrategy,i.e.anoverallapproachtoansweringaresearchquestion.Inadditiontodatagenerationandanalysismethods,aresearchstrategyincludesthegeneralset‐upofthe context in which the research is undertaken. Some well‐establishedresearchstrategiesaresurveys,experiments,casestud‐ies,actionresearch,groundedtheoryandethnography.
28
Surveys
A survey starts by generating data from a large group of objects(people,organisations,systems,etc.)oftenthroughuseofquestion‐nairesordocumentstudies.Thedataarethenanalysed,typicallybyaquantitativemethod, inorder to findgeneralisations.Anexampleof a survey is an investigation aiming at identifying problems thatpublic organisations experience when introducing workflow sys‐tems. Themain data generationmethod of the survey is question‐nairesdistributedamongasampleofmanagersandusersofwork‐flowsystems.
Experiments
Inanexperiment,aresearchercreatesanartificialenvironmentthatwillmakeitpossibletoisolateandstudyasmallnumberofobjects,thereby preventing other objects from influencing those under in‐vestigation. The researcher will then observe (a data generationmethod) thebehaviourandrelationshipsof theobjects in theenvi‐ronment and analyse the generated data (using a data analysismethod),oftenwiththepurposeofestablishingcauseandeffectre‐lationships.Anexampleofanexperimentisaresearcherinvestigat‐ingcausesoffailuresinanembeddedsystembysendingthesystemseveralcommandsinparallel.
Case studies
A case study investigates in detail one specific case of the generalphenomenonunderinvestigation,e.g.oneorganisation,onesystemsdevelopment project, or one mobile application. The purpose of acasestudyistopaintarichpictureofasingleobjectorsituationasabasis for obtaining a deep and comprehensive understanding ofsome general phenomenon. An example of a case study is a re‐searcherstudyingorganisationalchangeasaresultofintroducingabusiness process system in an organisation. To understand thechange,theresearcherfollowstheorganizationfora longperiodoftime. Shemakes use of observations, document studies (analysingsystemlogs),andinterviews.Thereby,shecanidentifydiscrepancies
29
betweentheactualexecutionofbusinessprocesses(usinganalysesof the system logs and observations) and user perceptions of theexecution.
Action research
Inactionresearch,aresearcher introducessomechange intoarealenvironment. For example, a researcher may actively introduce anewmethodorsysteminanorganizationtogetherwithitsemploy‐ees.Shethenstudiesandreflectsontheeffectsofthischangeusinganydatagenerationanddataanalysismethods.Anexampleofactionresearchisaresearcherintroducingamethodforidentifyinginnova‐tivee‐servicesinanorganisation.Theresearcherandtheemployeesoftheorganizationcanapplythemethodtogether,andtheresearch‐erwill collect data about the application. She can also perform anevaluationofthemethod, forexample,byinterviewingtheemploy‐ees.
Grounded theory
In grounded theory, a researcher generates theory or hypothesesfromdata.Shegathersdatafromapracticeandidentifiespatternsorstructuresinthecollecteddata,whichwillbecomethebaseforcre‐atingthetheoryorhypotheses.Amainthrustingroundedtheoryisthat the researcher should try to avoid applying her pre‐understandingortheoreticalknowledgeaboutthedata.Instead,thedatashould“speakforitself”totheresearcherandbethesole,oratleastprimary,basisforcreatingthetheory.
Ethnography
In ethnographic studies, the purpose is to understand the cultureandperspectivesofsomegroupofpeople,e.g.,systemsadministra‐tors ina largecompany.Theresearcher isnotadetachedobserverbut tries to become a part of the studied groupbyparticipating intheirdailyactivities.Thereby, shegainsadeepandcomprehensiveunderstandingofhowsomegroupsviewtheirpractice.Anexampleof an ethnographic study is a researcher working as an employee
30
duringatimeperiodinordertounderstandwhyagroupofemploy‐eesrefrainfromusingadecisionsupportsystem.
Can social science be used when doing design science?
Short answer: Yes
Long answer: Social science offers models and theories as well as re‐search strategies and methods, which can be used when doing design science. Social science models and theories are often key components of the knowledge base used when developing artefacts. As an example, agency theory is essential when developing auditing systems. Further‐more, research strategies and methods in social science are applicable in all the activities of the design science method. For example, inter‐views and observation can be used for requirements definition, and both experiments and case studies can be applied for evaluation.
Figure 3.1 Research strategies and research methods
Figure3.1showstherelationshipsbetweenresearchstrategies,datagenerationmethods, anddata analysismethods. It also reads from
31
toptobottom,asasimpleguidetotheinitialdecisionsinaresearchproject.First,theresearcherwillselectoneormoreresearchstrate‐giestobeapplied.Foreachresearchstrategy,shewillselectoneormore data generationmethods to use aswell as one ormore dataanalysismethods.Inpractice,thesedecisionswillnotbeasserialassuggestedherebutwillratherdevelopiniterationsduringtheentireresearchproject.
Aconcreteexamplecanbearesearcherwhointendstodevelopamethodfordesigningbusinessprocessmodels.Thepurposehereistodevelopanartefact (amethod). Inorder togather requirementsfor themethod,shechoosesasurveyasoneof theresearchstrate‐gies in the project, where she tries to get suggestions for require‐mentsfromexpertsinbusinessprocessmanagement.Forgatheringinformationfromtheexperts,shechoosestouseinterviews(adatageneration method) and she analyses the gathered interview datausing aqualitativemethod. In addition to the requirements elicita‐tion,shealsodecidestoevaluatethemethod.Forthispurpose,shechoosesanexperimentastheresearchstrategy.Shesetsupasitua‐tionwherestudents,aswellasexperiencedprocessdesigners,aretousethemethod.Sheobserves(adatagenerationmethod)thebehav‐iour of these peoplewhen using themethod and logs their results(another data generation method), which she later analyses usingstatistical(quantitative)methods.
Is a theory an artefact?
Short answer: No
Long answer: The answer may seem to be yes, as a theory is clearly cre‐ated by humans. However, to count as an artefact in design science, an object should not only be man‐made but also constructed with the in‐tention to be used for addressing a practical problem. Thus, a theory is not an artefact, as it is not intended to address practical problems but to explain and predict phenomena and behaviours in some domain. A definition by Bhattacherjee (2012) states: “A theory is a set of systemat‐ically interrelated constructs and propositions that are advanced to
32
explain and predict a certain phenomenon or behavior within certain boundary conditions and assumptions. Essentially, a theory is a system‐atic aggregation of theoretical propositions. While propositions connect two or three constructs at most, theories represent a system of multiple constructs and propositions.”
Caveat: Even if a theory cannot be used to directly address a practical problem, it can help to design artefacts that do address such problems. In the words of Kurt Lewin, often recognized as the founder of social psychology: “There is nothing so practical as a good theory”. In fact, design science requires that the design of artefacts always be based on a knowledge base, which preferably includes theories.
Further Reading
Therearemanybooksonresearchmethodologybothfornaturalandsocialscience.ThischaptercloselyfollowsthestructureofresearchstrategiesandmethodsproposedbyDenscombe(2010)anditisalsoclosetothestructuresuggestedbyBhattacherjee(2012).
33
4 A Design Science Method
Design science projects can be large undertakings, involvingmanypeopleandbeingcarriedoutoveranextendedperiodoftime.There‐fore,researcherscanbenefitfromdesignsciencemethodsthatsup‐porttheminstructuringtheirworkandensuringthequalityoftheirresults. Suchmethods can also support design researchers in pre‐sentingtheirworkinalogicalandeasilyunderstandableway.Inthissection,we introduce a generic design sciencemethod that can beadaptedtotheneedsofspecificresearchprojects.Themethodpro‐posedhereisnotanotherresearchstrategyormethod,suchasthoseintroduced inChapter3;rather it isaholisticapproachtoproblemsolvingbymeansofartefacts,offeringaframeworkofactivitiestobeperformedandquestionstobeanswered.
4.1 Activities in Design Science
Thedesignsciencemethodintroducedhereincludesfivemainactivi‐tiesthatrangefromprobleminvestigationandrequirementsdefini‐tion,throughartefactdevelopmenttodemonstrationandevaluation.TheseactivitiesareoutlinedbelowandelaboratedindetailinChap‐ters5to9.
34
Explicate Problem
TheExplicateProblemactivityisaboutinvestigatingandanalysingapracticalproblem.Itistobepreciselyformulatedandmotivatedbyshowingthatitissignificantforsomepractice.Theproblemshouldbeofgeneralinterest,i.e.,significantnotonlyforonelocalpractice.Furthermore, underlying causes to the problem may be identifiedandanalysed.
Outline Artefact and Define Requirements
TheOutlineArtefactandDefineRequirementsactivityistooutlineasolution to the explicated problem in the form of an artefact. Re‐quirements are tobedetermined,whichcanbe seenasa transfor‐mationof theproblem intodemandson theproposedartefact.Therequirementswillbedefinedprimarilyforfunctionalitybutalsoforconstructionandenvironment.
Design and Develop Artefact
TheDesignandDevelopArtefactactivityaims tocreateanartefactthat addresses the explicated problem and fulfils the defined re‐quirements.Designinganartefactincludesdeterminingitsfunction‐alityaswellasitsconstruction.
Demonstrate Artefact
TheDemonstrateArtefactactivityusesthedevelopedartefactinanillustrative or real‐life case, sometimes called a “proof of concept”,thereby proving the feasibility of the artefact. The demonstrationshall show that the artefact actually can solve an instance of theproblemaddressed.
Evaluate Artefact
TheEvaluateArtefactactivity is todeterminehowwell theartefactfulfilstherequirementsandtowhatextentitcansolve,oralleviate,thepracticalproblemthatmotivatedtheresearch.
Anoverviewoftheproposeddesignsciencemethodisshowninfig‐ure 4.1, which also indicates the results from each activity. The
35
method, as describedhere,may lookhighly sequential.However, adesignscienceprojectisalwayscarriedoutinaniterativeway,mov‐ingbackandforthbetweenalltheactivitiesofproblemexplication,requirements definition, development and evaluation. This way ofworkingisfullyinlinewiththeproposeddesignsciencemethod,asitdoesnotprescribeasequentialworkordering.Thearrowsinfig‐ure4.1shouldnotbeinterpretedastemporalorderingsbutasinput‐outputrelationships.Inotherwords,theactivitiesarenottobeseenastemporalgroupingsofworktobeperformedinsequencebutin‐steadaslogicalgroupingsofwork.Theseissuesarefurtherelaborat‐edinChapter11.
Is design science the same as action research?
Short answer: No
Long answer: The purpose and activities of design science are close to those of action research. Both aim to change and improve human prac‐tices and thus can be viewed as practice research, as pointed out by Goldkuhl (2012). Furthermore, both include problem solving and evalu‐ation. However, there are also differences. Design science addresses practical problems through the design and positioning of artefacts, while action research does not require an artefact to be part of the solu‐tion addressing the practical problems. Instead, action research ad‐dresses problems through psychological, social and organizational change. Furthermore, design science does not require a practical prob‐lem from a specific organisation as the starting point for the research, while this is required in action research. Finally, action research is a single research strategy, while design science can make use of many different research strategies, e.g. one for problem explication, another for requirements definition and a third for evaluation.
Caveat: It could be argued, however, that if an artefact is proposed as a solution in an action research project, action research becomes very similar to design science. Clearly, action research is a highly relevant research strategy for many design science projects, and there are also design science methods based on action research, e.g. (Sein et al. 2011).
36
Figure 4.1 Overview of the Design Science Method
37
4.2 Research Strategies and Methods in Design Science
Thepurposeofdesignscienceisnotonlytocreateartefactsbutalsotoanswerquestionsaboutthemandtheirenvironments.Inordertoensure that the answers express valid and reliable knowledge, theuseofresearchstrategiesandmethodsarevital.Withindesignsci‐ence,itispossibletouseanyresearchstrategyormethodtoanswerquestionsaboutartefacts. Inotherwords,noresearchstrategiesormethodscanbeexcludedinadvanceforadesignscienceproject,asanyofthemmaybevaluabledependingontheproject’scharacteris‐ticsandgoals.
Inlargedesignscienceprojects,itisevencommontouseseveralresearch strategies and methods, because different design scienceactivitiesmay require different approaches. For example, a surveycan be an appropriate instrument for eliciting requirements,whileanexperimentcanbethebestchoiceforevaluation.Figure4.2out‐lines commonly used research strategies for design science activi‐ties;itshouldbeseenasmerelyindicative,asanyresearchstrategycanbeusedinanyactivity.
For problem explication, surveys are effective instruments, astheycanbeusedtoquicklygatheropinionsandperceptionsaboutaproblem situation from a large number of stakeholders. However,the results from surveys are often superficial, as stakeholdersmaynot be prepared to spendmuch time and effort on describing andanalysingtheproblemunderconsideration.Adeeperunderstandingabout a problem and its context can be achieved through a casestudy, where researchers investigate the problem for an extendedperiodoftime,usinginterviews,observationanddocumentstudies.However, results fromcase studiesmaybedifficult to generalise iftheyarebasedonlyononelocalpractice.
38
Figure 4.2 Suggested use of research strategies in the design science method
39
For requirements definition, surveys are also effective instru‐ments.However,itmaybehardforstakeholderstosuggestrelevantrequirementswhen theartefact tobedesigned ishighly innovativeorcomplex. Insuchcases,actionresearchandcasestudiesmaybebetter alternatives, as they allow a researcher to identify require‐mentsherself,withouthavingtorelyonstakeholderstobeinventiveandcommitted.Adrawbackisthatrequirementsdefinitionmaybe‐come heavily dependent on the competence and experience of theresearcher.
For design and development, research strategies are often lessimportant,astheprimarygoalofthisactivityistoproduceanarte‐factandtoalesserextent,theknowledgeaboutit. Instead,creativemethodsaremorerelevant,e.g.,brainstorming,participativemodel‐ling,empatheticthinkingandlateralthinking.
Demonstration is about using an artefact in a specific case toshow its feasibility. Thus, the most obvious research strategies toapply are action research and case studies,whichwill typically in‐cludeinterviewsandobservationsofpeopleintheiruseofthearte‐fact.
Forevaluatinganartefact,experimentsarepopularinstruments,astheyallowaresearchertoachievehighinternalvaliditybycare‐fullycontrollingtheconditionsunderwhichanexperimentisunder‐taken. On the other hand, external validity can suffer, because theartificial setting of the experiment can be decidedly different fromthepractice inwhich theartefact is tobeused.Analternative is touse case studies or action research,where the artefact is used andevaluatedinalocalpractice.
4.3 Relating to a Knowledge Base
Theresultsfromadesignscienceprojectshouldbeoriginalandwellfounded.Therefore,itisrequiredtorelatebothdesignscienceactivi‐ties and their results to an existing knowledge base. Such aknowledgebasemayincludemodelsandtheoriesfromseveralfieldsof science. In the IT and information systems area, theories from
40
social and behavioural sciences are particularly relevant but alsomodels from formal sciences. Furthermore, a knowledge base willtypicallyincludeinformationaboutartefactssimilartotheoneunderconsideration.
Relevantknowledgemaybefoundinacademicpublications, in‐cludingtextbooksandresearchpapers in journalsandconferences.A good starting point for identifying these kinds of publications isGoogle Scholar.Other sourcesof knowledge includemagazine arti‐cles,whitepapers,fairtradepresentationsandartefactmanuals.
4.4 Typical Cases of Design Science Projects
Many design science projects do not undertake in depth all of thefiveactivitiesofthedesignsciencemethod.Instead,theymayfocuson one or two of the activities, while the others are treatedmorelightly.Basedontheirfocus, it ispossibletodistinguishbetweenatleastfivetypicalcasesofdesignscienceprojects.
Problem focused design science
Some design science projects focus on problem explication. Theycarefully investigate a problem situation and divide it into sub‐problems,andmayalsocarryoutarootcauseanalysis.Theytherebyemployresearchstrategiesandmethodsinarigorouswayandtypi‐cally include comprehensive empirical studies. These projects alsodefine requirements on an artefact based on more or less carefulinvestigations.Thedesignoftheartefactisonlyoutlinedandneitherdemonstration nor evaluation is carried out. Projects of this kindoftenhaveastrongsocialscienceflavourandthedesignelementisdownplayed. However, they can provide essential problem under‐standinguponwhichsubsequentdesignscienceprojectscanbuild.
Requirements focused design science
Other design science projects focus on requirements definition.Theseprojectsstartwithanexistingproblemthatissimplyacceptedas is, or slightly explicated. Careful and rigorous investigations are
41
carriedoutinordertocollectrequirements,whichwillinvolveliter‐aturestudies,aswellasinteractionwithrelevantstakeholders.Thedesignoftheartefactisonlyoutlinedandneitherdemonstrationnorevaluationiscarriedout.
Requirements and development focused design science
Many design science projects focus on a combination of require‐ments definition and artefact development. Such projects will notperformanyproblemexplicationbutmovedirectlytorequirementsdefinition,whichiscarriedoutinarigorousway.Theartefactisthendevelopedusingresearchaswellascreativemethods.Theviabilityof the artefact is demonstrated or a lightweight evaluation is per‐formed.
Development and evaluation focused design science
Itisalsocommonthatdesignscienceprojectsfocusondevelopmentandevaluation.Suchprojectswillneitherperformproblemexplica‐tionnorrequirementsdefinitionbutstart fromanexistingrequire‐mentsspecification.Thefocuswillbethedesignanddevelopmentofanartefactusingbothresearchandcreativemethods.Therewillalsobeademonstrationanda thoroughevaluationbymeansof experi‐ments,casestudies,orotherresearchstrategies.
Evaluation focused design science
Somedesignscienceprojectsfocusontheevaluationactivity.Thus,noartefactisdeveloped,norevenoutlinedandtherefore,itcouldbequestionedwhether such a project should count as design science,seetheboxbelowforadiscussion.Evaluationprojectsoftenincludecareful requirements definition, which results in a requirementsspecification used as a basis for the evaluation. The evaluation iscarriedoutinarigorouswayusingadequateresearchstrategiesandmethods.
42
Can an evaluation study be design science?
Short answer: Yes
Long answer: Consider a study that does not design and develop a new artefact but only evaluates an existing one. Can such a study count as design science? At first glance the answer seems to be no, as design science is about the development of novel artefacts. However, an eval‐uation study can be part of a larger design science undertaking, which may extend over a long time involving a number of research groups and individuals. One study may explicate a problem and suggest an artefact that can help to solve it; another study may define requirements on the artefact; a third study may develop the artefact; and yet another study may evaluate it. All of the studies are part of the same design science undertaking, although the same people do not necessarily carry them out. In this sense, the studies can all be viewed as design science contri‐butions.
The question about evaluation studies can be generalised to problem explication studies, requirements definition studies and artefact de‐signs. Using the same argument as above, all such studies can count as design science. Even if none of them covers all the design science activi‐ties, together they extend the knowledge about an artefact and its con‐text.
Caveat: Even if very narrow studies can still count as design science, they may not meet the requirements for a bachelor, master or PhD the‐sis within certain study programs.
4.5 Visualizing the Design Science Method
TheproposeddesignsciencemethodcanbevisualizedusingIDEF0,whichisatechniquefordescribingsystemsasanumberofinterre‐latedactivities,graphicallyrepresentedasboxes,seefigure4.3.Eachactivitycanbedecomposedintosub‐activities,whichthemselvescanbedecomposed into further sub‐activities, andsoon.Furthermore,channelsconveyingdataorobjectsarerelatedtoeachactivity.
43
Figure 4.3 IDEF0 diagram
FourkindsofchannelsareusedinIDEF0:input,output,controlandresources. Inputs are transformed or consumed by an activity toproduce outputs. Controls (e.g., blueprints, rules, guidelines,meth‐odsandinstruments)governanactivitytoproducecorrectoutputs,whileresourcesarethemeans(e.g.,genericanddomainknowledge)that support an activity. Channels are represented graphically asarrows; inputarrowspoint to the left sideof abox;output arrowsleavetherightsideofabox;controlarrowspointtothetopsideofabox;andresourcearrowspointtothebottomofabox.
Inthedesignsciencemethod,thechannelscanbedefinedasfol‐lows:
Input(arrowfromleft)–describeswhatknowledgeorobjectistheinputtoanactivity
Output(arrowtoright)–describeswhatknowledgeorobjectistheoutputfromanactivity
Controls (arrow from above) – describewhat knowledge isusedforgoverninganactivity,includingresearchstrategies,researchmethods,andcreativemethods
Resources(arrowfrombelow)–describewhatknowledgeisused as the basis of an activity, i.e. the knowledge base in‐cludingmodelsandtheories
44
Figure 4.4 IDEF0 diagram for the Design Science Method
Figure4.4displaysthedesignsciencemethodasanIDEF0diagram.Whenapplyingthemethodinadesignscienceproject, thediagram
45
maybeusedasatemplateandfilledinwiththespecificsofthepro‐ject.Thiswill result ina convenientoverviewof theproject,whichcanbeusedfordocumentationandcommunication.
Must a design science project always create a new artefact?
Short answer: No
Long answer: The purpose of a design science project is to address a practical problem by means of an artefact. The project may develop a new artefact from scratch or refine an existing one. The project can even repurpose an existing artefact, i.e. use it for a new purpose with‐out changing it. An existing artefact is used to solve another problem than that for which it was originally designed and developed. For exam‐ple, the anticoagulant chemical Warfarin was introduced as a rat poison but later repurposed as a blood thinning medicine. Gunpowder started out as a medical elixir in China centuries before it was repurposed for powering fireworks and firearms. Thus, a design science project does not have to create a new artefact as it may transform an existing one into something entirely different through repurposing. In terms of the design science method suggested in this book, such a project would skip the activity Design and Develop, while the other activities would still be addressed.
4.6 The Design Science Canvas
ThroughtheIDEF0representationofthedesignsciencemethod,itispossible toobtain a comprehensive, yet compactoverviewof ade‐sign science project. However, it is sometimes desirable to get anevenmorecompactdescriptionandforthispurpose,theDesignSci‐enceCanvas canbeused.TheCanvasoffersaconcise,easilyunder‐standableandvisuallyappealingoverviewofthekeycomponentsofadesignscienceproject.TheCanvasconsistsofarectangledividedinto a number of fields that describe the artefact under considera‐tion, the problem it addresses, the knowledge base used, the re‐quirements on the artefact, the research strategies and methodsusedintheproject,andtheresultsoftheproject.
46
Figure 4.5 The Design Science Canvas
47
AtemplatefortheDesignScienceCanvaswithdescriptionsofthe
fieldsispresentedinfigure4.5.AmoredetailedversionoftheCan‐vascanbefoundatgoo.gl/TKD5S.ThetoppartoftheCanvas(bluein the figure)defines theartefactunderconsideration, theproblembeingaddressedandtheknowledgebaseusedinthework.Themid‐dlepart (yellow in the figure)describes theactivitiesof thedesignsciencemethod,correspondingtoChapters5to9.Thebottompart(red in the figure) shows the results of a design science project intermsoftheconstruction,function,usabilityandeffectsofthearte‐fact.
TheDesignScienceCanvasmaybeusedinseveraldifferentsitu‐ations.Itcanbeusedasafirstsketchinordertoimprovecommuni‐cationandcommonunderstandingbetweenmembersofanewde‐signscienceproject.Itcanalsobeusedasamonitoringtoolforcon‐tinuouslyrecordingchangestotheplansforaproject.Furthermore,the Canvas can help to present the overall setup of a project tostakeholdersandotherinterestedparties.Asimilaruseistoincludethe Canvas in the final project report as part of an executive sum‐mary.Ingeneral,theCanvasisusefulwheneverthereisaneedtogeta compact and easily understandable overview of a design scienceproject.
Can the Design Science Canvas help to structure thesis work?
Short answer: Yes
Long answer: The Design Science Canvas can serve as a communication tool for the participants in a thesis work, including authors, employers, supervisors, examiners, fellow students, and other interested parties. In the initial stage, a draft of the Canvas can function as a plan for the the‐sis project, particularly by outlining those parts of the design science method that are in focus. Later on, the Canvas can be regularly updated to reflect the progress of the work, thereby becoming more and more detailed. The Canvas can also help to structure the thesis text, as the Canvas fields can work as summaries of key parts of the thesis. Thus, the
48
Canvas can be used to structure the thesis project, as well as the thesis text and it can support communication among the stakeholders of the project.
4.7 Running Example
Inordertoillustratetheuseofthedesignsciencemethod,afictitiousrunningexample ispresented inChapters5 to9. Inrecentyears,anumber of publishing houses have experienced decreasing sales ofbooks,asreadingbooksingeneral is indecline,at leastwhencom‐pared to other kinds of media consumption. In particular, youngpeople seem to be less and less interested in book reading. At thesametime, the Internetandmobile technologiesoffermanyoppor‐tunitiesforenrichingthebookreadingexperiencebutthesehavenotbeenfullyexploitedasyet.Publishinghousesbelievethattheyneedtomakeajointefforttoaddresstheproblemandgraspthenewop‐portunities. They also realise that the problem is complex and inneedofresearch if it is tobeadequatelyaddressed.Thepublishinghouses, together with a number of university researchers, havethereforedecidedtolaunchadesignscienceprojecttoaddressthisproblembydesigninganovel artefact thatwill increase interest inbook reading. The project is called DIGIREAD and is further de‐scribedinChapters5to9.
What are the similarities and differences between the scientific method and the design science method?
According to the Oxford English Dictionary, the scientific method is "a method or procedure that has characterized natural science since the 17th century, consisting in systematic observation, measurement, and experiment, and the formulation, testing, and modification of hypothe‐ses". One common version of the scientific method is the hypothetico‐deductive method, which is typically divided into four steps:
1 Ask a question. The researcher observes some phenomenon that is novel, surprising, or interesting for some other reason. She attempts to
49
capture the relevant aspects of the phenomenon by asking a question about it.
2 Form a hypothesis. The researcher comes up with and formulates a hypothesis that is able to answer the question. This hypothesis often includes a causal mechanism or a predictive model. The hypothesis may also be made up of a set of smaller, interrelated hypotheses.
3 Deduce predictions from the hypothesis. Assuming the hypothesis is true, the researcher identifies some consequences that must then hold, i.e. she makes a number of predictions.
4 Check the predictions. The researcher performs observations to de‐termine whether the predictions are correct or not; if correct, the hy‐pothesis is strengthened, otherwise it is weakened.
The design science method is aligned with the scientific method. Its first activity, Explicate Problem, is similar to Ask a question of the scientific method, as both investigate a situation that is experienced as challeng‐ing. The difference is that in the scientific method, a question is asked, while a practical problem is examined in design science. Step 2, Form a hypothesis, is similar to activities 2 and 3 of the design science method, which first identify requirements on an artefact and then design and develop it. However, in the scientific method an answer is formulated in the form of a hypothesis, while design science produces an artefact. Steps 3 and 4 of the scientific method correspond to Evaluate Artefact of the design science method, as both intend to show that the results produced are satisfactory. Summarizing, steps 1 and 2 of the scientific method, as well as activities 1 to 3 of the design science method, are about discovery and invention, i.e. creating a new hypothesis or arte‐fact. Steps 3 and 4 of the scientific method, as well as activities 4 and 5 of the design science method, are about justification, i.e. ensuring that the created hypothesis or artefact is adequate.
Further Reading
The design sciencemethod of this chapter is close to the one pro‐posedbyPeffersetal.(2007),thoughwehaveomittedthecommu‐nication activity and emphasized requirements elicitation. Other
50
similar methods have been proposed by March and Smith (1995)andVaishnaviandKuechler(2004).
51
5 Explicate Problem
ThefirstactivityofthedesignsciencemethodisExplicateProblem.Thegoalofthisactivityistoformulatepreciselytheinitialproblem,motivate its importance, and investigate its underlying causes. Inotherwords,itaddressesthequestion:
What is the problem experienced by some stakeholders of
a practice and why is it important?
The answer to this question primarily consists of descriptiveknowledge about the characteristics and environment of the prob‐lem.Sometimes,theanswerwillalsoincludeexplanatoryknowledgeaboutproblemcauses.
Asdiscussed inChapter1, aproblem is anundesirable state ofaffairs,ormoreprecisely,agapbetweenthecurrentstateandade‐sirablestate.Forexample,supposethatseveralcustomerscomplainabout the long delivery times of a company. The customers expectthe time fromorder placement to product delivery to be less thanoneweek(desirablestate) insteadof thecurrentthreeweeks(cur‐rentstate).Thus,thisisthegap,whichconstitutestheproblem.
The gap between the desirable and the current state is not al‐waysmadeexplicitwhenaproblemisdiscussed.Often,thegapisso
52
obviousthatknowledgeofthecurrentstateissufficienttoconcludethataproblemexists.Forexample,ifmanycustomersofacompanycomplainaboutdeliverytimes,itsmanagementwillrealisethatcus‐tomers are dissatisfied and that there exists a problem to be ad‐dressed. In this case, the desirable state is not explicitly stated. Inother cases, a problemmay become apparent onlywhen someonesuggestsamoredesirablestateofaffairs.Forexample,supposethatnocustomerhascomplainedaboutthedeliverytimes,butacompeti‐torstatesinamarketingcampaignthatitsdeliverytimeisonlythreedays fromorderplacement. If themanagement interprets this as athreat, therewill be aproblem, although the current statewasnotviewedasundesirableinitself.
Notonly threatsbut alsoopportunities canbeviewedasprob‐lems.Anexample isanorganizationthatreceives informationfromits ERP vendor thatmobile phones can be integratedwith its ERPsystem. Thereby, the employees can access the system from any‐where,whichwill increase their productivity. Therefore, the prob‐lemisthattheorganisationcurrentlydoesnotworkasproductivelyas possible, because its employees do not benefit from this oppor‐tunityofmobiletechnology.
TheactivityExplicateProblemcanbestructuredandvisualizedusingtheIDEF0technique,seefigure5.1.Theinputisaninitialprob‐lem that can be vaguely formulated. The output is an explicatedproblem,which is precisely defined,well‐motivated and put into acontext.Theresourcesusedby theactivityconsistofknowledge inthe research literature and otherwritten sources, aswell as infor‐mation from relevant stakeholders. The controls are primarily re‐searchstrategiesandmethodsbutmayalso includeotherpractice‐basedapproachestoproblemelicitationandrepresentation.
53
Figure 5.1 Explicate Problem
5.1 Sub‐activities of Explicate Problem
Whentheinitialpracticalproblemisobscurelyexpressedorincom‐pletelyunderstood,thereisaneedtoinvestigateit,sothatresearch‐erscansuggestanappropriatesolution.However,insomecases,theinitialproblemisalreadyunderstoodandclearlyarticulated,mean‐ingthattheExplicateProblemactivitywillbesmall.Problemexpli‐cationconsistsofthreemainsub‐activities:makingtheproblemdef‐inition as precise as possible, motivating the problem and findingrootcauses.
Define precisely
Aproblemshouldbedefinedaspreciselyaspossiblesothatdiffer‐ent people understand it in the sameway. A problem definition ismademoreprecisebyreducingthenumberofwaysinwhichitcanbe understood and interpreted. For example, a problem definitionsuchas“non‐integratedITsystemsresult in longdeliverytimesfororganisations” is vague and may be interpreted in many differentways.Itcanbemademoreprecisebybeingreformulatedas“thelackofintegrationofITsystemssupportingtheordermanagementpro‐
54
cessand thedeliveryprocess, results in longdelivery times foror‐ganisations”. The second definition has fewer possible interpreta‐tionsthanthefirstone.Ingeneral,preciseproblemdefinitionsaretobepreferredoverlesspreciseones,astheysupportpeoplebettertodevelopacommonviewofaproblem.However,highlypreciseprob‐lemdefinitions can sometimesbedifficult toquicklygraspandun‐derstand.
Motivate Problem
Aproblemshouldalwaysbewellmotivatedsothatpeoplecanagreeit is worthwhile to address. The problem should be significant forsomepractice,i.e.,viewedasimportantbystakeholderswhowanttofindasolutiontoit.Furthermore,theproblemshouldbeofgeneralinterest, i.e., itshouldnotmatteronlytoasinglelocalpractice.Theproblemshouldalsobechallenging,inthesensethatasolutiontoitdoesnotalreadyexist.Sometimesaproblemcanbeoriginal,whichisparticularlycommonwhentechnologicalinnovationshavecreatednewopportunities.Themotivationofaproblemmayalsoincludeitsethicalandsocietalconsequences.
Find Root Causes
In anearly stage, aproblemoftenexpresses some symptomofun‐derlying causes that are as yet unknown. In order to address theproblem effectively, a so‐called root cause analysis can be per‐formed, where the underlying causes are identified, analysed andrepresented. By addressing these causes, better results can beachieved than by only treating the symptoms of the problem. Forexample,an initialproblemmaybeexpressedas“thecompanyhasdeliverytimesthataretoolong”.Arootcauseanalysismayindicatethat themain reason for thisproblem is the lackof integrationbe‐tweentheITsystemsmanagingtheordermanagementprocessandthedeliveryprocess.Consequently, time‐consumingmanualadmin‐istrativeworkhastocompensateforthelackofsystemsintegration.Thus, theunderlyingproblemcanbedefinedas “non‐integrated IT
55
systemsresultinlongdeliverytimesfortheorganisation”.Thefocusmaythenshiftfromtheinitialproblemtothisunderlyingproblem.
OnewidespreadtoolforrepresentingproblemcausesistheIshi‐kawa diagram (also called cause‐effect diagram or fishbone dia‐gram),seefigure5.2foranexample.AnIshikawadiagramisagraph‐icaltoolusedtoinvestigateandrepresentpotentialcausesofaprob‐lem. It consists of amainhorizontal line representing the problemand associated slanting lines representing direct problem causes,which in their turnmaybe related to additional lines representingindirectproblemcauses.Thecausescanalsobeclassifiedintodiffer‐entcategories,asindicatedinfigure5.2.
Figure 5.2 Ishikawa diagram
5.2 Resources for Explicate Problem
The results of the activity should be based on and comparedwithexistingrelatedwork.Therefore,theresearcherneedstoinvestigatewhatpreviousresearchhasaddressedsimilarproblemsandexistingsolutions. Not just research literature can be used but also othersources,e.g.,newspaperarticlesandwhitepapers.
56
Insomecases,aresearchercanbaseaproblemexplicationsolelyon literature, but usually she also needs to directly study partici‐pants and stakeholders of relevant practices. Stakeholders in thepracticesmaythemselvesexpressviewsandopinionsaboutaprob‐lem,whichthenaretobeinterpretedbytheresearcher.Shecanalsogain a better understanding of the practices by observing partici‐pants in their daily activity. Furthermore, the researcherwill oftenneed to study different groups of stakeholders, e.g.managers, em‐ployees and customers, as they may have different views andknowledgeaboutvariousaspectsoftheproblem.Bycombiningcon‐tributions from different stakeholder groups, the researcher canachieveadeeperandmorecompleteexplicationoftheproblem.
5.3 Strategies and Methods for Explicate Problem
Research Strategies for Explicate Problem
Surveys
Surveys can be used for eliciting problem statements from a largegroupof stakeholders.Thereby, theyprovide anoverviewof prob‐lems experiencedby, for example,managers, employees, end‐usersandcustomers.Inmanycases,differentstakeholdershavedifferentviewsof theproblemat hand, and a survey canmake these differ‐encesexplicit.However,asurveyistypicallynotaneffectiveinstru‐ment for eliciting a deep and elaborated problem analysis fromstakeholders.
Case studies
Case studies can provide a deep understanding of the practice inwhich the initialproblememerged.Thisestablishesa firmgraspoftherootcausestotheproblemaswellasthestakeholders’viewsontheproblem.However, case studies are complex undertakings thatheavilyrelyontheskillsandexperiencesoftheresearcherperform‐ing them. This dependency on the individual researchermay be a
57
drawback, as she may have interests and preconceptions that canbiastheresearchwork.
Action research
Action research requires an active engagement of researchers in apractice. The competence and experiences of the researchers mayoffer freshperspectivesontheproblemthatarenotobvious to thestakeholdersofthepractice.Furthermore,newandmoreimportantproblemscanemergewhentheresearchersarediscussingopportu‐nities and presenting solutions to the stakeholders. However, thedependencyontheresearchersisstrongduetotheiractivepartici‐pation in thepractice.Therefore, there is a risk that their interestsand preconceptions will have too much influence on the problemexplication.
Grounded Theory
Groundedtheoryisaresearchstrategyinwhichpureempiricalfactshaveastrongimpactontheexplicationofaproblem.Theresearch‐ers start by gathering facts about the domain under consideration.Basedonthesefacts,theysuggestafirstproblemexplication,whichistestedagainstfurtherempiricalfactsfromthedomain,resultinginarefinedproblemexplication.Theiterationsbetweenfactgatheringandproblemexplicationrefinementcontinueuntilfurtherempiricalfacts have no effect on the problem explication. An advantage ofgroundedtheoryisthelackofanyrestrictionbyanyspecifictheoret‐icalviewthatmaylimittheresearchers.However,thisisalsoadis‐advantage,asatheoreticallenscansupporttheresearchersinfind‐ingnewperspectivesontheproblem.
Ethnography
The research strategy ethnography allows a researcher to under‐stand thecultureofapractice indepth.Thereby, she isable toseetheproblemnotonlyasanoutsiderbutalsofromthestakeholders’point of view. Furthermore, based on her competence and experi‐ence, the researcher may understand the structures behind thestakeholders’ views and actions, which they themselvesmight not
58
recognise. This knowledge can allow the researcher to arrive at adeepandrichproblemexplication.However,becauseethnographicalstudies are time‐consuming, shemayonly be able to understand alimitednumberofstakeholders,whileotherstakeholdersmaynotbeconsidered.Theoutcomeofthisresearchstrategyalsoheavilyreliesonthecompetenceandexperienceoftheresearcher.
Research Methods for Explicate Problems
Interviews
Interviews allow a researcher to engage in a dialogue with a re‐spondentinordertoexplicateaprobleminaninteractiveandcrea‐tiveway. This is possible because the researcher, based on the re‐spondent’sinitialanswers,canaskfollowupquestions.Adrawbackof interviewsis thedependencyontheperspectiveandinterestsofthe respondent, but this problem canbemitigatedby interviewingseveralrespondents.
Group discussions
Agroupdiscussionisaresearchmethodwhereseveralrespondentsinconversationsmayinspireeachothertoidentifyanddefineprob‐lemsinadomain.However,thereisariskthatdominantindividualsinsuchadiscussionhavetoogreatanimpactsothatotheropinionsarenotvoiced.Therefore,itmaybenecessaryforaskilledmodera‐tortomanageagroupdiscussion.
Questionnaires
Aquestionnaireisaformthatcontainspredefinedwrittenquestions.A main benefit of using questionnaires for data generation is thattheycaneasily,andwithlowcost,bedistributedtoalargenumberof respondents. A drawback is that a researcher and a respondentcannotdiscussaproblemsituationinformallyandcreatively.Anoth‐erdrawbackisthatrespondentscaninterpretthewrittenquestionsofaquestionnaireindifferentways.
59
Observation studies
In an observation study, researchers can observe the behaviour ofpeople in a practice. A benefit of the method is that researchers,based on their competence and experience, can identify problemsandcircumstancesthatarenotapparenttothepeopleunderobser‐vation. A drawback is that the method requires highly skilled re‐searchersto interprettheactionsandinteractionsofthepeoplein‐vestigated.Thereisalsoariskthattheinterestsandpreconceptionsof the researchersmay influence their interpretations inunwantedways.
Document studies
Adocumentstudyisaformofanobservationstudybutthefocusisonwritten documents, not actions.Written documents can exposecontradictions inapracticeand, therefore,beavaluablesource foridentifying and defining problems. However, the method requiresskilled researchers for interpreting the documents. There is also ariskthatsomedocumentsonlyshowtheofficialviewofsomeactorandmayhideexistingproblems.
5.4 Guidelines for Explicate Problem
The results of the problem explication will govern the rest of theresearchprocess.Therefore,athoroughproblemexplicationwillbevaluable for all the other design science activities. The followingguidelinescanbeusedtosupportproblemexplication.
Position theproblem!Clarify inwhich practice the problemappears.
Formulate theproblemprecisely! Describe the problem in aprecisebutalsoconcise,easilyunderstandablemanner.
Motivatetheproblem!Explainwhytheproblemisimportantandforwhom.
Ensuretheproblemisofgeneralinterest!Makeclearthattheproblemisofinterestnotonlytoalocalpractice.
60
Ensuretheproblem issolvable!Defineandanalyzetheprob‐lemsothatitbecomessmallenoughtobesolved.
Specifythesourcesoftheproblem!Describetheliteratureandthestakeholdersthathavepreviouslyidentified,studiedandexperiencedtheproblem.
Describehowyouhaveexplicated theproblem!Explainwhatyou have done to explicate the problem, in particular howyouhavereviewedthestakeholdersandresearchliterature.
5.5 Running Example
ThefirstactivityforDIGIREAD,presentedinSection4.7asarunningexample, was to explicate the problem. There were three sub‐activities: defining the problem precisely, motivating the problemandfindingtherootcauses.
Thefirstsub‐activitywastodefinetheproblemprecisely inor‐der toreducethenumberofways inwhich itmightbeunderstoodandinterpreted.Theinitialproblemwasformulated:bookreadingisindecline.Inordertoformulatethisstillmoreprecisely,athoroughempiricalsurveycouldthenhavebeencarriedout.However,inthiscase, the sub‐activity was based on public statistics showing thatpeople between the ages of 15 ‐ 25 in particular spend less timereadingbookstodaycomparedtopeopleofthesameageeither5or10yearsago.Amorepreciseproblemwasthereforeformulatedas:bookreadingisindeclineamongpeoplebetweentheagesof15‐25.
Thesecondsub‐activitywastomotivatetheproblemsothatthestakeholders can agree it isworthwhile to address it. For thepub‐lishinghouses,thisisclearlyamajorproblemasitthreatensfuturesales.
Thethirdsub‐activitywastofindtherootcausesoftheproblemsothatbetterresultscouldbeachievedthanbysimplytreatingthesymptomsoftheproblem.Inthiscaseathoroughinvestigationwascarriedoutby the researchers, anda surveywas chosenas the re‐searchstrategy.Thesurveytargetedyoungpeoplewhohadgivenupbook reading in favour of social media interaction, web browsing,
61
watchingtelevisionorotherkindsofmediaconsumption.Fivehun‐dred randomly selected persons between the age of 15 ‐ 25 wereaskedabouttheirmediapreferences.Thetenpercentleastcommit‐ted to book readingwas then selected for a second questionnaire.This questionnaire consisted mainly of open questions, where therespondentswere encouraged to explainwhy they had abandonedreading books. The answerswere analysed using content analysis,andanumberofrootcauseswere identified.Finally,media theory,inparticularcriticalmediatheory,wasusedtoidentifyfurthercaus‐es for the current decline in reading. The most important causesidentifiedwerethefollowing:
Paperbooksareawkwardtohandle Itiswearisometoreadlongtexts Booksarenotinteractive Booksandbookcontentscannoteasilybeshared Bookcontentsarenoteasilycombined Manypeopleseepublishinghousesaspromoting theirown
interests
TheExplicateProblemactivityissummarizedinfigure5.3.
Figure 5.3 Explicate Problem in the running example
62
Further Reading
AclassictextonproblemanalysisandsolvingwaswrittenbyPolya(1973),althoughitisquitemathematicallyoriented.Ritchey(2007)givesaneasilyaccessibleintroductiontowickedproblems.Ishikawadiagramsaredescribedon thewebpageCause andEffectAnalysis(2012).
63
6 Outline Artefact
and Define Requirements
ThesecondactivityofthemethodisOutlineArtefactandDefineRe‐quirements. The goal is to identify andoutline an artefact that canaddresstheexplicatedproblemandtodefinerequirements.Inotherwords,theactivityaddressesthequestion:
What artefact can be a solution for the explicated problem
and which requirements on this artefact are important for
the stakeholders?
Answeringthisquestioncanbeviewedasanextendedproblemex‐plication.Inotherwords,inthisactivity,researcherswillcontinuetoexplicatefurthertheproblem,buttheywilldosousingtheproposedsolutionoutlineasapairofglassesforguidingtheirproblemexami‐nation. Thus, the question is to be answered by descriptiveknowledge that specifies which requirements on the artefact areimportantforthestakeholders.
The requirementswill primarily address the function and con‐structionoftheartefactbutalsotherelationshipstoitsenvironment.Functionalrequirementsdependontheproblemtobeaddressed,as
64
wellastheneedsandwantsof thestakeholders.Someexamplesoffunctionalrequirementsforanelectronichealthrecordsystemcouldbe to provide storage of X‐rays, to enable doctors to enter infor‐mation thatwill be unavailable to patients, or to allow patients toenterinformationontheirself‐medication.Ascanbeseenfromthisexample,functionalrequirementsareoftenveryspecifictothesitua‐tion at hand. In contrast, requirements pertaining to constructionandenvironmentareoftenmoregeneric.Examplesonconstructionrequirements for the health record system could be that it shouldhave a coherent andmodular design. Examples onenvironment re‐quirementscouldbethatthesystemshouldbeavailableondifferentplatforms and be easy to adapt to changes.Non‐functional require‐ments is a termencompassingboth construction requirements andenvironmentrequirements.
Figure 6.1 Outline Artefact and Define Requirements
The activity Outline Artefact and Define Requirements can bestructuredandvisualizedusingtheIDEF0technique,seefigure6.1.Theinputisanexplicatedproblemprovidedbythepreviousactivity.Theoutput isanartefactoutlineandasetofrequirements.There‐sources used by the activity consist of knowledge in the researchliterature and other written sources, as well as assertions from
65
stakeholders. The controls are primarily research strategies andmethods but may also include practice‐based approaches to re‐quirementselicitationandanalysis.
6.1 Sub‐activities of Outline Artefact and Define Re‐
quirements
The activity includes two main sub‐activities: outline artefact anddefinerequirements.
Outline Artefact
The sub‐activity outline artefact starts by choosing which artefacttypeshouldbedesignedinordertosolvetheproblem,i.e.choosingwhetherthesolutionshouldbeaconstruct,amodel,amethodoraninstantiation.Thischoiceissometimessimpleduetothecharacteris‐ticsoftheexplicatedproblem.Inothercases,itcanbemoredifficulttochoosetheartefacttypetobedesigned.Forexample,ifITsystemsneed to be integrated tomake a processmore efficient, a solutioncanbeamethodforintegratingITsystems,amodelofanintegrationarchitecture, or an instantiation in the form of an integration tool.When the artefact type has been chosen, the artefact is to be de‐scribedonanoverviewlevel.
Define Requirements
Thesecondsub‐activityistodefinetherequirementsontheoutlinedartefact.Therequirements to includedependonthecharacteristicsof the problem, the outlined solution,technological opportunities,previous research including documented solutions to similar prob‐lems,andstakeholderinterestsandopinions.
Should requirements be formulated in terms of a solution?
Short answer: No
Long answer: Consider the following two requirements for an electronic health record system:
66
1 Physicians should be able to search for health records based on any patient characteristic 2 Physicians should be able to search for health records based on any patient characteristic using a relational database management system
Which requirement is to be preferred? The second one is more specific and, therefore, may seem to be preferable. However, this requirement restricts the designer to one single solution, a relational database man‐agement system, even if this would not be the best way of providing search capabilities to physicians. In general, including solutions in re‐quirements formulations restricts the designer in devising the best pos‐sible solution. Therefore, requirements should not be expressed in terms of specific solutions.
6.2 Resources for Outline Artefact and Define Require‐
ments
Theresultsoftheactivityshouldalwaysbebasedonandrelatedtoexistingwork.Theresearcherneedstoreportonpreviousresearchthathasbeen carriedout to solve similarproblems,what artefactshave been designed and developed in that research, and what re‐quirementshavebeenaddressed.Evenifnootherartefacthasbeendesignedpreviouslytosolvethesameproblem,orfocusonthesamerequirements as in the current work, the researcher should stillspecifywhethersimilarsolutionsexistandinwhatwaystheydifferfromtheproposedartefact.Thereby,therequirementswillbegivena context that supports both researchers and practitioners in as‐sessingtheoriginalityandsignificanceoftheartefact.
Anotherbasisfortheactivityconsistsoftheinterestsandopin‐ions of stakeholders. An investigation to determine these may in‐cludemanydifferentgroupsofstakeholders inoneorseveral localpractices. The stakeholders may hold differing opinions about therequirements, including their relative importance. Therefore, it isoften worthwhile to ask the stakeholders to rank suggested re‐quirementswithregardtoimportance.
67
6.3 Research Strategies and Methods for Outline Arte‐
fact and Define Requirements
Research Strategies for Outline Artefact and Define Requirements
Surveys
Surveys canbeused for eliciting requirementsdirectly fromstake‐holders.Theyoffera relatively low‐costapproach, in termsof timeas well as other resources, to identifying requirements. Therefore,surveysmakeitpossibletoinvestigatetheneedsandwantsofmanypeople and different kinds of stakeholders, such asmanagers, em‐ployees, end‐users, and customers. This comprehensive coverageincreases the likelihood of finding all potentially relevant require‐ments and ensures that all stakeholder groups’ voices are heard.However, there is a risk that surveys will result in incomplete re‐quirements when stakeholders are not prepared to spend a suffi‐cient amountof timeandeffortonproviding information. Further‐more, the stakeholdersmaymiss identifying requirementsbecausetheyhavelimitedknowledgeorarebiasedinvariousways.Inotherwords, surveys as an instrument for requirements elicitation aredependent on the commitment, knowledge and interests of thestakeholders.
Case studies
Case studies can overcome some of the limitations of surveys forrequirementselicitation. Inparticular,casestudiesofferopportuni‐tiesforinvestigatingstakeholderneedsandrequirementsaswellastheirpracticeingreaterdepthoveranextendedperiodoftime.Fur‐thermore, a researcher can identify requirements even if a stake‐holder does not explicitly state them. The deeper understandingprovidedbyacasestudyandthereduceddependencyonstakehold‐erscanresultinmorecompleteandrelevantrequirements.Howev‐er,casestudiesaretime‐consuming,theiroutcomesdependheavilyonthecompetenceoftheresearcherandtheycanalsobebiasedbyher interests and preconceptions. Furthermore, a case study is al‐
68
wayscarriedoutinasinglelocalpractice,whichcanlimitthegener‐alisabilityoftheresults.
Action research
The advantages and disadvantages of action research for require‐ments elicitation are similar to thoseof case studies.However, thedependenceontheresearcherisevengreaterinactionresearchthanin case studies, as she is to actively participate and intervene in apractice.Thereisariskthatsheforcesherownviewsonthestake‐holders and comes upwith requirements that are not particularlyrelevantforthem.Ontheotherhand,theresearchermaybeessen‐tialforsuggestingrequirementswhenherknowledgeexceedsthatofthestakeholders.Thisisoftenthecasewhenanovelartefactistobedesignedandintroducedinapractice,anditsstakeholdershaveonlyalimitedunderstandingofthepotentialofthenewartefact.There‐searcher can then inform and guide the stakeholders so that theybecomeabletoidentifyandarticulaterelevantrequirements.
Theoretical analysis
Insomesituations,itmaynotbefeasibletocarryoutempiricalstud‐iesforelicitingrequirements,andinsteadtheoreticalanalysishastobeused.Suchsituationsprimarilyoccurwhentheartefacttobede‐signedistrulynovel,anditisstillunclearhowtouseitinpractice.Atheoretical analysis may then suggest preliminary requirementsbasedonpossibleusagescenarios.Anobviousdrawbackofthisap‐proach is that itmaybecomehighly speculative anddependent onthe competence, imaginationandpreconceptionsof the researcher.Thus,therequirementsidentifiedthroughatheoreticalanalysisareusually notwell‐founded, but they can still function as a first steptowardsdesigningahighlynovelartefact.
Research Methods for Outline Artefact and Define Requirements
Interviews
Interviewsmaybethemostcommonmethodforgatheringrequire‐ments. Interviews usually take a direct approach to requirements
69
elicitationbyaskingstakeholdersaboutfeaturesthattheywouldliketo see included in the outlined artefact andwhich explicit require‐mentstheysuggest.Interviewscanbehighlyefficientandresultinalarge number of requirements in a short time. However, theymayeasilybecomestaleandstiflecreativity,whichisobviouslycounter‐productive for identifyingrequirements.Tosomeextent, thisprob‐lemcanbereducedbyusingunstructuredinterviewsthatencouragetherespondenttotakemoreinitiative.Anotherissueisthataninter‐viewmayonly be effective if the respondent is competent and ap‐pliessufficienttimeandeffort,whichmaynotalwaysbethecase.Inadditiontodirectlyaskingforrequirements,interviewscanbeusedforincreasingtheunderstandingoftherespondent’spracticeaswellasherattitudetotheartefactunderconsideration.
Group discussions
Groupdiscussionscanovercomesomeofthedisadvantagesofinter‐viewsfordirectrequirementselicitation.Whenpeoplemeet,discussandbrainstormingroups, thechances increasethatmore imagina‐tive and creative requirementswill be suggested.This is especiallysoifpeoplewithdifferentbackgroundsandcompetencesareinclud‐ed in thesamegroup; theywillbeable tosurprise, inspireanden‐courageeachother to comeupwithnovel requirements.However,asalwaysingroupdiscussions,thereisariskthatoneorafewper‐sonswill dominate,whichmay reduce thewillingness of others tooffer suggestions for requirements. A moderator who makes surethateveryonegetsachancetocontributecanmitigatethistosomeextent.
Questionnaires
Questionnaireshave the sameadvantagesanddisadvantages as in‐terviewsbutoftentoanevenhigherdegree.Questionnairesare in‐expensive to distribute,whichmakes it possible to get suggestionsforrequirementsfrommanystakeholders.However,fillinginaques‐tionnaire does not invitemuch creativity,meaning that innovativerequirements rarely canbe expected.On theotherhand, question‐
70
naires canbe effective for ranking the importanceof requirementsthathavealreadybeenidentifiedthroughothermethods.
Observation studies
Observation studiescanbeused forgettingabetterunderstandingofthepractice inwhichanartefactwillbeused,whichcanprovidecluesforadditionalrequirements.Inthisway,observationisusedasameansforaresearchertogeneraterequirementsherself.Observa‐tionstudiesareeffectivewhenitcannotbeexpectedthatthestake‐holders themselves are able to generate all relevant requirements,typicallybecausetheyhavea limitedunderstandingof theoutlinedartefact.Thevalueof the requirementsproduced throughobserva‐tionstudiesreliesheavilyonthecompetenceoftheresearcher.
Document studies
Documentstudiescancomplementothermethodsforrequirementselicitation,orbeanalternativewhenaccess to stakeholders is lim‐ited.Documentstudiescan improve theunderstandingof theprac‐ticeunderconsideration,whichmayprovidecluesforrequirements,analogouslytoobservationstudies.
6.4 Guidelines for Outline Artefact and Define Require‐
ments
Specifywhatartefacttobuild!Specifytheartefacttype(con‐struct,model,method, instantiation)and itsgeneralcharac‐teristics.
Formulate each requirement clearly! Describe each require‐mentinaprecise,conciseandeasilyunderstandableway.
Justifyeachrequirement!Foreachrequirement,explainwhyitisneededandrelateittotheproblem.
Berealisticbutalsooriginal!Ensurethatit isrealistictode‐velopanartefactfulfillingtherequirementsbutalsotrytobeoriginal.
71
Specify the sources of the requirements! Describe the litera‐tureand the stakeholders thathave contributed todefiningtherequirements.
Describe how you have defined the requirements! Explainwhatyouhavedonetodefinetherequirements,inparticularhowyouhavereviewedthestakeholdersandresearchlitera‐ture.
6.5 Properties for Generic Requirements
Mostrequirementsonanartefactaresituationspecific,inparticularthoseconcerning its function.However, therearealsoanumberofgeneric requirements that are relevant for almost any artefact, e.g.thatitshouldbeeasytouseorhaveamodulardesign.Thissectionintroduces a set of properties that are useful when defining suchgenericrequirements.Basedonthedistinctionbetweenconstructionandenvironment,thepropertiesaredividedintointernalpropertiesthatdescribetheinnerstructureoftheartefact,andexternalproper‐tiesthatdescribetherelationshipbetweentheartefactanditsenvi‐ronment. Furthermore, theexternalproperties aredivided intous‐agepropertiesthatdescribehowtheartefactworksandisperceivedinusesituations;managementpropertiesthatdescribehowthear‐tefact is managed over time; and generic external properties thatmainlydescribehow the artefact is structurally related to its envi‐ronment.
The suggestedproperties canbeusedas inspiration for formu‐latinggenericrequirementsinanearlystageofadesignsciencepro‐ject.Theycanalsobeusedasvalidationsupportofalreadyformulat‐ed requirements. Thereby important butmissed requirements canbeidentified.
Internal Properties
Internalpropertiesdescribetheinnerstructureofanartefact.
Coherence ‐thedegree towhich thepartsof anartefactarelogically,orderlyandconsistentlyrelated.Coherenceislowif
72
an artefact includes parts that, in some sense, do not fit inwiththerestoftheartefact.
Consistence (only formodels) ‐thedegreetowhichamodelisfreefromconflicts.
Modularity ‐ thedegree towhich an artefact is divided intocomponents that may be separated and recombined. Com‐mon requirements related to modularity are low coupling,thatmodulesarenotoverlyrelatedwitheachother;highco‐hesion, thatmodules arehighly related internally; andhighcomposability, thatmodules can be easily replaced and re‐combined.
Conciseness‐theabsenceofredundantcomponentsinanar‐tefact,i.e.componentsthefunctionsofwhichcanbederivedfromothercomponents.
Elegance ‐ the degree to which an artefact is pleasing andgracefulinappearanceorstyle.
External Properties
External properties describe the relationships of an artefact to itsenvironment,forexampletousersorotherartefacts.
Usage
Usagepropertiesdescribehowtheartefactworksandisperceivedinusesituations.
Comprehensibility ‐ the ease with which an artefact can beunderstood or comprehended by a user (also called under‐standability).
Learnability‐theeasewithwhichausercanlearntouseanartefact.
Usability ‐ theeasewithwhichausercanuseanartefact toachieveaparticulargoal.
Customisability ‐the degree to which an artefact can beadaptedtothespecificneedsofalocalpracticeoruser.
73
Suitability ‐ the degree towhich an artefact is tailored to aspecificpractice, focusingonlyon its essential aspects (alsocalledinherenceorprecision).
Accessibility‐thedegreetowhichanartefactisaccessiblebyasmanyusersaspossible.
Traceability(onlyformethods)‐theabilitytoverifythehis‐toryofusingamethodbymeansofdocumentation.
Management
Management properties describe how an artefact ismanaged overtime.
Maintainability‐theeasewithwhichanartefactcanbemain‐tained in order to correct defects, meet new requirements,makefuturemaintenanceeasier,orcopewithachangeden‐vironment.
Flexibility ‐ the easewithwhich an artefact can be adaptedwhenexternalchangesoccur(similartomaintainability;re‐latednotionsareconfigurability,evolvabilityandextensibil‐ity).
Accountability‐theeasewithwhichanactorcanbemadeac‐countablefortheworkingsofanartefact(asimilarnotionisauditability).
Generic
Generic propertiesmainly describe how the artefact is structurallyrelatedtoitsenvironment.
Expressiveness(onlyforconstructsandmodels)‐thedegreetowhicha setof constructsoramodel is capableof repre‐sentingtheentitiesofinterestinadomain.
Correctness(onlyformodels)‐thedegreetowhichamodelcorresponds to the domain it represents (also called accu‐rateness).
Generality ‐ the degree towhich an artefact is relevant notonlyforalocal,butalsoforaglobalpractice.
74
Interoperability(onlyforinstantiations)‐theabilityofanar‐tefact toworktogetherwithotherartefacts, inparticular toexchange data (related notions are openness, compatibilityandstandardscompliance).
Autonomy(onlyforinstantiations)‐thecapacityofthearte‐facttofunctionwithouttheinvolvementofanothersystem.
Proximity(onlyformodels)‐thedegreetowhichindepend‐entaspectsofadomainarecapturedbydifferentconstructs,andrelatedaspectsarerepresentedbyrelatedconstructs.
Effectiveness ‐ the degree to which an artefact is able toachieveitsgoals(aspecialcaseiscompleteness).
Efficiency ‐ thedegreetowhichanartefact iseffectivewith‐outwastingtime,effortorexpense.
Robustness(onlyforinstantiations)‐theabilityofanartefacttocopewithfailures,errorsandotherproblemsduringexe‐cution (related notions are degradability, survivability andsafety).
6.6 Running Example
ThesecondDIGIREADactivitywastooutlineanartefactanddefinetherequirements.Firsttheartefactwasoutlined.Thismeantdecid‐ing artefact type and the basic characteristics of the artefact. Theresearchproject came to the conclusion that a possible solution tothe problem could be the introduction of a novel service that pro‐vides interactive and collaborative book reading usingmultimedia.Thus,theartefactwasaninstantiationintheformofaservice,whichwillprovideinteractiveandcollaborativebookreadingusingmulti‐media.
In order to define the requirements of the artefact, a survey ofuserneedswasinitiated.Afocusgroupwassetuptogeneratedata,whichincludedfrequentbookreadersaswellasnon‐readers.Ausa‐bility and a social media expert also participated. The group mettwiceandproducedanextensivelistofpossiblerequirements,mostofwhichpertainedtothefunctionalityoftheservice.Emergingtheo‐
75
ries about social media were also investigated to help refine theidentified requirements. Existing services for book sharing werestudied, e.g. Amazon and Rethink Books. In order to prioritize therequirements, a questionnaire was distributed to the same peoplewhohadparticipatedintheproblemexplicationstudy.Examplesofthemosthighlyprioritizedrequirementsfortheservicewere:
It should be possible to share book snippets over variouschannels(functionalrequirement)
Theownerofabookshouldbeabletolendittoothersusingtheservice(functionalrequirement)
It shouldbepossible to read the first chaptersofbooks forfree(functionalrequirement)
Itshouldbepossible towriteandsharebookreviews inde‐pendentlyofanybookseller(functionalrequirement)
Theserviceshouldsupportsearchinganddiscoveringbooksbasedonreviews(functionalrequirement)
It should be possible to include othermedia, such as audioandvideo,inthetextflow(functionalrequirement)
It should be possible to move seamlessly between devices(environmentalrequirement,usabilityandinteroperability)
Theserviceshouldintegratewithsocialmediaservicessuchas Facebook, Twitter and Google+ (environment require‐ment,interoperability)
The service should be independent of platform and easy toadapt to mobile platforms such as Android and iOS (envi‐ronmentrequirement,interoperability)
Theserviceshouldbeeasytouse(environmentrequirement,usability)
The Outline Artefact and Define Requirements activity is summa‐rizedinfigure6.2.
76
Figure 6.2 Outline Artefact and Define Requirements in the example
Further Reading
Requirementselicitationhasforalongtimebeenstudiedintheareaof requirements engineering. An established textbook in the areawaswrittenbySommervilleandSawyer (1997).A recent textbookonrequirementsengineeringwaswrittenbyPohl(2010).
77
7 Design and Develop Artefact
ThethirdactivityofthemethodisDesignandDevelopArtefact,thegoalofwhich is tocreateanartefact fulfillingtherequirementsde‐rived from the previous activity. This includes designing both thefunctionality and construction of the artefact. In other words, theactivitycanbedescribedas:
Create an artefact that addresses the explicated problem
and fulfils the defined requirements
Theresultof thisactivitywillprimarilybeprescriptiveknowledge,whichcanbeembeddedinthecreatedartefact,seeSection2.2.Fur‐thermore,theactivitywillproducedescriptiveknowledgeaboutthedesigndecisionstakenandtheirrationale.
TheactivityDesignandDevelopcanbestructuredandvisualizedusing the IDEF0 technique, see figure 7.1. The input is an artefactoutlineandasetofrequirementsasprovidedbythepreviousactivi‐ty. The output is an artefact fulfilling these requirements andknowledge about it. The resources used by the activity consist ofknowledgefromtheresearchliteratureandotherwrittensourcesaswellasknowledgeembedded inartefactsandassertions fromrele‐vantstakeholders.Thecontrolscan includeresearchstrategiesand
78
methods,andmayalsoincludeanypractice‐baseddevelopmentap‐proaches.
Figure 7.1 Artefact Design and Development
7.1 Sub‐activities of Design and Develop Artefact
TheactivityDesignandDevelopArtefactincludestwosub‐activities.The first, Generate, produces new possible solutions,while the se‐condone,Searchandselect,evaluatesandselectsfromthegenerat‐edsolutions.Thus,DesignandDevelopArtefactcanbeviewedasasearchprocessthroughasolutionspace(Simon,1996).Accordingtothis view, the designer startswith setting up a solution space con‐sistingofpossiblesolutionsandthenpruningit.Inpractice,thesub‐activitiesarecarriedoutinparallelanditeratively.
Generate
Generatinginnovativeandusefuldesignsolutionsistosomeextentalways dependent on luck. However, as Louis Pasteur observed in1854, “Chance only favours the preparedmind”. Themind can be‐comepreparedthroughexploratorystudiesofpotentialusersoftheartefactintheirpractices,usingactionresearch,andethnographyaswellasnaturalisticobservation.
79
There also exist several practice based approaches for solutiongeneration.Oneof them isempathetic thinking inwhichadesignertries “to see theworld through the eyes of others, understand theworld through their experiences, and feel the world through theiremotions” (Brown, 2009). Empathetic thinking can help to expandthedesigner’smind, therebysupporting the creationofnovel solu‐tions.AnotherapproachforgeneratingsolutionsisdeBono’slateralthinking(deBono,1973),whosuggestsanon‐traditionalwayofrea‐soning. Instead of focusing on logical step‐by‐step arguments, deBono suggests a spectrum of practices for generating fresh ideasincludingrandomgeneration,provocativegenerationsuchaswish‐ful thinkingandexaggeration,challengegenerationwhereanythingcanbequestioned,anddisprovalofgenerallyacceptedtruths.
One established instrument for generating ideas in groups isbrainstorming,wheretheparticipantsaimtoproduceasmanyideasaspossible,attempt towithholdcriticism,encouragenewandunu‐sual ideas, and try to integrate and improve ideas that have beenproposed.Brainstormingisagenericinstrumentforideagenerationthathasbeenappliedinmanyareas.Forexample,agoalofabrain‐stormingsessioncouldbetoidentifyimportantactivitiesinameth‐odformanagingrisksinagilesoftwaredevelopmentprojects.First,thegroupcould identifypossibleactivitiesbytryingtosuggestandwritedownasmanyaspossiblewithoutanycriticism.Thenextstepcouldbetoeliminateduplicatedactivitiesororganisesmalleractivi‐tiesintolargerones.Finally,thegroupcanprioritiseamongthesug‐gested activities in order to select the most useful ones, therebymovingtothesearchandselectsub‐activity.
IntheareaofITandinformationsystems,therearealsoseveralother more specialized instruments, like participative modelling,walk‐throughsandpairdesign.Participativemodellingresemblestosomeextentbrainstormingbutusesspecifictechniquesforexpress‐ingideas,suchasgoal,processandconceptualmodels.Italsomakesuse of certain tools for visualization and cooperativemodel devel‐opment,e.g.whiteboards,plasticsheetsandpost‐its.Awalk‐through
80
isakindofpeerreview,whereadesignerleadsmembersofadevel‐opmentteamandotherstakeholdersthroughanartefact, forexam‐ple amethod. The participantsmay ask questions, point out prob‐lems,suggestalternativesolutions,orprovidefeedbackinanyotherform.Pairdesign,basedontheagilepracticeofpairprogramming,isawayofworkingwhere twodesignerswork togetherandsimulta‐neouslydevelopanartefact.Typically,onedesignermakesadesigndecision,e.g.introducesanactivityintoamethod,andtheotherde‐signer immediately reviews it and suggestspossible improvementsorpotentialproblems.Theinstrumentsintroducedherearecertain‐lynotexhaustivebut theyprovidearepresentativesampleof tech‐niquesusedforsupportingcooperativedesignwork.
Search and Select
When a number of solutions have been generated, the designermakesafirstdesigndecision,therebypruningawaypartsoftheso‐lution space. She continues andmakes additional design decisions,which further narrows the space. The process goes on and can beviewed as a systematic exploration of the solution space. At eachstepthedesignerisguidedinherdecisionmakingbysomecriteriaand the requirementson theartefact.Theguidancecan sometimestaketheformofagoodnessorutilityfunction,whichcanbeusedtodeterminewhichdesigndecisionstomake.SomeauthorshaveevenarguedthatthesearchprocesscanbemoreorlessautomatedusingAI techniques (Simon, 1996). Though this has not proved feasible,the view of design as a search process is useful in supporting thecommunication, documentation, planning and structuringof designwork.
Generatingsolutionsrequiresanotherkindofthinkingthanthatneeded for selecting from given solutions. The difference is some‐timesexpressedinthenotionsofconvergentanddivergentthinking.Convergentthinkingindesignaimsatdecidingamongexistingalter‐natives (sub‐activity Search and select),whiledivergent thinking isabout generating new alternatives from which to choose (sub‐
81
activityGenerate).Inotherwords,divergentthinkingcreateschoicesin a solution space and is highly innovative and imaginative,whileconvergentthinkingmakeschoices,whichcanbemoreanalyticandrational.
7.2 Resources for Design and Develop Artefact
Designinganddevelopinganartefactisacombinationofreusingandadaptingcomponentsfromexistingsolutions,inventingnewcompo‐nents,andcombiningtheminaninnovativeway.Therefore,relatedsolutions need to be analysed by studying previous work, both inresearchliteratureandartefactsusedinpractice.Eveniftheartefactto be designed is highly original, the researcher should relate andcompareittoexistingsolutions.
Anotherbasisforthisactivityconsistsofideasandopinionsgen‐erated by stakeholders. While stakeholder input is almost alwaysessential for defining requirements, its value for design and devel‐opment issometimes limited.Stakeholdersmaynotbeable tocon‐tributeiftheartefactbeingdesignedhasahighlycomplexconstruc‐tion. However, if the artefact has a simpler structure, experiencedusers may provide significant contributions, as witnessed by thephenomenonofOpeninnovation(Chesbrough,2005).
Regardlessoftheresourcesusedincreatingadesign,itisvalua‐bletodocumentthedesignrationale,i.e.alistingofandargumenta‐tion about the decisionsmade during the design process. A designrationaleshouldcontainthereasonsandjustificationsbehinddesigndecisions,alternativedecisionsconsidered,andtheargumentslead‐ingto thedecisions.Adesignrationalecansupportcommunicationduringasingleproject,but it isalsouseful forfacilitatingreusebe‐tweendifferentprojects.Infact,adesignrationalecanbeoneofthemostvaluableoutcomesofadesignscienceproject,asitwillrecordthe reasoning behind design decisions including potential pitfalls.Thisknowledgecanbeofgreatvaluetosubsequentprojects,inpar‐ticular itmayhelp themavoiddead ends andother kindsof prob‐lems.
82
7.3 Research Strategies and Methods for Design and
Develop Artefact
Design and Develop Artefact differs from the other design scienceactivities in that it does not primarily aim to answer questions byproducing descriptive or explanatory knowledge. Instead, its mainpurpose is to produce prescriptive knowledge by creating an arte‐fact.Therefore, researchstrategiesandmethodsare less importanthere than in theotheractivities.However, thisdoesnot imply thatresearchmethods arewithout value for developing an artefact. Onthecontrary,interviews,observationstudiesandotherdatagenera‐tionmethods can be highly effective in producing ideas for designsolutions. The point is rather that it is not critical that researchmethods are used for coming up with possible solutions; any ap‐proachforgeneratingsolutionsisadmissibleaslongasitworks.
7.4 Guidelines for Design and Develop Artefact
Clearly describe each component of the artefact! Describeboththefunctionalityandconstructionofeachartefactcom‐ponent.
Justifyeachcomponentoftheartefact!Explainthepurposeofeachartefactcomponent,inparticularwhichrequirement(s)itaddresses.
Describetheuseoftheartefact!Describehowtheartefactanditscomponentsareintendedtobeusedinitsintendedprac‐tice.
Clarifytheoriginality!Describeinwhatrespectstheartefactisdifferentfromexistingoneswithrespecttobothfunction‐alityandconstruction.
Specifythesourcesoftheartefactdesign!Describethelitera‐ture and the stakeholders that have contributed to compo‐nentsoftheartefactand/orinspiredthedesignofnewcom‐ponents.
83
Describehow youhavedesignedanddeveloped theartefact!Explainwhatyouhavedonetodesignanddevelopthearte‐fact, inparticularhowyouhave reviewed the stakeholders,existingsolutions,andresearchliterature.
7.5 Running Example
Thethirdactivity inDIGIREADwastodesignanddevelopthearte‐fact.Thisactivityconsistedofgeneratingandselectingsolutions.
Theprojectfirstgeneratedalternativesolutionstowardsprovid‐ing interactive and collaborative book reading using multimedia.Thiswasdonebydesigningasetofservicefunctions,theiruserin‐terface (UI) representations in the form of UI components and anoverall architectureof the service. Thiswas carriedoutbya smallgroupofsoftwareengineersandthesameusabilityexpertwhopar‐ticipated in the requirements definition activity. The group madefrequentuseofpairdesigntogeneratetheservicefunctions,theUIcomponentsandthearchitecture,andwasclosely intertwinedwiththe requirements definition activity. As a knowledge base for theactivity,thegroupmadeuseofsoftwareengineeringtechniquesandprinciplesforserviceorientedarchitectures.
Thesecondsub‐activitywastoselectoneofthealternativesolu‐tions generated, for further development. i.e., the service functionsanduserinterfacecomponents.Thiswassupportedbypeerreviewsintheformofwalk‐throughs.Thegroupdecidedtodesignalayeredarchitecturefortheservice.Thereasonforthischoicewasimprovedflexibilityandmaintainability,thoughperformancecouldsuffer.TheDesignandDevelopactivityissummarizedinfigure7.2.
84
Figure 7.2 Design and Develop Artefact in the running example
Further Reading
Many good books exist on the art of designing artefacts, e.g. thosewrittenbyBrown(2009)andCross(2011),whileaclassical text is(Simon, 1996). The relationships between design, explanation andexplorationarediscussedbyGibbonsandBunderson(2005).
85
8 Demonstrate Artefact
The fourth activity of the method is Demonstrate Artefact, whichaims to demonstrate the use of the artefact in one case, therebyproving its feasibility. In other words, the activity addresses thequestion:
How can the developed artefact be used to address the ex‐
plicated problem in one case?
The answer to this question will primarily consist of descriptiveknowledgedescribinghow theartefactworks inone case, but alsoexplanatoryknowledgeexplainingwhytheartefactworks.
Ademonstration shows that theartefact in fact can solve someaspects of a problem in one illustrative or real‐life case. A demon‐strationcanbeseenasaweakformofevaluation;iftheartefactcansolvesomeaspectsofaprobleminonecaseitmightbeabletodosoinother cases aswell. Furthermore, ademonstration canalsohelpcommunicate the ideabehind theartefact toanaudience inavividandconvincingway.
TheactivityDemonstrateArtefactcanbestructuredandvisual‐izedusingtheIDEF0technique,seefigure8.1.Theinput isanarte‐factprovidedbythepreviousactivity.Theoutputisademonstrated
86
artefactincludinginformationontheworkingsoftheartefactinonecase. The resources used by the activity consist of domain specificknowledgeabouttheartefactanditsenvironment.Thecontrolswillvaryfromcasetocase.
Figure 8.1 Artefact Demonstration
8.1 Sub‐activities of Demonstrate Artefact
The first sub‐activity is to choose or construct a case onwhich toapplytheartefact.Thiscasecanbeafictitiousonedevelopedbytheresearcherswhodesignedtheartefact,awell‐documentedcasefromliterature,areallifecase,oracombinationofthese.Casesfromreallifetypicallyprovidebetterexternalvalidity,butfictitiouscasescansometimesbepreferableastheycanbedesignedtodemonstratetheviability of the artefact under extreme conditions.The secondsub‐activity is to apply the artefact to the chosen case, which includesdocumentingtheoutcomeoftheapplication.
8.2 Resources for Demonstrate Artefact
Theresourceneededforthisactivityisprimarilyknowledgeregard‐ingthecaseinwhichtoapplytheartefact.
87
8.3 Research Strategies and Methods for Demonstrate
Artefact
Aresearchstrategyisdecideddependingonthecasechosenandthecharacteristicsoftheartefact.Clearly,actionresearchandcasestudyare natural choices when a real life case is used. Experiments areusefulforfictitiousaswellasliteraturecases.
8.4 Guidelines for Demonstrate Artefact
Justifythechoiceofcase!Explainwhythechosencaseisrep‐resentative of the problem and challenging enough to offeranadequatetestbed.
Make clearhowmuchof theartefact is tested!Describe thecomponentsandaspectsoftheartefactthatareactuallyusedinthedemonstration.
8.5 Running Example
Thedemonstrationofthedesignedanddevelopedserviceforinter‐activeandcollaborativebookreadingintheDIGIREADprojectcon‐sisted of two sub‐activities: choose or design a case and apply theartefacttothecase.
Theresearchersdecidedtodesignacaseintheformofanexper‐iment.Thecasedesignincludedanumberoftaskstobecarriedoutbyusers,includingwritingreviews,sharingbooksnippetsviaTwit‐ter,discoveringnewbooksbasedonreviews,andswitchingbetweenreadingandlisteningtothesametext.
Tenpresumptiveusersthencarriedouttheexperimentoverthecourseof a day.Theuserswere given access to aprototype of theservice,whichprovidedmorethan90%oftherequiredfunctionalityand included approximately 100 books. The users carried out thetasks described above and the researchers then logged all serviceinteractions and analysed them using quantitative methods. Theexperiment functioned as a proof‐of‐concept showing that the ser‐
88
vice couldactuallybeusedas intended. TheDemonstrateArtefactactivityissummarizedinfigure8.2.
Figure 8.2 Demonstrate Artefact in the running example
89
9 Evaluate Artefact
ThefifthactivityisEvaluateArtefact,whichaimstodeterminehowwelltheartefactisabletosolvetheexplicatedproblemandtowhatextent it fulfils the requirements. In other words, the activity ad‐dressesthequestion:
How well does the artefact solve the explicated problem
and fulfil the defined requirements?
The answer to this question will primarily consist of descriptiveknowledge,butmayalsoincludeexplanatoryknowledgeexplainingwhytheartefactisabletosolvetheproblem.
The activity Evaluate Artefact can be structured and visualizedusing the IDEF0 technique, see figure 5.1. The input is an artefactprovidedbythepreviousactivity.Theoutputisanevaluatedartefactincludinginformationonhowwelltheartefactworksandwhy.Theresources used depend on the evaluation strategy and can includeexpertsaswellassitesforcasestudiesoractionresearch.Thecon‐trolscan includeanyresearchstrategyormethod thatcanbeusedforevaluation.
90
Figure 9.1 Evaluate Artefact
9.1 Sub‐activities of Evaluate Artefact
TheactivityEvaluateArtefact includes two sub‐activities.The first,Chooseevaluationstrategy,determineshowtheevaluation is tobecarried out, while the second, Carry out evaluation, actually per‐formstheevaluation.
Choose evaluation strategy
There are twomain strategies: ex ante and ex post evaluation. Exante evaluationmeans that the artefact is evaluatedwithout beingused,whileexpostevaluationrequirestheartefacttobeemployed.
Anexanteevaluationoftenmakesuseof interviews,whereex‐perts express their views on an artefact. The experts base theseviews on their general knowledge aswell as experience of similarartefactsandtheirapplications.Forsomerequirements,e.g.,modu‐larity and coherence, an ex ante evaluation can provide both validandreliableresults.Forotherrequirements,e.g.,usabilityandeffec‐tiveness, theviewsof theexpertscanberatherspeculative. Ingen‐eral,themaindrawbackofanexanteevaluationisitsheavyreliance
91
onthesubjectivejudgmentsofexperts,whilethemainbenefitisitslowcost.
Anexanteevaluationsometimesconsistsofargumentsprovidedby the researchers who developed the artefact. In this case, theyevaluate it by reasoning and arguing that it fulfils the defined re‐quirementsandcansolvetheexplicatedproblem.Thisformofeval‐uation is sometimes called informed argument. A common line ofreasoningusinginformedargumentistoclaimthattheartefactful‐filsarequirementbecauseithasacertainconstruction.Forexample,itcanbearguedthatanewMMORPGiseasytolearnbecauseitsin‐terfaceissimilartothatofWorldofWarcraft.Informedargumentisobviously aweak formof evaluation, as itmay easily be biasedbythe backgrounds and interests of the researchers. However, in‐formedargument is inexpensiveand isoftenusedwhenevaluatinghighlyinnovativeandstill immatureartefacts.Researchersmayuseinformed arguments at informal presentations or research work‐shops. Based on feedback from other researchers, the artefact canthenberefinedbeforeastrongerformofevaluationiscarriedout.
Whileexanteevaluationsareoftenweak,expostevaluationscanbeconsiderablystronger.Suchevaluationsrequirethatanartefactisactuallyputintooperationbeforebeingevaluated.Incontrasttoexanteevaluations,anexpostevaluationcanpreferablybeemployedtoevaluatetheusabilityofanartefactincludingitscomprehensibil‐ityandlearnability.Expostevaluationsarealsoappropriate for in‐vestigatingtheeffectsanartefactmayhaveonitsenvironment,e.g.howanewERPsystemmay influence thepowerrelationships inaworkplace.
Amainbenefitofexpostevaluationsis,asmentionedabove,thatthey can be employed to evaluate the usability and effects of arte‐facts.Theirmaindrawbackisthelowinternalvaliditywhenanarte‐factisemployedinacomplexenvironment,suchasanorganisation.Insuchacase, it canbenext to impossible todistinguish the influ‐encesof theartefact fromthoseofcontingent factors.Forexample,the alleged effects of an ERP system on power relationships may
92
insteadbeduetoaneconomiccrisis,otherchangeprojects,newem‐ployees,orotherfactors.
Carry out evaluation
Theevaluationistobecarriedoutanddocumentedaccordingtotherequirements of the selected research strategies and methods. Inparticular,thevalidityandreliabilityoftheevaluationaretobedis‐cussed.
9.2 Resources for Evaluate Artefact
Forexanteevaluations,themostimportantresourceisanumberofexpertswho shouldhave competence aswell as the time andwill‐ingnesstoevaluatetheartefact.Obviously,theserequirementsmayneedatrade‐off,asthemostqualifiedexpertstypicallyarethebusi‐estones.Expostevaluationsoftenrequireoneorseveralsiteswherethe artefact can be employed. Getting access to such sites can bechallenging,forexamplewhenanewITsystemistobeevaluatedinanorganisation.
9.3 Research Strategies and Research Methods for Eval‐
uate Artefact
Research Strategies for Evaluate Artefact
Surveys
Surveys can be used for gathering feedback on an artefact from alargenumberofstakeholdersandexperts.Therefore,theresultsofasurveyhave fairchancesofbeinggeneralizable,providedthatsam‐pling and data analysis have been carried out in a rigorous way.However,adrawbackofsurveysisthattheyoftenonlyresultinshal‐low responses from respondents, as they may not be prepared tospendmuchtimeandeffortonansweringsurveyquestions.
93
Experiments
Experiments are popular instruments for evaluating an artefact, astheyallowaresearchertoachievehighinternalvaliditybycarefullycontrollingtheconditionsunderwhichanexperimentiscarriedout.On the other hand, external validity can suffer, since the artificialsettingof theexperimentcanbedecidedlydifferent fromtheprac‐ticeinwhichtheartefactistobeused.
Case studies
Casestudiesallowaresearchertomakeadeepevaluationofanarte‐factandunderstandthereasonsforitssuccessorfailure,atleastinone case. This understanding can aid the researcher in coming upwith improved redesigns of the artefact, which can be used in aniterativedevelopmentprocess.However,casestudiesrequiremucheffort and their outcomes depend heavily on the competence andexperienceof theresearcher.Theycanalsobebiasedbyher inter‐estsandpreconceptions.Furthermore,carryingoutacasestudyononlyasinglesitecanlimitthegeneralizabilityoftheresults.
Ethnography
The research strategy ethnography allows a researcher to under‐standnotonlyhowapractice influences theuseofanartefact,butalso how the artefact may change the practice. Therefore, the re‐searchermayidentify interplaysbetweenartefactandpracticethatmightnot beapparent to itsparticipants.Adrawbackof ethnogra‐phy is that it is time‐consuming and requires highly qualified re‐searcherstobeeffective.
Theoretical Analysis
Theoreticalanalysiscanbeusedtoverify thatanartefact fulfilsre‐quirements on internal properties, like coherence and consistence.Theoreticalanalysis,intheguiseofinformedarguments,canalsobeused to show that less formal requirements are fulfilled. Informedarguments typically start from the construction of an artefact andthenarguethattheseensurethatsomerequirementissatisfied.
94
Research Methods for Evaluate Artefact
Interviews
Interviewsareeffectiveinstrumentsforgatheringstakeholderopin‐ions and perceptions about the use and value of an artefact. Inter‐views also allow a researcher to delve deeper into stakeholderviews, as she can ask follow‐up questionswhen needed. However,results from interviews are always dependent on the perspectiveand competence of the respondent, which must be taken into ac‐countwhen interpreting interviewanswers.Furthermore,respond‐entsusuallywanttobeniceandpolitewhenmeetingaresearcherinperson and assessing her results,meaning that theymaywithholdcriticismandexpressmorepositiveviewsthantheyactuallyhold.
Questionnaires
Questionnairescanbehighlyefficientforgatheringtheopinionsandperceptions of many stakeholders about an artefact. However, theanswersareoftensuperficialanddonotallowaresearchertogainadeepinsightintotheviewsofindividualrespondents.
Observation studies
While interviews and questionnaires are effective tools for under‐standing the subjective views and perceptions of stakeholders, ob‐servations offer researchers an instrument for a more objectiveevaluation. For example, an interview can reveal whether or notsome respondents perceive a new information service as easilylearnable.However,perceivedlearnabilityshouldnotuncriticallybeequatedwithlearnability.Observationsmayverywellshowthattheserviceisnoteasilylearnableeventhoughtherespondentsperceiveitso,orviceversa.
95
9.4 Guidelines for Evaluate Artefact
Evaluateeveryrequirement!Everyrequirement identified inthe activity Outline Artefact and Define Requirements shallbeevaluated.
Evaluatehow theartefactcansolve theproblem! Investigatenotonlyhowwelltheartefactfulfilstherequirementsbutal‐sotowhatextentitcanaddresstheproblem.
Describehow youhave evaluated theartefact! Explainwhatyouhavedonetoevaluatetheartefact,inparticularhowyouhavestudiedstakeholdersusingtheartefact.
9.5 Ethical Consequences
Evaluation of an artefact’s effects, as discussed so far, has focusedsolely on the intendedeffects of using the artefact.However, unin‐tendedeffectsalsoneedtobeconsidered,inparticularwheninvesti‐gating the ethical consequences of an artefact and its use. Ethicalinvestigationscanbecarriedoutinmanydifferentwaysandcanbebasedonbothutilitariananddeontologicaltheories.Evenifthereisnosinglebestwaytoinvestigatetheethicalaspectsofartefactuse,itis still meaningful to ask for practical guidelines that can assist inthis task. The remainder of this section outlines a deontologicallybasedapproachthatcansupportadesignerinidentifyingtheethicalconsequencesofartefactuse.
The first step of the approach aims at finding individuals andgroupsofpeoplethatcanbeaffectedbytheuseofanartefact.Thiscanbedonebyfirst identifyingall thepractices thatare influencedby the artefact. Thiswould not only include the intended practice,seesection1.3,butalsootherpractices,wheretheuseoftheartefactmaycausevarioussideeffects.Withineachpractice,individualsandgroupsthatcanbeaffectedareidentified.
Thenextstepistoinvestigate,foreachindividualandgroup,inwhichrespectsitcanbeaffected.Thiscancertainlybedoneinmanydifferentways,whereoneistoconsiderbasichumanrightsandde‐
96
termine how they can be influenced by the artefact use. A recentproposalforthemostfundamentalhumanrightshasbeenformulat‐edbyMarthaNussbaumintheformofalistofcentralhumancapa‐bilities. Nussbaum’s capability approach is based on the view thatethicalvaluesshouldfocusonhumanfunctioning,i.e.whatpeopledoor are able to do. This is in contrast to utilitarian approaches thatfocusonhumanpreferences,i.e.whatpeopleprefertodo.Nussbaumarguesthatthelatterareunstableandeasilydistorted,implyingthathuman functioning is a safer ground for ethical values. Nussbaum(2007)offersatentativelistofcentralhumancapabilities.
“1. Life.Beingable to live to the endof ahuman life of normallength;notdyingprematurely,orbeforeone'slifeissoreducedastobenotworthliving.
2.BodilyHealth.Beingabletohavegoodhealth,includingrepro‐ductivehealth;tobeadequatelynourished;tohaveadequateshelter.
3.BodilyIntegrity.Beingabletomovefreelyfromplacetoplace;tobesecureagainstviolentassault,includingsexualassaultanddo‐mesticviolence;havingopportunitiesforsexualsatisfactionandforchoiceinmattersofreproduction.
4.Senses,Imagination,andThought.Beingabletousethesenses,to imagine, think,andreason‐and todo these things ina "trulyhu‐man"way,awayinformedandcultivatedbyanadequateeducation,including,butbynomeanslimitedto,literacyandbasicmathemati‐calandscientifictraining.Beingabletouseimaginationandthoughtinconnectionwithexperiencingandproducingworksandeventsofone'sownchoice,religious,literary,musical,andsoforth.Beingableto use one's mind in ways protected by guarantees of freedom ofexpression with respect to both political and artistic speech, andfreedomofreligiousexercise.Beingabletohavepleasurableexperi‐encesandtoavoidnonbeneficialpain.
5.Emotions.Beingabletohaveattachmentstothingsandpeopleoutsideourselves;tolovethosewholoveandcareforus,togrieveattheir absence; in general, to love, to grieve, to experience longing,gratitude, and justified anger.Not having one's emotional develop‐
97
mentblightedbyfearandanxiety.(Supportingthiscapabilitymeanssupportingformsofhumanassociationthatcanbeshowntobecru‐cialintheirdevelopment.)
6.PracticalReason.Beingabletoformaconceptionofthegoodand to engage in critical reflection about theplanningof one's life.(This entails protection for the liberty of conscience and religiousobservance.)
7.Affiliation.A.Beingabletolivewithandtowardothers,torec‐ognizeandshowconcernforotherhumanbeings,toengageinvari‐ousformsofsocialinteraction;tobeabletoimaginethesituationofanother. (Protecting this capability means protecting institutionsthat constitute and nourish such forms of affiliation, and also pro‐tectingthefreedomofassemblyandpoliticalspeech.)B.Havingthesocial bases of self‐respect and nonhumiliation; being able to betreatedasadignifiedbeingwhoseworth isequal to thatofothers.Thisentailsprovisionsofnondiscriminationonthebasisofrace,sex,sexualorientation,ethnicity,caste,religion,nationalorigin.
8.OtherSpecies.Beingabletolivewithconcernforandinrela‐tiontoanimals,plants,andtheworldofnature.
9.Play.Beingableto laugh,toplay,toenjoyrecreationalactivi‐ties.
10. Control over One's Environment. A. Political. Being able toparticipateeffectivelyinpoliticalchoicesthatgovernone'slife;hav‐ingtherightofpoliticalparticipation,protectionsoffreespeechandassociation.B.Material.Beingable toholdproperty(both landandmovablegoods),andhavingpropertyrightsonanequalbasiswithothers;havingtherighttoseekemploymentonanequalbasiswithothers;havingthefreedomfromunwarrantedsearchandseizure.Inwork,beingabletoworkasahumanbeing,exercisingpracticalrea‐sonandenteringintomeaningfulrelationshipsofmutualrecognitionwithotherworkers.”
Inordertoidentifyanddocumenttheethicalconsequencesofar‐tefactuse,adesignercanfillinatableliketheoneinfigure9.2.For
98
eachgroupor individualandeachcapability, thedesigner is tode‐terminetheeffectsoftheuseoftheartefact.
Figure 9.2 Capability based template for ethical analysis
9.6 Running Example
ThefifthDIGIREADactivitywastoevaluatethedesignedanddevel‐opedartefact.Anexanteevaluationstrategywaschosen,i.e.thear‐tefactwasevaluatedwithoutbeingused.
The ex ante evaluation of the artefact was then carried out, inwhich experts with different competences participated. Three ex‐pertswereinterviewed,onefromtheareaofsocialmedia,onefromusability andone from software engineering.A semi‐structured in‐terviewofabout45minuteswascarriedoutwitheachexpert.Theinterviewswerevideorecordedandthenindependentlyanalysedbytwomembersoftheresearchteam.Theevaluationwasbasedonthedefinedrequirementsoftheartefact.Themainoutcomeoftheeval‐uationwasaconcernaboutusability,astheproposedartefactcouldbehighlycomplex.
Inadditiontotheexpertinterviews,theresearchersalsoprovid‐edinformedargumentsasapartoftheevaluation.Theseargumentsprimarilyshowedthat theconstructionof theartefactensuredthat
99
the functional requirements were at least partially fulfilled. Fur‐thermore, itwasarguedthattheproposedarchitectureshouldpro‐vide adequate interoperability. The Evaluate Artefact activity issummarizedinfigure9.3.
Figure 9.3 Evaluate Artefact in the running example
Further Reading
Agreatdealofliteratureexistsonevaluationinvariousdomains.ForHuman Computer Interaction a comprehensive introduction in theform of a presentation is given by Dix (2009). The evaluation ofmodelsandmethodsin informationsystemsisdiscussedbyMoody(2003; 2005). Clements et al. (2001) investigate the evaluation ofsoftware architectures. Pries‐Heje et al. (2008) discuss the distinc‐tionbetween ex ante and ex post evaluations in design science re‐search. The box above on design theory is based on the work byWallsetal.(2004)andGregorandJones(2007).
100
10 Communicate Artefact Knowledge
Designscienceresultsaretypicallytobecommunicatedtoresearchaswell as practitioner communities,whichmay include both tech‐nology‐orientedandmanagement‐orientedaudiences.Furthermore,somedesignscienceresultscanbeofsuchabroadinterestthattheyareworthwhiletocommunicatetothegeneralpublic.
Communicatingresults to researchers requiresattention to rig‐oursothattheycanevaluatetheresultsandbuildontheminfuturework. In particular, the knowledge base should be carefully de‐scribedaswellasitsrelationshiptotheresultsproduced.Thechoiceofresearchstrategiesandmethodsshouldbewelljustified.Theap‐plicationofthechosenresearchstrategiesandmethodsshouldalsobedescribedindetail includingdiscussionsonvalidityandreliabil‐ity.
Theseconcernsaboutmethodologyandrelatedresearcharelessrelevantforcommunicationtopractitioners.Thefocusisthenratheron problem and practice aswell as concrete outcomes in terms ofconstruction,function,usabilityandeffects.
Technology‐orientedaudiencesbenefitfromextensivedetailsontheconstructionofanartefact,i.e.itscomponentsandtheirrelation‐
101
ships.Thisallowspractitionerstoconstructandimplementthearte‐factinapractice,andresearcherstofurtherdeveloptheartefact.
Management‐orientedaudiencesareprimarily interested in theproblemtheartefactaddresses,whatbenefitsitcanbringtoaprac‐tice,howeasyitistouse,anditsoveralleffects,e.g.onefficiencyandagility in an organisational setting. Knowledge about these aspectswillenablemanagerstodeterminewhetherorhowtoapplythearte‐fact.Theconstructionoftheartefactislessrelevant,thoughmanag‐ers will still need a basic knowledge of the inner workings of theartefact in order to appreciate its significance and understand itsapplication. When communicating to the public, the main focus isoftenontheeffectsofanovelartefact,includingitsethicalandsocie‐talconsequences.Theinterestsofdifferentaudiencesaregraphicallyillustratedinfigure10.1.
Figure 10.1 Different interests for different audiences
When disseminating design science results, the researcher has toselect the right communication channels depending on the target.For research communities, results are primarily communicated
102
throughacademicjournals,conferencesandworkshops.Workshopsoftenaccept immatureworkandprovidea forumfor feedbackanddiscussionaboutpreliminaryresults,while journalsmainlypublishmatureworkthatincludescarefulevaluation.Forpractitionercom‐munities,researchresultscanbecommunicatedthroughtradefairs,whitepapers,magazinearticles,blogs,etc.Regardlessofthetargetofthe communication, it is valuable to present results in a clear andeasily understandable structure. For example that provided by thedesign sciencemethod as depicted by the IDEF0diagrams and thedesignsciencecanvas.ThisstructureisinmanyrespectsalsosimilartotheIMRAD(Introduction,Methods,ResultsAndDiscussion)struc‐ture,whichisfrequentlyusedfororganisingempiricalresearchpa‐pers.
Is design theory the same as design science?
Short answer: No
Long answer: In this book, design science has been presented as a knowledge building endeavour, which can be carried out according to the design science method presented in Chapters 4 to 9. In contrast, a design theory focuses on the results of the design science activities and shows how these can be structured. A design theory consists of a num‐ber of interrelated components that describe abstract artefacts like models and methods (each component is here exemplified using Codd’s relational database theory): Purpose and scope. The set of requirements or goals that specifies the type of artefact to which the theory applies and in conjunction also de‐fines its scope and boundaries. (Improved database technology is need‐ed for increasing productivity as existing approaches of managing data persistence are failing.)
Constructs. Representations of the entities of interest in the theory. These entities can be physical as well as abstract. (Attribute, tuple, n‐ary relation, domain of values.)
Principle of form and function. The abstract blueprint or architecture that describes an artefact. For a model, its components and their rela‐tionships and functions would be described. For a method, its steps and
103
their order and purpose would be given. (A relation is defined over a domain of values and includes attributes and tuples.)
Artefact mutability. The changes in state of the artefact anticipated in the theory, that is the degree of artefact change encompassed by the theory. (The relational model allows for easy adaptation and change to base tables, while user views appear unchanged.)
Testable propositions. Testable propositions about instantiations of the artefact under consideration. Typically, these propositions state that if a model or method is instantiated then it will work, or it will have certain characteristics. (A relational database can perform as well as a non‐relational database.)
Justificatory knowledge. The underlying knowledge or theories that provide a basis and justification for the design. (Set theory and behav‐ioural science about human cognitive processes.)
Principles of implementation. Processes for implementing, i.e. instanti‐ating, the artefact to which the theory applies (model or method) in specific situations. (Guidelines on how to create a relational database through normalization procedures.)
Expository instantiation. An implementation of the artefact that can assist in representing the theory both as an expository device and for purposes of testing. An instantiation can support communication about the artefact of a design theory. (A working relational database with ta‐bles filled with data.)
While the design science method focuses on organising research activi‐ties, a design theory aims at structuring the knowledge about an arte‐fact. However, they are closely related in the sense that applying the design science method should result in a design theory. The Explicate Problem and Define Requirements activities should result in defining the purpose and scope of the design theory. The Design and Develop activi‐ty should identify the constructs and specify the form and function of the artefact. Testable propositions should be deduced by the Evaluate Artefact activity, and an expository instantiation should be shown by the Demonstrate Artefact activity. Justificatory knowledge is to be ad‐dressed by all the activities of the design science method. Artefact mu‐
104
tability and principles of implementation can also be addressed by sev‐eral activities.
Further Reading
Anestablishedandeasy to read textbookon scientific communica‐tionwaswrittenbyBoothetal.(2008).Sörensen(2005)haswrittenanentertainingpaperonhowtoandhownottowritescientificarti‐cles.Dixit(2011)hasprovidedaconciseintroductiontotheIMRADformat.
105
11 The Design Science Method and
Systems Development
A quick look at the design science method may suggest that it isnothingmore than yet another version of the traditional waterfallmodelforsoftwaredevelopment.Thewaterfallmodelisasequentialdesign process, where progress is viewed as a steady flow down‐wardsinanumberofconsecutivephases,seefigure11.1.
Figure 11.1 The Waterfall Model
Theresultsproduced inonephasearehandedoveras input tothenextphase,meaningthataphasecanonlybestartediftheprevi‐ousphasehasbeenfullycompleted.Thephasesindifferentversions
106
ofthewaterfallmodelmayvary,buttypicallyincluderequirementsspecification, design, construction, testing, installation, andmainte‐nance.
Thewaterfallmodelhasbeenheavilycriticizedforbeinginade‐quate forguiding largesoftwareprojects inunstableenvironments.Inanearlystageofthesoftwaredevelopmentprocessitisoftenim‐possibletodefinepreciseandcompleterequirements.Insuchsitua‐tions,rapidprototypingisanessentialtoolforelicitingrequirementsfrom stakeholders. By creating and testing prototypes of the soft‐ware tobedeveloped, stakeholderswill improve theirunderstand‐ingandbeabletoarticulatemoreexactandrelevantrequirements.Thus, theremay bemany short iterations between specification ofrequirements,design,construction,andtesting.Forexample,basedon some stakeholders’ requirements, a first draft of a prototype isdesigned, constructed and tested. The results of the test will spurneworrefinedrequirements,whichwillbethestartofanewitera‐tionwith a refined design of the prototype, followed by additionaltests. The result of the tests will give rise to new and refined re‐quirements,whichstartanotheriteration,andsoon,seefigure11.2.Furthermore,theenvironmentmaychangeduringthedevelopmentprocess,e.g.duetotechnologicaldevelopmentsorgovernmentregu‐lations,whichimpliesthattheremaybeaneedtoiterateallthewaybacktorequirementsspecification.Thus,thedevelopmentprocessishighly iterative as it needs to obtain complex requirements andmanageachangingenvironment.
107
Figure 11.2 A Waterfall Model with a simple iteration
11.1 Temporal and Logical Groupings of Work
Asthedesignsciencemethodmaylookverymuchlikeasequentialdesignprocess,seefigure4.1,itcouldbearguedthatitsuffersfromthesameweaknessesasthewaterfallmodel.Inotherwords,itcouldbeclaimedthatthedesignsciencemethoddoesnotcaterforvaguerequirements,changingenvironments,shiftingstakeholderinterests,unclearproblemsituations,etc.Ifthisweretrue,themethodwouldclearlynotbeofmuchinterest.However,themethoddoesnotpre‐scribeasequentialwayofworking.Theactivitiesarenottobeseenas temporal groupings of work to be performed in sequence, butinsteadas logicalgroupingsofwork.Thus, thearrows in figure4.1shouldnotbeinterpretedastemporalorderingsbutasinput‐outputrelationships.These interpretationscanbemadeclearerbyplacingthem in the context of the RUP (Rational Unified Process) frame‐work.
RUPisstructuredintwodimensions,phasesanddisciplines,thatcapture the serial and iterative aspects of software development,respectively, see figure 11.3. The phases represent the sequentialstagesthataprojecttraversesovertime,whilethedisciplinescorre‐spond to the logical activities that are carriedoutduringaproject.RUPidentifiesfourphases:inception,elaboration,construction,andtransition.The inceptionphaseaimsatachievingconsensusamongthestakeholdersabouttheobjectivesoftheproject;theelaboration
108
phase specifies requirements in detail and outlines an architectureforthesystemtobebuilt; theconstructionphasefocusesondevel‐opingthesystemsothat it isreadyfordeployment;andthetransi‐tionphasedeliversthesystemintoproduction.
Figure 11.3 Disciplines and phases in RUP
Whilethephasescapturetheserialaspectsofaproject inRUP, thedisciplines address the iterative aspects. During each phase, thesoftwaredeveloperswillalternatebetween(almost)allofthedisci‐plines.Forexample,duringthe inceptionphasethedeveloperswilltypicallyaddressasubsetoftherequirements,carryoutsomebusi‐nessmodeling,returntotherequirementsandrevisethem,suggestaninitialdesign,onceagainrevisetherequirements,dosomecodingand preliminary tests, and then improve the design. Furthermore,theinceptionphasecanbebrokendownintoseveraliterations,eachof which only addresses a small portion of the system to be built,therebymakingRUPevenmoreiterative.therebymakingRUPevenmoreiterative.Similarly,alltheotherphaseswillalsoincludesever‐aldisciplines.
The relationshipsbetweenphases anddisciplines are shown infigure 11.3,which is sometimes called a “hump chart”. The humps
109
foreachdisciplinedisplayhowmucheffortthatisspentonthatdis‐cipline in thevariousphases.Forexample, thediagramshows thatmostoftheworkonrequirementsiscarriedoutintheinceptionandelaboration phases, but it continues all theway into the transitionphase. The sizes and placements of the humpsmay certainly varyfromprojecttoprojectandcanbetailoredasrequired.
Figure11.4showsaversionofthehumpchart,whichisclosetoawaterfallmodel requiring that, forexample, allworkonbusinessmodelingandrequirementsbecompletedbeforeanyworkondesigncanbestarted.
Figure 11.4 A Waterfall version of RUP
Anotherversionofthehumpchartisgiveninfigure11.5,requiringthatalldisciplinesaregivenequalattentioninallphases.Thiskindofdiagramwouldreflectanextremelyagiledevelopmentprocess.
110
Figure 11.5 An Agile version of RUP
Returningtothedesignsciencemethod, itsactivitiesareanalogoustothedisciplinesofRUP,notthephases.Inotherwords,theactivi‐tiesarenottobesequentiallyorderedintimebutcanbeperformedinparallelandinanyorder.Theactivitiesonlytellwhattasksneedtobedoneinthemethod,whatinputtheyconsume,andwhatoutputtheyproduce.Theactivitiesarelogicalgroupings,nottemporalones.
Design and Appropriation
People often adapt and use artefacts in ways that their designers never intended. For example, email was designed to support communication between people, but it is nowadays regularly used by people to send reminders to themselves or to store safety copies of documents. Dix (2007) describes this process of technology adaptation as follows:
“These improvisations and adaptations around technology are not a sign of failure, things the designer forgot, but show that the technology has been domesticated, that the users understand and are comfortable enough with the technology to use it in their own ways. At this point we
111
know the technology has become the users' own not simply what the designer gave to them. This is appropriation.“
Designers should be aware that the artefacts they create may be ap‐propriated in unanticipated ways. They could even try to design in order to facilitate appropriation. Dix (2007) suggests a number of guidelines for this, three of which are:
Allow interpretation – “Don’t make everything in the system or prod‐uct have a fixed meaning, but include elements where users can add their own meanings.”
Provide visibility – ”Make the functioning of the system obvious to the users so that they can know the likely effects of actions and so make the system do what they would like.”
Support not control – ”As a designer you want to do things right, to make them as efficient and optimal as possible. However, if you opti‐mise for one task you typically make others more difficult. In some situ‐ations, such as very repetitive tasks, then designing explicitly for the task may be the correct thing to do, perhaps taking the user step by step through the activities. However, more often the tasks description is incomplete and approximate, in particular ignoring exceptions. Instead of designing a system to do the task you can instead design a system so that the task can be done.”
11.2 Differences between DS and Systems Development
Comparing the activities of thedesign sciencemethod to thedisci‐plinesofRUP,theylookquitesimilar.ExplicateProblemissimilartoBusiness Modelling, Outline Artefact and Define Requirements issimilartoRequirements,andsoon.Thisobservationholdsnotonlyfor RUP but also for many other systems development methods.Thus, some new questions arise: Is the design sciencemethod notredundant?Could itnotbereplacedbysomesystemsdevelopmentmethod?Why introduceanewkindofprocesswhenRUP is there?Theanswerstothesequestionsdependonthedifferentpurposesofdesignanddesignscience,whichwerealreadyintroducedinChapter1.
The purpose of design is to create an artefact that fulfils theneeds and requirements of some stakeholders, possibly only for a
112
local practice. Design science, in contrast, aims at producing andcommunicatingnewknowledgethatisrelevantforaglobalpractice.These differences in purpose give rise to three additional require‐mentsondesignscience.
First, thepurposeofcreatingnewandgeneralizableknowledgerequires design science projects to make use of rigorous researchstrategiesandmethods.Suchmethodsareessential forcreatingre‐sultsthatcanbecriticallydiscussed,evaluated,andvalidated.Apos‐sibleobjectiontothis lineofreasoningisthatmanysystemsdevel‐opmentmethodsalreadyincludetoolsandtechniquesthatarequitesimilartoresearchmethods,e.g.techniquesforinterviewsorpartic‐ipative requirements elicitation. It is true that some of these toolsand techniquesarecloselyalignedwithestablishedresearchmeth‐ods,butnotalwaysastheyonlyaimatproducingeffectivesolutions,not knowledge for global practices. Thus, the design researcherneeds to determinewhether the tools and techniques offered by asystemsdevelopmentmethodare indeedsufficient for thepurposeof producing knowledge, or whether additional research methodsarerequired.
Thesecondadditionalrequirementondesignsciencestatesthattheresultsproducedmustberelatedtoanexistingknowledgebase,therebyensuringthat theyarebothwell foundedandoriginal. It isnotsufficientjusttoproducesomeknowledge;italsohastobeinte‐gratedwith previous knowledge in the area and shown to providenovelinsights.
Thirdly, the new knowledge should be communicated to bothpractitioners and researchers.This requirementon communicationdoesnotexistinmostsystemsdevelopmentmethods,whereinsteadthereareactivitiesdevotedtodeploymentandmaintenance.
In a project for developing an IT system, the design sciencemethodcanbeusedintandemwithanysystemsdevelopmentmeth‐od.A typical scenario could look as follows. First, a systemsdevel‐opment method is chosen. Second, the relevant parts (disciplines,workflows, tasks or whatever they may be called) of the chosen
113
method aremapped to the activities of the design sciencemethod.Third,thetoolsandtechniquesusedbythemethodareinspectedinordertodeterminehowcloselytheymatchadequateresearchmeth‐ods.Fourth,thetoolsandtechniquesareadaptedorcomplementedso that sufficiently rigorous research methods are used. Fifth, theworkoftheprojectiscarriedoutinaccordancewiththechosensys‐temsdevelopmentmethodaswellastheselectedresearchmethods.Inparallel, theprojectwill includework to relate the results to anexistingknowledgebase.Finally,theresultswillbecommunicatedtorelevantaudiences.Thus, theprojectwillprimarilybegovernedbythe chosen systems development method, but there will also bemodified or additional activities for validating and communicatingtheknowledgeproduced.
Can companies use design science in their product development?
Short answer: Yes
Long answer: When companies need reliable knowledge about their customers, they can benefit from using design science. For example, if a company wants solid knowledge about the requirements of their cus‐tomers, it can make use of research methods for requirements defini‐tion, including interviews and questionnaires. Likewise, if a company wants to understand how their customers perceive its products, re‐search methods for evaluation can be relevant.
Some companies are well known for using research methods in their product development. One example is Google that regularly tests and evaluates its service offerings by means of advanced quantitative data analysis methods. Before a product launch, Google rolls out the new service to a limited number of users, collects data by recording their interactions with the service, and evaluates the records in order to im‐prove it.
Caveat: The goal of companies is not to produce knowledge but to cre‐ate value for their shareholders and other stakeholders. Thus, scientifi‐cally founded knowledge is never an end in itself for a company but only a means to other goals. In many cases, it is still worthwhile to use re‐
114
search methods to obtain reliable knowledge that can be used as a basis for decision making. However, research methods come with a cost, in particular in terms of time. For example, carefully evaluating a product before launching it takes valuable time, which may result in a competi‐tor being first to market. Therefore, companies often deliberately choose to refrain from scientific investigations in order to improve speed and flexibility.
Further Reading
Atextbookonmethodologies for informationsystemsdevelopmentwas written by Avison and Fitzgerald (2006). Kroll and Kruchten(2003) havewritten an easily readable introduction to RUP. Basicprinciples behind agile development were proposed by Beck andBeedle(2012).
115
12 Research Paradigms
Aresearchparadigmisasetofcommonlyheldbeliefsandassump‐tionswithinaresearchcommunityaboutontological,epistemologi‐cal, and methodological concerns. Such a paradigm constitutes amentalmodelthat influencesandstructureshowthemembersofaresearch community perceive their field of study. A research para‐digmanswersontologicalquestionsaboutthenatureofreality,whatentitiesexist,andhowthesearerelatedandinteractwitheachother.Aresearchparadigmalsoaddressesepistemologicalquestionsabouttheways inwhichpeoplecanknowaboutreality, i.e.howtheycangain knowledge about the world. Finally, a research paradigm an‐swersmethodologicalquestionsaboutlegitimatewaysofinvestigat‐ingrealityandhowtoconfirmthattheknowledgegeneratedisvalid.Summarizing, ontology asks “What is in theworld?”; epistemologyasks“Whatcanweknowabouttheworldandhowshouldweobtainthatknowledge?”;andmethodologyasks“Whichprocedurescanbeusedtoobtainknowledge?”.
Thetwomostestablishedresearchparadigms in theareaof in‐formation systems are positivism and interpretivism. This chapterintroduces these paradigms and discusses their relationship to de‐
116
signscience,seealso figure12.1,whichrelatesresearchparadigmstoempiricalresearchstrategiesandmethods.
Figure 12.1 Research paradigms
12.1 Positivism
Positivismoriginatedwiththe19thcenturysociologistandphiloso‐pherAugusteComte,whoattemptedtoestablishsociologyasasci‐encebyapplyinganaturalscienceviewonsocialphenomena.Comteintroducedpositivismasareactiontotheologicalandmetaphysicalworldviewsthatembracedauthority,divinerevelationandtraditionaslegitimateknowledgesources.Incontrast,positivismonlyacceptsknowledge that isbasedonsense,experienceandpositiveverifica‐tion.
Ontologically, positivism assumes a reality that exists inde‐pendentlyofhumanactionsandexperiences.Asfornaturalscience,the goal of social science should be to identify regularities amongphenomenaintheworldandexplainthemthroughcauseandeffectrelationships.Preferably,theseregularitiesandexplanationscanand
117
shouldbeindependentofcontext,i.e.theyshouldbegeneralandnotdependontime,placeorpeople.
Epistemologically, positivism claims that objective knowledgeabout the socialworld is obtainable, but only through observationand experimentation. Social inquiry should be objective, and a re‐searchershouldassumetheroleofadisinterestedobserverwho isseparate from the subjects being investigated. In other words, theresearcher should keep at arm’s length from the phenomena shestudiesinordertoensurethatherbackgroundandinterestsdonotbiasherfindings.
Methodologically, positivist researchers strive for an objectiveand value‐free investigation where they distance themselves fromthe entities being studied. Preferred instruments for collecting re‐search evidence are large quantitative studies including interviewsand questionnaires. Experiments are also highly valued researchstrategies,astheycanprovideobjectiveknowledge.
Positivism as described above may seem natural and com‐monsensical.However,thisisexactlytheproblemwiththepositivistparadigmaccording to interpretivism.Whilepositivismmaybeap‐propriatefornaturalscience,interpretivistsarguethatitfailstocap‐tureessentialaspectsofthesocialworld,inparticularthesubjectiveconstructionofsocialphenomena.
12.2 Interpretivism
Interpretivismemergedasareactiontopositivismatthebeginningofthe20thcentury.OneofitsforerunnerswastheGermansociolo‐gistMaxWeber,whoclaimedthatthesocialworld, includingsocialactions, can only be understood through grasping the subjectivemeaningsandpurposesthatpeopleattachtotheiractions.
Ontologically,interpretivismarguesthatincontrasttothenatu‐ralworld,thesocialworlddoesnotexist“outthere”,independentofhumanactionsandintentions.Instead,thesocialworldisconstruct‐edbypeoplewhocarryoutsocialactionsandgivemeaningstothem.For example, housing contracts are created through social actions
118
involvingtwoormoreindividuals,anationalgovernment,andpos‐sibly other actors. The meaning of such a contract is not given inadvancebutisdeterminedbytheactorswhoenterit.Furthermore,themeaningofthecontractmaychangeduringitslifetimeasnegoti‐atedbytheactorsinvolved.Itmayalsobethecasethattheactorsdonot fully agreeon themeaningof the contract butmake theirowndifferentinterpretations.Inextremecases,suchdisagreementsmayresultincourttrialswhereathirdparty,thecourt,decideshowthecontractistobeinterpreted.Thisexampleillustratesthatevensucha straightforward social phenomenon as a housing contract is notpartofastable,objectivelyexistingreality.Instead,itisconstructedand continuouslymodified through the actions and interpretationsoftheactorswhohaveaninterestinit.
Any social phenomenon emerges as a result of the interactionsandlivedexperiencesofhumans.Thisholdstrueformembershipsinclubs, employment contracts, marriages, money, debts, holidays,religions,etc.Allofthesephenomenaarecreatedbypeopleandtheirmeaning depends on the actions, intentions and understanding ofthe individuals who participate in them. Thus, the social reality ismuchmoreelusiveandfluidthanthephysicalone,asitdependsonpeoplewithalltheirwhims,prejudicesandothersubjectivities.
The ontological differences between the natural and the socialworldhaveepistemologicalconsequences.Associalphenomenaaregrounded in the actions, experiences and subjective meanings ofpeople,onlysuperficialknowledgecanbeobtainedbystudyingpeo‐plelikeobjects.Instead,aresearchershouldviewpeopleassubjectswhoactivelycreatethesocialworld.Theresearchercanonlyachieveadeepunderstandingofasocialphenomenonbyactivelyparticipat‐ing inthatphenomenontogetherwiththepeoplewhoactuallycre‐ate it. She shouldnot be a detachedobserver but rather try to be‐comeamemberofthecultureorgroupbeingstudiedbyparticipat‐ingintheirdailypractices.Inshort,sheneedstoentertheshoesoftheother.Thisisinstarkcontrasttothepositivistidealofseparatingthe researcher from her object of study. Such a separation is self‐
119
defeatingaccordingto interpretivism,as thesubjectiveknower isapartandsourceofthesocialrealitysheknows.
Methodologically, interpretivist researchersprefer case studies,actionresearchandethnography,astheseresearchstrategiesallowthem to arrive at an empathetic or participatory understanding ofsocialphenomena.Theresearcherscangetclosetothepeoplewhoparticipate in the phenomena being studied and thereby come tounderstand their views and interpretations. However, this kind ofmethodology has been criticized for producing subjective researchresults,i.e.resultsthatarehighlydependentontheskillsandexpe‐riences of the individual researcher. There is a risk that two re‐searchers with different backgrounds and interests may arrive atverydifferent results.Oneanswer to this criticism is that an inter‐pretivistresearchershouldalwaysacknowledgehersubjectivityandexplicitlydiscusshowitcouldaffectthevalidityofherresults.
Positivism and Interpretivism in the Movies ‐ Tim Burton’s The Nightmare Before Christmas
As the arguments of interpretivism may seem somewhat abstruse on a first encounter, it can be helpful to make them concrete by putting them into some familiar context. One such context is provided by the 1993 stop motion movie The Nightmare Before Christmas produced by Tim Burton. The movie illustrates how difficult it is to understand and interpret the habits and symbols of foreign cultures.
The main character of the movie is Jack Skellington, the Pumpkin King of Halloween Town, a place inhabited by ghouls, ghosts, vampires, witches and other monsters. One day Jack stumbles upon a tree, which is a portal to Christmas Town. He enters the tree and finds himself tele‐ported to a friendly and joyful world, completely different from his own Halloween Town. The following video track records his impressions: http://goo.gl/yCV88.
The video shows how hard it is for Jack to make any sense of the people, activities and things he discovers in Christmas Town, as they are so unrelated to his own experiences. Still, Jack wants to bring the
120
Christmas spirit to his own people, and he calls for a town meeting when he returns to Halloween Town http://goo.gl/zkSQt.
In the meeting, Jack presents his findings on Christmas to the Hallow‐een Town residents. They may understand some basic facts about the holiday, but as Jack complains in the end of the scene “they don’t un‐derstand that special kind of feeling in Christmasland”. The problem is that the town people immediately try to interpret Christmas in their own cultural terms. For Halloween Town residents, a present in a box is a pox. A Christmas sock contains a rotten foot. Toys bite, snap and ex‐plode. Without even thinking about it, the Halloween Town people as‐sume that their own views and experiences can be used to understand the world of Christmas. However, as any interpretivist would be happy to point out to them, this does not work. Gift giving, present wrapping and other Christmas habits are social phenomena that are created and given meanings by those who participate in them. They cannot be un‐derstood by outsiders who just observe them in a casual and disinter‐ested manner. But understanding is what Jack seeks, as he calls out in the next video “What does it mean?”: http://goo.gl/r3GcP.
Jack seeks meaning in two ways. He first tries the positivist road with observation and analysis. He puts toys into vials, inspects rag dolls, and learns Christmas rhymes by heart. In some deleted scenes, Jack also dissects a teddy bear with a scalpel, dissolves candy in test tubes, and formulates equations about Christmas notions: goo.gl/j8Xxk. However, Jack soon realizes that these kinds of investigations will not grant him any understanding of Christmas, “something’s there I cannot see”. So he then switches to an interpretivist mode, summarized in the phrase “Just because I cannot see it doesn’t mean I can’t believe it”. Christmas is not understood by observing Christmas objects. Christmas is understood by believing it. Christmas is understood by making it. And the people of Halloween Town start making Christmas: http://goo.gl/7YVFe.
They become participants of the Christmas tradition by believing it, by making it, by living it. No longer are they passive observers, but in‐stead active creators. So through their lived experiences they come to understand Christmas. Or do they? Of course they do not, and their failure to understand drives the rest of the movie forward to its tragic end. Someone may argue that this lack of understanding is a general rule and that people are so caught up in their own culture that they
121
always interpret other cultures in terms of their own. They cannot step outside and truly understand a foreign culture, or at least their under‐standing is so limited and subjective that it cannot be trusted. Is this a fair conclusion? The movie offers an answer, but it would be a spoiler to give it here...
12.3 The Roles of Positivism and Interpretivism
While the differences between positivism and interpretivism mayseemdeepandsubstantive, ithasbeenarguedthatmanyof theal‐leged differences are spurious, in particular the ontological ones(Weber,2004).Furthermore,someofthepositionsascribedtoposi‐tivistsareoutdatedandnot supportedbyanyone today.Positivistswouldnowadayscertainlyagreethatsomepartsofrealityaresocial‐ly constructed, includingmost of the research objects in the infor‐mation systemsarea.Historically, however, some researchershaveassumed thatmany complex social phenomena are immutable andindependentoftime,placeandperson.Forexample,phenomenalikechildhood,genderandhonourhavebeen treatedasstable,welles‐tablished, and independent of culture and time. Clearly, this is notthecase–justcomparethegenderstereotypesoftoday’sScandina‐via with those of 19th century Prussia. Researchers who have as‐sumed that social phenomena are culturally independent have inpracticeascribedtheirownconceptionstopeopleinothercultures,who have actually held widely different views. As interpretivistshavepointedout,thisisasurerecipeforbadresearch.
Epistemologically, interpretivists do understand and even ap‐preciate the positivist ambition for objectivity. However, they con‐tend that objective research results are only attainable if the re‐searcher restricts herself to surface level phenomena in the socialworld,suchascorrelationsbetween incomeandeducation level. Inorder to obtain deeper knowledge, the researcher needs to engagecloser with people to understand their views and interpretations.Obviously, this compromises objectivity, but the interpretivist seesthisasapricewellworthpayingfortheinsightsgained.Apositivist
122
may be less inclined to make this trade off and might argue thatmany results obtained through interpretivist investigations are sosubjective that theyarenext touseless.Furthermore, thepositivistmayclaim that typicallypositivist research strategies also canpro‐videdeepknowledgeonsocialphenomena,e.g.ingeniouslydesignedexperimentssuchasversionsof theprisoner’sdilemmaor theulti‐matumgame.
Another epistemological issue is the social construction ofknowledge.Positivistswouldhere acknowledge that any construct,framework,modelortheoryissociallyconstructed,whetheritssub‐jectmatterissociallyconstructedornot.Thus,researchresultsarealways influencedby the culture, experience andhistory of the re‐searcher.ThisrecognitionisacornerstoneinKuhn’saccountofhowresearchers develop theories not only in the social but also in thephysical and life sciences (Kuhn, 2012). Thus, the ephemeral andculturallydependentnatureofknowledgeisacknowledgedbyposi‐tivistandinterpretivistresearchersalike.
Asdiscussedearlier,positiviststendtopreferresearchstrategieslikeexperiments,surveysand fieldstudies,while interpretivists fo‐cus on case studies, action research and ethnographic studies. Thepros and cons of these research strategies follow the same line ofreasoningasintheprevioussection.Simplifyingsomewhat,theposi‐tivistonesprovidereliablebutshallowknowledge,while the inter‐pretivistonesofferdeepbutunreliableknowledge.Inordertoover‐come the weaknesses of the respective research strategies, it hasbeen suggested to combine them. The interpretivist ones shouldthen be used for generating and suggesting research results,whilethepositivistonesshouldbeusedforverifyingtheseresults.
Critical Theory and Design Science
Most social sciences have as their goal solely to explain and understand social and cultural phenomena. Critical theories, by contrast, have also added the practical and ethical goal of seeking human emancipation. As expressed by Horkheimer (1975), research should “liberate human be‐
123
ings from the circumstances that enslave them”. Critical theories are not only concerned with how things are but also how they might be or should be. Critical theories are critical in the sense that they question hidden assumptions and purposes in academic theories as well as in political ideologies and cultural practices. By exposing these assump‐tions, critical theories can make people aware of the oppressing effects of established theories, ideologies and practices. Such an awareness is the first step towards change that can set people free.
As critical theories aim to decrease domination and increase freedom in all their forms, they have been applied in many contexts, including inquires on race, gender, ethnicity and identity. In an analysis of gender within some tradition or practice, a critical theory may ask questions about the equal status of women, whether they are demeaned or ideal‐ized, and whether they are ignored or patronized. The answers to these questions can help people start to understand if and how women are subordinated in the tradition.
According to critical theory, two of the most common forms of domi‐nation in modern society are alienation and reification. Alienation is about the psychological effects of exploitation and the division of la‐bour, where an individual is disconnected from the tools of production as well as her fellow workers. In other words, people lose control of their lives and selves by not being able to control their work. Reification is about transforming social relations into things, or more precisely so‐cial relations become mediated and expressed by things, in particular commodities and markets. For example, the patriarchal relation be‐tween a farmer and his farmhand is replaced by an impersonal relation based on money and market valuations.
Critical theories and design science share the conception that describ‐ing and explaining the world is not sufficient; it also needs to be changed. Critical theories bring about this change by raising people’s awareness of patterns of domination, thereby spurring them to action. Design science causes changes by developing and introducing new arte‐facts. However, critical theories and design science differ widely with respect to their goals. While critical theories aim at liberating people from oppressing structures, design science projects have the more lim‐ited purpose of satisfying stakeholder needs and wants. In other words, critical theories attempt to realize the universal values of liberty and
124
autonomy, while design science projects are governed by the particular interests of stakeholders of a practice.
These differences in goals may even give rise to conflicts. An example could be a design science project, where a new ERP system is to be de‐signed and introduced in a company. One group of stakeholders, the management, claims that the goal of the new system is to improve effi‐ciency and productivity by eliminating routine work activities. However, a critical theorist could argue that the hidden motive of management is to better monitor, control and evaluate employees, which would con‐tribute to their alienation and reduce their autonomy. The critical theo‐rist could even claim that the design science researchers are manage‐ment lackeys who help strengthen controlling and oppressive structures in the company. This example illustrates that a design science research‐er needs to take a broad outlook during problem explication and re‐quirements definition. She has to acknowledge the viewpoints of multi‐ple stakeholders, and probe into the problem situation using different research methods, in order to get a rich and unbiased understanding.
12.4 Research Paradigms and Design Science
In design science, just as in empirical research, both the positivistand interpretivist paradigm can be applied. It is even possible, in‐deed common, tomake use of both paradigms in the same designscienceproject.Duringproblemexplicationandrequirementsdefini‐tion, a researcher may choose an interpretivist stance in order toobtainadeepunderstandingoftheneedsandwantsofstakeholders.Thisunderstandingwill helpher todesignanddevelopan artefactthatishighlyrelevantforthestakeholdersofapractice.Forevalua‐tion,however, shemayvaluerigourmorehighlyanddecide tousepositivistmethods likeextensivesurveysandexperiments inordertoarriveatobjectivelyvalidresults.Still,interpretivistmethodsmayalso be appropriate for evaluation, especially if there is a need tounderstand the subjective experiences of users. In summary, a de‐sign science project can use and benefit from positivist as well asinterpretivist strategies and methods, depending on the goal andcontextoftheproject.
125
What design science is and is not
Design science is not design – the purposes of design and design science differ with respect to generalizability and knowledge contribution, as pointed out in Section 1.5. Design science is not a research strategy – instead, different research strategies can be used in design science projects, as pointed out in Sec‐tion 4.2. Design science is not action research – design science always creates or positions an artefact, which is not required in action research, see Sec‐tion 4.1. Design science is not a research paradigm – a design science project can benefit from both positivist and interpretivist approaches, as discussed in Section 12.4.
So, what is design science? The definition from Section 1.4 states that design science is the scientific study and creation of artefacts as they are developed and used by people with the goal of solving practical problems. Thus, design science is a way of studying artefacts. Design science takes a problem solving stance, starting from problems experi‐enced by people in practices and then trying to solve them. It does so by creating, positioning and repurposing artefacts that can function as so‐lutions to the problems. In this work, design science will include empiri‐cal studies like problem explication and evaluation, but this is not the whole story. In addition, design science includes problem solving in the form of artefact design.
Further Reading
Oates (2005)discusses researchparadigms in the context of infor‐mationsystemsandcomputing.Anintroductiontointerpretivismisgiven by Burr (2003). Vaishnavi and Kuechler (2004) discuss re‐searchparadigmsandtheirrelevancetodesignscience.Baskerville(2008)discusseswhatdesignscienceisnot.
126
References
Avison,D.andFitzgerald,G.(2006).InformationSystemsDevelopment:Methodologies,Techniques&Tools.4thed.McGraw‐Hill.
Baskerville,R.(2008).Whatdesignscienceisnot.EuropeanJournalofIn‐formationSystems,17(5),p.441–443.
Beck,K.andBeedle,M.(2001).TheAgileManifesto.[Online].Availableat:http://agilemanifesto.org/[Accessed:16July2012].
Bennett,J.G.(1985).DramaticUniverse:ManandHisNaturev.3.Newed.ClaymontCommunications,U.S.
Bhattacherjee,A.(2012).SocialScienceResearch:Principles,Methods,andPractices.2nded.CreateSpace.
Boehm,B.W.(1981).SoftwareEngineeringEconomics.1sted.PrenticeHall.
Bolstorff,P.andRosenbaum,R.G.(2007).SupplyChainExcellence:AHand‐bookforDramaticImprovementUsingtheScorModel.AMACOMDivAmeri‐canMgmtAssn.
Booth,W.C.,Colomb,G.G.andWilliams,J.M.(2008).TheCraftofResearch.Thirded.UniversityOfChicagoPress.
deBono,E.(1973).LateralThinking:CreativityStepbyStep.HarperColo‐phon.
127
Brown,T.(2009).ChangebyDesign:HowDesignThinkingTransformsOr‐ganizationsandInspiresInnovation.HarperBusiness.
Burr,V.(2003).SocialConstructionism.2nded.Routledge.
CauseandEffectAnalysis(FishboneDiagrams)‐ProblemSolvingToolsfromMindTools.com.[Online].Availableat:http://www.mindtools.com/pages/article/newTMC_03.htm[Accessed:30June2012].
Cetina,K.K.,Schatzki,T.R.andSavigny,E.von(eds.).(2001).ThePracticeTurninContemporaryTheory.Routledge.
Checkland,P.(1999).SystemsThinking,SystemsPractice:Includesa30‐YearRetrospective.1sted.Wiley.
Chesbrough,H.W.(2005).OpenInnovation:TheNewImperativeforCreat‐ingAndProfitingfromTechnology.FirstTradePapered.HarvardBusinessReviewPress.
Clements,P.,Kazman,R.andKlein,M.(2001).EvaluatingSoftwareArchitec‐tures:MethodsandCaseStudies.1sted.Addison‐WesleyProfessional.
Cohn,M.(2009).SucceedingwithAgile:SoftwareDevelopmentUsingScrum.1sted.Addison‐WesleyProfessional.
Cross,N.(2011).DesignThinking:UnderstandingHowDesignersThinkandWork.BergPublishers.
Day,J.D.andZimmermann,H.(1983).TheOSIreferencemodel.Proceed‐ingsoftheIEEE,71(12),p.1334–1340.
Delone,W.H.andMcLean,E.R.(2003).TheDeLoneandMcLeanModelofInformationSystemsSuccess:ATen‐YearUpdate.JournalofManagementInformationSystems,19(4),p.9–30.
Denscombe,M.(2010).TheGoodResearchGuide:forsmall‐scalesocialre‐searchprojects.4thed.OpenUniversityPress.
Dix,A.(2007).Designingforappropriation.In:Proceedingsofthe21stBrit‐ishHCIGroupAnnualConferenceonPeopleandComputers:HCI...butnotasweknowit‐Volume2,BCS‐HCI’07,2007,Swinton,UK,UK:BritishCom‐puterSociety,p.27–30.
128
Dix,A.(2009).HCI3e‐Ch9:Evaluationtechniques.Technology.[Online].Availableat:http://www.slideshare.net/alanjohndix/hci‐3e‐ch‐9‐evaluation‐techniques[Accessed:1July2012].
Dixit,A.(2011).The‘IMRAD’format:aconciseandclearwaytopresentascientificstudy.IndianJournalofMedicalSpecialities.[Online].Availableat:http://www.ijms.in/articles/2/2/the‐imrad‐format‐a‐concise‐and‐clear‐way‐to‐present‐a‐scientific‐study.html[Accessed:1July2012].
Eisenhardt,K.M.(1989).Agencytheory:Anassessmentandreview.TheAcademyofManagementReview,14(1),p.57–74.
Gibbons,A.S.andBunderson,C.V.(2005).Explore,Explain,Design.In:EncyclopediaofSocialManagement,Elsevier,p.927–938.
Goldkuhl,G.(2012).FromActionResearchtoPracticeResearch.Austral‐asianJournalofInformationSystems,17(2).
Gregor,S.(2006).Thenatureoftheoryininformationsystems.MISQuar‐terly,30(3),p.611–642.
Gregor,S.andJones,D.(2007).TheAnatomyofaDesignScienceTheory.JournaloftheAssociationforInformationSystems,8(5),p.312–335.
Hevner,A.R.,March,S.T.,Park,J.andRam,S.(2004).Designscienceininformationsystemsresearch.MISQuarterly,28(1),p.75–105
Horkheimer,M.(1975).CriticalTheory:SelectedEssays.1sted.ContinuumPublishingCorporation.
Kroll,P.andKruchten,P.(2003).TheRationalUnifiedProcessMadeEasy:APractitioner’sGuidetotheRUP:APractitioner’sGuidetotheRUP.1sted.Addison‐WesleyProfessional.
Kuhn,T.S.(2012).TheStructureofScientificRevolutions:50thAnniversaryEdition.Fourthed.UniversityOfChicagoPress.
Madhavan,R.andGrover,R.(1998).FromEmbeddedKnowledgetoEmbod‐iedKnowledge:NewProductDevelopmentasKnowledgeManagement.Journalofmarketing:Aquarterlypublicationoftheamericanmarketingassociation,62(4),p.1–12.
March,S.T.andSmith,G.F.(1995).Designandnaturalscienceresearchoninformationtechnology.DecisionSupportSystems,15(4),p.251–266.
129
Moody,D.(2003).TheMethodEvaluationModel:ATheoreticalModelforValidatingInformationSystemsDesignMethods.ECIS2003Proceedings.
Moody,D.L.(2005).Theoreticalandpracticalissuesinevaluatingthequali‐tyofconceptualmodels:currentstateandfuturedirections.Data&Know‐ledgeEngineering,55(3),p.243–276.
Nussbaum,M.C.(2007).FrontiersofJustice:Disability,Nationality,SpeciesMembership(TannerLecturesofHumanValues.BelknapPressofHarvardUniversityPress.
Oates,B.J.(2005).ResearchingInformationSystemsandComputing.1sted.SagePublicationsLtd.
Österle,H.,Becker,J.,Frank,U.,Hess,T.,Karagiannis,D.,Krcmar,H.,Loos,P.,Mertens,P.,Oberweis,A.andSinz,E.J.(2010).Memorandumondesign‐orientedinformationsystemsresearch.EuropeanJournalofInformationSystems,20(1),p.7–10.
Peffers,K.,Tuunanen,T.,Rothenberger,M.A.andChatterjee,S.(2007).ADesignScienceResearchMethodologyforInformationSystemsResearch.JournalofManagementInformationSystems,24(3),p.45–77.
Pohl,K.(2010).RequirementsEngineering:Fundamentals,Principles,andTechniques.1sted.SpringerPublishingCompany,Incorporated.
Polya,G.(1973).HowtoSolveit.2nded.OpenUniversityPress.
Pries‐Heje,J.,Baskerville,R.andVenable,J.(2008).StrategiesforDesignScienceResearchEvaluation.ECIS2008Proceedings.
Ritchey,T.(2007).WickedProblems‐StructuringSocialMesseswithMor‐phologicalAnalysis.[Online].Availableat:http://www.swemorph.com/pdf/wp.pdf[Accessed:30June2012].
Sein,M.,Henfridsson,O.,Purao,S.,Rossi,M.andLindgren,R.(2011).ActionDesignResearch.ManagementInformationSystemsQuarterly,35(1),p.37–56.
Simon,H.A.(1996).TheSciencesoftheArtificial‐3rdEdition.thirded.TheMITPress.
Sommerville,I.andSawyer,P.(1997).RequirementsEngineering:AGoodPracticeGuide.1sted.NewYork,NY,USA:JohnWiley&Sons,Inc.
130
Sörensen,C.(2002).ThisisNotanArticle.[Online].Availableat:http://mobility.lse.ac.uk/download/Sorensen2005b.pdf[Accessed:27July2012].
Vaishnavi,V.andKuechler,W.(2004).DesignScienceResearchinInfor‐mationSystems.[Online].Availableat:http://www.desrist.org/desrist/[Accessed:27July2012].
Venkatesh,V.andDavis,F.D.(2000).ATheoreticalExtensionoftheTech‐nologyAcceptanceModel:FourLongitudinalFieldStudies.Manage.Sci.,46(2),p.186–204.
Walls,J.G.,Widmeyer,G.R.andElSawy,O.A.(2004).AssessingInfor‐mationSystemDesignTheoryinPerspective:HowUsefulWasour1992InitialRendition?JournalofInformationTechnologyTheoryandApplication,6(2),p.43–58.
Weber,R.(2004).Editor’scomments:therhetoricofpositivismversusin‐terpretivism:apersonalview.MISQuarterly,28(1),p.iii–xii.
Wikipediacontributors.(2012).Knowledge.Wikipedia,thefreeencyclope‐dia,WikimediaFoundation,Inc.[Online].Availableat:http://en.wikipedia.org/w/index.php?title=Knowledge&oldid=502162616[Accessed:15July2012].
131
Appendix
Template for a Design Science Thesis
Thisappendixprovidesatemplatethatcanbeusedforstructuringadesignsciencethesisatbachelor,masterorPhDlevel.Thefirstpartsofthetemplateaddressthegeneralbackgroundandmethodology.Itthenmoveson to the fivedesignscienceactivitiesandendswithapartondiscussionandconclusions. Insomesense,adesignsciencethesiscanbeviewedas includingfivemini theses,oneforeachde‐signscienceactivity.Thus,thetemplateincludesfiveparts,PartIVtoPartVIII,correspondingtotheseactivities,whereeachpartaddress‐es its own research question. However, inmany theses, some partcanbeomittedand/ortwopartscanbemergedintoasinglechapter.Forexample,PartIVandpartVcanbemergedintoachapter“Prob‐lemandRequirements”;PartVIcanformaseparatechapter“Devel‐opment”;PartVII canbeomitted; andPartVIII can formachapterentitled “Evaluation”. In this case, Part IV to Part VIII will be ad‐dressedinonlythreechapters.Otherpartscanalsobemergedintoonechapter,andsometimesonepartcangiverise toseveralchap‐ters.
132
Abstract
Theabstractistosummarizethemainpointsofthethesis.Itshouldbrieflydescribethebackground,practicalproblemtobeaddressed,goal,choiceofmethods,applicationofmethods,results,andconclu‐sions. The abstract should be self‐contained so that it can be readindependently from the rest of the thesis. The abstract should beconcisebutyetdetailedenoughthatareadercandecidewhetherthepaperisofinteresttoher.
Part I – Introduction
Background
Thebackground section is toprovide concise information requiredforunderstandingtheproblemandthegoalofthethesis.Theback‐ground typically includesadescriptionof thepractice inwhich theproblemarises.
Problem
Theproblemsection is todescribethepracticalproblemthatmoti‐vatesthethesis.Apracticalproblemisoftenasituationthatinvolvesorcausessignificantdifficulties,disadvantagesorriskstopeopleororganizations,suchaspeoplebeingexposedtohealthhazards,busi‐nesseslosingmoneyorcitizensreceivingpoorservicefromgovern‐mentagencies.Apracticalproblemcanalsobeaboutnewpossibili‐ties,e.g.howtabletscouldbeusedinhealthcare.Aproblemshouldbe of general interest. The problem should be described in such away that the reader becomes convinced that it is important to ad‐dress it.Anexampleofaproblemis“Manyelderlyexperiencediffi‐culties inusinghigh‐speedscanningsystemsforself‐serviceingro‐cerystoresandrefrainfromusingthem.”
133
Goal
Thegoalsectionistoformulatethegoalofthethesis.Agoalcanbeformulated as a question, e.g. "How can an architecture for high‐speed scanning systems for self‐service in grocery stores be de‐signedso that the systemsare suitable for theelderly?"Agoal canalsobemoredirectlyformulated,e.g."Thegoalistodesignanarchi‐tecture for high‐speed scanning systems for self‐service in grocerystoresthataresuitablefortheelderly."Thegoalshouldrelatetotheproblem,i.e.theproblemshouldbeaddressedbyachievingthegoal.Agoalshouldbeofgeneral interest, i.e. itshouldbeinterestingnotonlyforalocalpractice.Thesectionshouldmakeclearwhattypeofartefactistobedeveloped,suchasamodeloramethod.
Part II Extended Background
This part should include two sub‐parts, The first sub‐part shouldinclude a description ofwhat knowledge from the knowledgebasethathasbeenusedasabasis for theresearchcarriedout. These‐cond sub–part should include a description and discussion of re‐searchthatisrelatedtotheresearchpresentedinthethesis.Differ‐ences between thework of the thesis and relatedwork should bedescribed. Part II should give the reader basic background infor‐mation for understanding the research presented in the thesis aswellasabaseforvalidatingtheoriginalityofthepresentedresearch.
Part III Method
Thispartistofirstexplainwhydesignscienceisanappropriateap‐proach for the thesis. It should then provide a description of themethods used, particularly the research strategies and researchmethodschosen.Ifcreativemethodsorsystemsdevelopmentmeth‐odshavebeenused,theseshouldalsobedescribed.Allchoicesshallbe justified, and the advantages and disadvantages of the choicesshallbedescribed.Thepart isalso todiscussalternativesandwhythesewerenotused.Relevantethicalconsiderationsaretobemade.
134
Part IV Explicate Problem
Thispartistopreciselyformulatetheinitialproblemthatwasintro‐duced inPart I,explain itssignificanceandpossiblyanalyze itsun‐derlying causes. This part can sometimes be omitted if the initialproblemissufficientlywelldescribedandanalyzedinPartI.Howev‐er,therearealsothesesinwhichproblemexplicationisalargeandimportantpart.
Research question
This section is to formulate the research question that forms thebasisoftheproblemexplication.Thiscanbeexpressedasavariationofthefollowingquestion:"Whatistheproblemexperiencedbysomestakeholdersofapracticeandwhyisitimportant?"
Method Application
Iftheproblemexplicationwasbasedonaliteraturestudy,thissec‐tionshouldspecifythedatabasesandsearchqueriesthatwereused.If an empirical study was undertaken, the section should describehowthechosenresearchstrategiesandmethodswereapplied.Forexample, if interviewswereused fordatageneration, thedesignoftheinterviewquestionsshouldbediscussedaswellastheset‐upoftheinterviewsituation.Furthermore,thesectionistoshowthatthemethodapplicationiseffective,i.e.thatithelpsanswertheresearchquestionofthispart.
Results and analysis
This section is to describe and analyze the results of the problemexplication.Conclusionsaredrawnthatformthebasisforansweringtheresearchquestionofthispart.Thecertaintyoftheconclusionsisto be discussed, for example in terms of validity, reliability, andtransferability.
Part V: Outline Artefact and Define Requirements
This part is to outline a solution to the explicated problem in theformofanartefactandidentifytherequirementsforthisartefact.
135
Research question
This section shall formulate the research question that forms thebasis for the requirements definition. This can be expressed as avariationonthefollowingquestion:"Whatartefactcanbeasolutionfortheexplicatedproblemandwhichrequirementsonthisartefactareimportantforthestakeholders?"
Method Application
SeePartIVabove.
Results and analysis
SeePartIVabove.
Part VI Design and Develop Artefact
Thissection istodescribeboththedevelopedartefactandthepro‐cessfollowedinitsdevelopment.Ifthethesisonlyevaluatesanarte‐fact,thispartonlyincludesadescriptionofit.
Development process
Thissectionistodescribehowthedevelopmentoftheartefactwascarried out, particularly themethods used. The use of systems de‐velopmentmethods,creativemethodsaswellasresearchstrategiesandresearchmethodsistobedescribedhere.Designdecisionsandthereasonsforthese(designrationale)aretobediscussed.
Artefact Description
Thissection is todescribe thedevelopedartefact.The functionalityandconstructionoftheartefactshouldbedescribed.
Part VII Demonstration
Thispart istoshowhowthedevelopedartefactcanbeusedinonecase.Thepartisomittedifnodemonstrationwascarriedout.
Research question
This section shall formulate the research question that forms thebasisforthedemonstration.Thiscanbeexpressedasavariationon
136
thefollowingquestion:"Howcanthedevelopedartefactbeusedtoaddresstheexplicatedprobleminonecase?"
Method Application
This section is to describe how the chosen research strategies andresearchmethodswereapplied.Furthermore,thesectionistoshowthatthemethodapplicationiseffective, i.e. that ithelpsanswertheresearchquestionofthispart.
Results and analysis
SeePartIV.
Part VIII Evaluation
Thispart is toevaluate thedevelopedartefact,addressingboth thedefinedrequirementsandtheexplicatedproblem.
Research question
This section is to formulate the research question that forms thebasisfortheevaluation.Thiscanbeexpressedasavariationonthefollowingquestion:"Howwelldoestheartefactsolvetheexplicatedproblemandfulfilthedefinedrequirements?"
Method Application
If an empirical study was undertaken, the section should describehowthechosenresearchstrategiesandmethodswereapplied.Fur‐thermore,thesectionistoshowthatthemethodapplicationiseffec‐tive,i.e.thatithelpsanswertheresearchquestionofthispart.
Results and analysis
SeePartIV.
Part IX Discussion
Thissectionistosummarizethefindingsandconclusionsofthepre‐cedingparts.Adiscussionoffutureresearchisalsotobeincludedinthispart.Significanceandoriginality is tobediscussed.Ethicalandsocietalimplicationsofusingtheartefactshouldbediscussed.
137
References
Thereferencelistshouldbeformattedaccordingtosomeestablishedreferencesystem.
Appendices
Not mandatory. An appendix includes material that supports thethesis,butforsomereasonitisnotappropriatetobeincludedinthemain text. An appendix may include questionnaire forms and rawdata.