(1994, Second edition. Thousand Oaks: Sage) By Robert K. in Su!!ary General characteristics "hen to use case study !ethod# 1) The ty$e o% research &uestion: ty$ica''y to ans(er &uestions 'ike )ho(* or )(hy* +) ,-tent o% contro' o.er beha.ioura' e.ents: (hen in.estigator has a 'itt'e/no $ossibi'ity to contro' the e.ents 0) 1enera' circu!stances o% the $heno!enon to be studied: conte!$orary $heno!enon in a rea'2'i%e conte-t 3ase study is an e!$irica' in&uiry, in (hich: 2Focus is on a conte!$orary $heno!enon (ithin its rea'2'i%e conte-t 4 boundaries bet(een $heno!enon and its conte-t are not c'ear'y e.ident Suitab'e %or studying co!$'e- socia' $heno!ena 2Procedural characteristics in the situation inc'ude: 5any .ariab'es o% interest6 !u'ti$'e sources o% e.idence6 theoretica' $ro$ositions to guide the co''ection and ana'ysis o% data 2Types of case studies !ight be: e-$'anatory6 e-$'oratory6 descri$ti.e 2Designs can be sing'e2 or !u'ti$'e2case studies 2Used methods can be &ua'itati.e, &uantitati.e, or both Ty$ica' criticis!s to(ards case studies 4 correcting ans(ers: 27ack o% syste!atic hand'ing o% data 28 Syste!atic re$orting o% a'' e.idence 29o basis %or scienti%ic genera'i:ation 28 ;ur$ose is to genera'i:e to theoretica' $ro$ositions, not to $o$u'ation as in statistica' research 2Take too 'ong, end u$ (ith unreadab'e docu!ents 28 Ti!e 'i!its 4 (riting %or!u'a de$end on the choices o% in.estigators Case study research design 3entra' co!$onents o% a case study design 4 their %unctions: 1) < study=s &uestions > )ho(*, )(hy* +) Study=s (theoretica') $ro$ositions > $ointing attention, 'i!iting sco$e, suggesting $ossib'e 'inks bet(een $heno!ena 0) Study=s units o% ana'ysis > !ain units !ust be at the sa!e 'e.e' as the study &uestions 4 ty$ica''y co!$arab'e to those $re.ious'y studied 4) 7ogic 'inking the data to the $ro$ositions > !atching $ieces o% in%or!ation to ri.a' $atterns that can be deri.ed %ro! the $ro$ositions ?) 3riteria %or inter$reting the %indings > iteration bet(een $ro$ositions and data, !atching su%%icient'y contrasting ri.a' $atterns to data6 there is no $recise (ay o% setting the criteria 1 Research design 'inks the data to be co''ected and conc'usions to be dra(n to the initia' &uestions o% the study > it $ro.ides a conce$tua' %ra!e(ork 4 an action $'an %or getting %ro! &uestions to set o% conc'usions. Preliminary theory & blueprint of the study: ha.ing so'.ed the %i.e ste$s !entioned abo.e 'eads to the %or!ation o% a 'oose theory 4 a b'ue$rint re'ated to the to$ic o% study 2nitial theory & understanding of !hat is being studied is necessary before any field contacts6 the co!$'ete research design e!bodies a )theory* o% (hat is being studied Deciding bet!een e"planatory# e"ploratory and descripti$e designs% 2@e$ends on the richness o% the ri.a' $ro$ositions in theories re'ated to the to$ic o% the study6 richest theories a''o( e-$'anatory designs 28 Search %or theoretica' $ro$ositions that can be e'aborated to co.er study &uestions, $ro$ositions, units o% ana'ysis, data2$ro$osition 'inks 4 criteria o% inter$retation 28 Re.ie(ing 'iterature, discussing (ith in.estigators, asking cha''enging &uestions, thinking (hat is to be 'earned %ro! the study 28 Being a(are o% the range o% theories 4 se'ecting the re&uired 'e.e' (indi.idua', organi:ation, societa') 28 3onstruction o% the design / conce$tua' %ra!e(ork takes ti!e 4 can be di%%icu't, but is a crucia' ste$ %or the success o% the study Selection of the cases: 23ases shou'd be se'ected in the sa!e (ay as the to$ic o% an e-$eri!ent is se'ected 28 @e.e'o$ed $re'i!inary theory is used as a te!$'ate (ith (hich to co!$are the characteristics 4 e!$irica' %indings %ro! the case(s) 28 Se'ected cases shou'd re%'ect characteristics 4 $rob'e!s identi%ied in the under'ying theoretica' $ro$ositions / conce$tua' %ra!e(ork The le$el of generali&ation of the study results A a$$ro$riate'y de.e'o$ed $re'i!inary theory / study design '5ode o% genera'i:ation A theory2re'ated analytic generali&ation, not statistica' '<na'ytic genera'i:ation $ossib'e %ro! one or !ore cases BO" 5<9 OC 3<S,S 4 D9ETS OC <9<7SES# 1) Case > re$resents so!eho( the interesting to$ic o% the study e!$irica''y (e.g. the ro'e o% the United States in the (or'd econo!y) +) Unit of analysis > is the actua' source o% in%or!ation: indi.idua', organi:ationa' docu!ent, arti%act, %or e-a!$'e (e.g. the ca$ita' %'o( bet(een countries, an econo!ic $o'icy) 1) 3ases: <) Single cases > i% case see!s to re$resent a critica' test to e-isting theory6 rare or uni&ue e.ents 28 E!$ortant to se'ect case 4 unit o% ana'ysis $ro$er'y + B) Multiple cases > i% a )re$'ication 'ogic* is su$$osed to re.ea' su$$ort %or theoretica''y i. Si!i'ar resu'ts ii. 3ontrasting resu'ts %or $redictab'e reasons 28 Theoretica' %ra!e(ork !ust identi%y c'ear'y the conditions, (hen a $articu'ar $heno!enon is 'ike'y to be %ound (i.) and (hen it is not 'ike'y (ii.) 28 Theoretica' %ra!e(ork is the .ehic'e %or genera'i:ing to ne( cases6 i% e!$irica' cases do not (ork as $redicted, !odi%ications !ust be !ade to the theory 28 9u!ber o% case re$'ications de$ends u$on the certainty (anted to achie.e 4 richness o% the under'ying theoretica' $ro$ositions +) Dnits o% ana'ysis: <) (olistic designs > inc'ude a sing'e unit o% ana'ysis6 i% ai! is to study the g'oba' nature o% the $heno!enon6 (hen no 'ogica' sub2units can be $ointed 28 danger o% abstractness B) )mbedded designs > inc'ude !u'ti$'e units o% ana'ysis6 study !ay inc'ude !ain 4 s!a''er units on di%%erent 'e.e's 28 'ooking %or consistent $atterns o% e.idence across units, but (ithin a case 9ote: The fle"ibility o% case study design is in se'ecting cases di%%erent %ro! those initia''y identi%ied, not in changing the $ur$ose or obFecti.es o% the study to suit the cases Conducting case studies @esired ski''s o% the in.estigator: 21ood kno('edge o% the $heno!enon (28co''ection $rocedures are not routini:ed) 2Sensiti.ity %or no.e' 4 une-$ected issues in data co''ection 2<sking good &uestions 2Being a good )'istener* 2<da$ti.eness 4 %'e-ibi'ity * case study protocol shou'd be essentia' $art o% e.ery case study $roFect ' < $rotoco' contains the instru!ent %or the research, but a'so the $rocedures and genera' ru'es that shoud be %o''o(ed using the instru!ent: ' O.er.ie( o% the study $roFect (obFecti.es, issues, readings, 'iterature 4 research) ' Cie'd $rocedures (access to %ie'd sites, sources o% in%or!ation) ' 3ase study &uestions $osed to E9G,STE1<TORS6 key c'assi%ications6 suggestions %or 'ike'y sources o% e.idence (not the 'itera' &uestions to be asked) ' < guide %or the case study re$ort ' 9ote: 3ase study $rotoco' shou'd be co2authored by a'' study in.estigators E!$ortant to identi%y di%%erent 'e.e's o% &uestions: <) Sing'e case 2'e.e' 1. Huestions asked o% s$eci%ic inter.ie(ees 0 +. Huestions asked o% the indi.idua' case B) 5ore genera' 'e.e's 1. Cindings across !u'ti$'e cases +. Cindings across an entire study (inc'uding re.ie(ed 'iterature) 0. 9or!ati.e &uestions about $o'icy reco!!endations 4 conc'usions So!eti!es a $i'ot study !ay $ro.ide use%u' he'$ Collecting the evidence Si- sources o% e.idence: 1) @ocu!ents ('etters, agendas, $rogress re$orts) +) <rchi.a' records (Ser.ice records, organi:ationa' charts, budgets etc.) 0) Enter.ie(s (ty$ica''y o$en2ended, but a'so %ocused, structured 4 sur.eys are $ossib'e) 4) @irect obser.ations (%or!a' or casua'6 use%u' to ha.e !u'ti$'e obser.ers) ?) ;artici$ant obser.ation (assu!ing a ro'e in the situation 4 getting an inside .ie( o% the e.ents) I) ;hysica' arte%acts Three $rinci$'es o% data co''ection: +, Use multiple sources of e$idence 2Triangu'ation A searching con.erging %indings %ro! di%%erent sources 28 increases construct .a'idity -, Create a case study database 2< database se$arate %ro! the %ina' re$ort to be (ritten, containing ' 3ase study notes (c'ear 4 a.ai'ab'e %or 'ater use) ' 3ase study docu!ents ' Tabu'ar !ateria's (co''ected 4 created) ' 9arrati.es (initia' o$en2ended ans(ers to the study &uestions suggested by in.estigators) ., Maintain a chain of e$idence '/ The 'ink bet(een initia' study &uestions and case study $rocedure shou'd be $ointed out in the case study $rotoco', as a'so the circu!stances o% the e.idence to be co''ected '/ ;utting the data co''ection to $ractice on the basis o% the $rotoco' 28 <ctua' e.idence storage in the database %or 'ater checks (s$eci%ic co''ection circu!stances indicated) 28 Su%%icient citing o% the case study data base 4 e.idence in the %ina' re$ort and conc'usions to be dra(n Analysing the evidence <na'ysis o% e.idence is one the 'east de.e'o$ed and !ost di%%icu't as$ects o% doing case studies. 5ost i!$ortant is to ha.e a genera' ana'ytic strategy, (hich he'$s to choose a!ong di%%erent techni&ues. En absence $re'i!inary techni&ues > !atri-es, 4 tabu'ation o% %re&uencies, te!$ora' sche!es etc. > can be tried out to get the ana'ysis started. T!o general analytic strategies: 1) Re'ying on theoretical propositions: theoretica' orientation guiding the ana'ysis6 %o''o(ing theoretica' $ro$ositions that ha.e %or!ed the design o% the case study 28 he'$s to %ocus attention on certain data 4 to ignore other data +) @e.e'o$ing a case description: a descri$ti.e %ra!e(ork %or organi:ing the case study6 ana'ysis organi:ed on the basis o% descri$tion o% the genera' characteristics and re'ations o% the $heno!enon in &uestion *nalytic techni0ues to be used as $art o% the genera' strategy: 1) Pattern matching (e-$'anatory / descri$ti.e) 23o!$aring e!$irica''y based $atterns (ith $redicted one(s) a) Expected outcomes as a $attern: co!$aring i% the initia''y $redicted resu'ts ha.e been %ound and a'ternati.e $atterns are absent b) Rival explanations as $atterns: searching i% so!e o% the theoretica''y sa'ient e-$'aining conditions !ight be articu'ated in e!$irica' %indings6 then the $resence o% certain e-$'anation shou'd e-c'ude the $resence o% others c) Simpler patterns: $attern !atching is $ossib'e a'so (ith on'y %e( .ariab'es, i% the deri.ed $atterns are $redicted to ha.e enough c'ear di%%erences +) Explanation-building (!ain'y e-$'anatory) 2<na'y:ing case study data by bui'ding an e-$'anation about the case 4 identi%ying a set o% causa' 'inks 2,-$'anation is a resu't o% series o% iterations: Enitia' theoretica' state!ent 28 3o!$aring %indings o% an initia' case 28 re.ising state!ent 28 3o!$aring detai's o% the case 28 Re.ising 28 3o!$aring to other additiona' cases 29ote: @anger o% dri%ting a(ay %ro! origina' to$ic o% interest 0) Time-series analysis 2)Bo(*2 and )(hy*2 &uestions about re'ationshi$s 4 changes o% e.ents o.er ti!e 2Edenti%ying theoretica''y $ro$osed se&uences o% an e.ent that are e-$ected to 'ead to a certain outco!e6 identi%ication o% e.ents !ust be done be%ore the onset o% the in.estigation 283o!$aring this trend (ith the trend o% e!$irica' data $oints 28 3o!$aring (ith so!e ri.a' trend to ru'e a'ternati.es out <na'ysis shou'd sho( that: Et re'ied on a'' the re'e.ant e.idence6 a'' !aFor ri.a' inter$retations are dea't6 !ost signi%icant issue o% the study is addressed6 $rior e-$ert kno('edge is brought to the study ?