The  Transformation  of  Accident  Investigation    From  Finding  Cause  to  Sensemaking             Proefschrift  ter  verkrijging  van  de  graad  van  doctor  aan  Tilburg   University  op  gezag  van  de  rector  magnificus,  prof.dr.  E.H.L.  Aarts,  in   het  openbaar  te  verdedigen  ten  overstaan  van  een  door  het  college   voor  promoties  aangewezen  commissie  in  de  Ruth  First  zaal  van  de   Universiteit  op  dinsdag  1  september  2015  om  10.15  uur   door   Ivan  Alexander  Pupulidy   geboren  op  30  mei  1958  te  Weissenburg,  Duitsland Promotores:     Prof.dr.  J.B.  Rijsman             Prof.dr.ir.  G.  van  Dijk                 Copromotor:     Dr.  D.  Whitney     Overige  leden     promotiecommissie:     Prof.dr.  J.  Goedee             Prof.dr.  L.  Bibard             Prof.dr.  J.A.J.  Luijten             Prof.dr.em.  E.  Schein             Prof.mr.  L.  Witvliet             Dr.  P.  Spierings Table  of  Contents       Abstract  ...........................................................................................................................................  i   Acknowledgements  ........................................................................................................................  ii   Introduction  ...................................................................................................................................  1   Chapter  1:  History  of  the  USDA  Forest  Service  ..............................................................................  9   Chapter  2:  The  Evolution  of  Accident  Investigation  .....................................................................  27   Chapter  3:  The  Serious  Accident  Investigation  Guide—Pressure  to  Standardize  the  Investigation   Process  .........................................................................................................................................  40   Chapter  4:  The  Norcross  Case  Study  ............................................................................................  54   Chapter  5:  The  Panther  Case  Study  .............................................................................................  68   Chapter  6:  The  Importance  of  Sensemaking  Communities  to  Accident  Prevention  ....................  87   Chapter  7:  Learning  from  Error  ....................................................................................................  95   Chapter  8:  Agreeing  to  the  Concepts  of  the  Coordinated  Response  Protocol  and  Learning   Review  ........................................................................................................................................  111   Chapter  9:  The  Saddleback  Case  Study  ......................................................................................  120   Chapter  10:  Reflections  on  Transformation  through  the  Lens  of  Social  Construction  ...............  149   Chapter  11:  Summary  and  Conclusions  .....................................................................................  162   Bibliography  of  References  ........................................................................................................  185   Appendix  ....................................................................................................................................  199 TRANSFORMATION  OF  ACCIDENT  INVESTIGATION  FROM  FINDING  CAUSE  TO  SENSEMAKING   Abstract   This  dissertation  introduces  a  network  of  practices  that  transformed  the  United  States   Department  of  Agriculture  (USDA)  Forest  Service  accident  investigation.  This  is  an   exceptionally  important  topic  to  the  Forest  Service  for  several  reasons.  First,  the  Chief  of   the  Forest  Service  has  committed  to  creating  a  “zero  fatality  organization,”  and  the   organizational  response  to  accidents  is  believed  to  play  a  significant  role  in  achieving  this   goal  (Tidwell,  2013).  Second,  the  previous  method  of  investigation  created  second   victims;  these  were  workers  who  were  blamed  or  cited  as  having  caused  the  accident.   This  outcome  was  not  intentional;  however,  the  process  demanded  the  identification  of   cause,  and  cause  was  translated  into  blame.  Third,  the  linear  traditional  method  of   investigation  was  overly  simplistic  and  eroded  the  confidence  that  the  workforce  had   concerning  the  organization.  Fourth,  the  fatality  accident  rate  for  wildland  firefighting   operations  was  “unacceptable”  (Tidwell,  2013)—the  wildland1  firefighting  community  lost   1,075  firefighters  between  1910  and  2014  (this  number  does  not  include  off  duty  deaths).     Under  the  traditional  method  of  accident  investigation,  the  accident  rate  increased.     This  dissertation  uses  case  studies  to  show  the  interweaving  of  organizational  and   individual  journeys,  each  of  which  began  with  the  strength  to  inquire  and  to  challenge   assumptions.  The  case  studies  show  how  constructed  realities,  including  my  own,  were   challenged  through  inquiry  and  how  four  practices  emerged  that  supported  sensemaking   at  both  the  field  and  organizational  leadership  levels  of  the  organization.  The  application   of  a  single  one  of  these  practices  can  improve  investigative  processes;  however,  as  the   last  case  study  demonstrates,  together  they  form  a  network  that  transformed  Forest   Service  investigations.   There  is  also  a  realization  that  this  was—and  in  many  ways—still  is  a  learning  journey.  The   process  of  change  spanned  eight  years  and  the  journey  is  not  yet  complete.  In  that  eight-­‐ year  period,  the  Forest  Service  has  accepted  new  processes  for  the  review  of  accidents   and  incidents.  The  Learning  Review  process,  which  replaced  accident  investigation,   embraces  four  practices  designed  to  engage  a  wide  range  of  participants  through   targeted  learning  products.  Where  we  used  to  construct  accident  investigation  reports  to   place  the  incidents  behind  us,  these  new  learning  products  are  designed  to  invigorate   communities  of  practice  to  discuss,  question,  and  explore  the  incidents  in  ongoing   dialogues  that  add  perspectives,  knowledge,  and  experience  in  order  to  develop   applicable  lessons  learned.   I  did  not  begin  my  journey  as  a  social  constructionist  and  only  discovered  this  orientation   once  I  was  well  along  the  path.  I  realized  almost  all  the  organizational  and  individual   transformation  practices  represented  the  application  of  constructionist  concepts.                                                                                                                               1  Wildland  firefighting  is  differentiated  from  structural  firefighting.  Wildland  fires  burn  in  forest  and   grasslands,  whereas  the  term  structural  firefighting  is  specific  to  houses  and  other  manmade  structures.   I v a n   P u p u l i d y     i TRANSFORMATION  OF  ACCIDENT  INVESTIGATION  FROM  FINDING  CAUSE  TO  SENSEMAKING   Acknowledgements   When  I  first  began  to  write  this  dissertation  I  did  not  understand  that  the  transformation  I   was  about  to  expose  was  as  much  about  me  as  it  was  about  the  USDA  Forest  Service.  I   learned  that  change  can  be  so  incremental  as  to  be  imperceptible.  I  also  learned  the   importance  of  an  emerging  sense  that  we  should  be  less  drawn  to  finding  specific   solutions  to  problems  in  complex  systems.  Instead,  the  focus  of  our  effort  should  be  to   facilitate  collaboration  across  hierarchical  boundaries  through  dialogues  that  result  in   creative  pathways  that  fit  specific  situations.  Then  we  must  dedicate  effort  to  reflect— only  in  reflection  can  we  learn.     I  want  to  thank  those  who  helped  conceptually,  spiritually,  and  technically;  that  list  would   likely  fill  the  page.  It  would  not  be  reasonable  to  continue  without  mentioning  some  of   the  most  critical  members  of  the  cast  of  people  who  supported  me  and  contributed  to   this  dissertation:  Crista,  my  wife,  best  friend,  and  editor;  Heather,  my  dear  Canadian   friend  and  conceptual  cohort;  John,  for  challenging  my  ideas  and  being  really  clever  about   it;    Ben,  for  beginning  his  journey  with  me  and  for  trusting;  Curtis,  Heath,  Wayne,  Gwen,   and  Jay  for  helping  to  bring  ideas  into  practice;  Sidney,  for  lighting  a  fire  that  would  not   go  out;  Todd,  for  changing  me  and  changing  with  me;  Diana,  for  holding  my  feet  to  the   fire;  and  every  person  who  challenged  my  ideas,  ran  small  experiments,  provided   feedback,  and  otherwise  agreed  or  disagreed—we  have  all  come  a  long  way.   There  is  a  greater  thanks  that  must  be  offered—to  my  mom—she  already  knows  what   she  did  to  inspire  the  drive  that  I  needed  just  to  get  this  thing  done!   There  is  another  group  of  people  who  deserve  the  utmost  recognition:  the  lost  friends   who,  in  the  pursuit  of  their  dreams,  were  with  me  at  dinner  one  night  and  were  gone  the   next.  Alongside  them  are  all  the  other  firefighters,  frontline  operators,  and  pilots  who   have  been  made  into  second  victims  by  a  process  from  which  justice  cannot  be  achieved   and  learning  is  of  little  or  no  interest.  In  particular  is  Pete,  whose  story  about  the  Thirty-­‐ Mile  investigation  was  etched  into  my  mind!       May  the  only  things  we  bury  be  “root-­‐cause”  and  “blame.”   I v a n   P u p u l i d y     ii TRANSFORMATION  OF  ACCIDENT  INVESTIGATION  FROM  FINDING  CAUSE  TO  SENSEMAKING   Introduction   Navigating  this  Journey     This  journey  can  be  described  as  moving  through  three  main  areas  of  study:  The  first   focused  on  how  the  USDA  Forest  Service  arrived  at  the  process  delineated  by  the  Serious   Accident  Investigation  Guide  (SAIG).  The  second  addressed  academic  research  that  could   be  used  to  frame  a  different  approach  to  investigation.  The  third  was  an  empirical   exploration  of  the  application  of  theory  and  practice  during  actual  investigations.  These   areas  of  study  are  viewed  through  the  lens  of  social  construction  (Chapter  10).   Why  was  transformation  needed?  Traditional  models  of  investigation  ignored  the  voices   of  participants,  communities  and  leaders  in  an  effort  to  resolve  the  event  to  a  single   narrative  that  made  sense  to  the  team.  Narratives,  while  espousing  to  be  unbiased,   offered  a  plausible  explanation  that  was  represented  as  a  factual  report.  The  stories   created  with  this  methodology  were  more  linear,  plausible  and  less  messy  than  the   complex  events  they  were  modeling  (Dekker  2002).  Often  lost  in  the  process  were   valuable  perspectives  that  offered  the  context  needed  for  those  outside  either  the  event   or  the  investigation  to  make  sense  of  the  event  themselves.  In  this  way,  learning  became   explicit  rather  than  transactional.  Facts  were  offered  through  reports  that  drew   conclusions,  made  assumptions  and  defined  cause  in  terms  of  the  judgment  of  actions   and  decisions,  leaving  little  room  for  individual  or  group  sensemaking.  However,  when   the  conclusions  of  these  investigations  were  deemed  to  be  a  surprise  or  well  outside  the   societal  construct  of  reality,  they  were  challenged.  This  dissertation  recounts  a  series  of   such  challenges.   The  case  studies  explored  in  this  dissertation  evoke  questions  that  could  not  be  explained   or  understood  using  the  formal  guidance  or  training  that  had  been  provided  to  me  in  the   SAIG  or  formal  accident  investigation  courses.  Together  the  training  and  written  guidance   formed  a  process  that  advocated  the  search  for  what  was  absent  in  the  system,   environment  or  people  involved  in  the  incident.  This  approach  avoids  consideration  of   positive  aspects  of  individual  performance  explored  through  positive  questions  (Whitney,   Trosten-­‐Bloom,  &  Cooperrider,  2010).  “Positive  questions  are  keys  to  treasure  troves  of   best  practices,  success  stories  and  creativity”  (Whitney,  Trosten-­‐Bloom,  &  Rader,  2010).     Each  accident  posed  unique  issues,  concerns,  and  opportunities  to  the  assigned  teams,   which  required  conversations  that  explored  contextual  influences,  adaptive  responses   and  interrelationships.  Actions  were  not  seen  as  negative  contributors  to  the  event,   rather  they  were  explored  as  the  best-­‐fit  solutions  that  were  developed  by  well-­‐meaning   individuals.  The  conversation  that  emerged  in  the  investigative  teams  was  open,   unscripted  dialogue,  largely  absent  the  prescriptive  causal  guidance  of  the  SAIG.    An   I v a n   P u p u l i d y     1 TRANSFORMATION  OF  ACCIDENT  INVESTIGATION  FROM  FINDING  CAUSE  TO  SENSEMAKING   alternative  approach  to  investigation  emerged  that  was  based  on  inquiry  and  advocated   for  sensemaking  and  learning  to  take  place  at  multiple  levels  of  the  organization.   My  Evolving  Role  in  Accident  Investigation   My  initiation  to  the  investigative  process  came  through  military  investigation  training,   where  I  was  fully  accredited  in  the  organizationally  approved  processes  common  to   traditional  accident  investigations.  It  was  in  the  Coast  Guard  that  I  received  my  first   experience  with  investigation,  a  helicopter  fatality.  In  these  early  years,  I  worked   diligently  to  bring  individual  flare  to  my  creations  and  created  factual  reports  (stories)   that  I  fervently  believed  would  result  in  corrections  and  fixes  to  specific  problems   uncovered  during  investigations.  I  also  believed  that  my  work  would  prevent  the  next   accident.  There  were  a  lot  of  statistics  that  seemed  to  point  to  success  and  served  to   reaffirm  that  the  process  was  working.  Moreover,  I  liked  what  I  did.     Some  context  is  therefore  needed  to  understand  why  I  became  compelled  to  move  away   from  this  path  and  to  influence  change  in  the  way  we  conducted  investigations.  I  was  a   Coast  Guard  pilot  for  10  years  and  during  that  time,  I  knew  three  people  who  died  in   aircraft  accidents.  The  Coast  Guard  touted  the  best  flight  safety  record  in  the  military,  and   statistics  proved  that  our  accident  rate  was  better  than  most  aviation  operations.  I   believed  that  it  was  our  actions  and  layered  defenses  that  were  delivering  these  great   results  and  that  the  investigation  process  produced  many  of  these  defenses.   After  the  Coast  Guard,  I  joined  the  Forest  Service  as  a  lead  plane  pilot.  Lead  plane  pilots   fly  low-­‐level  over  fires;  establish  tactics;  scout  routes  for  heavy  air  tankers  loaded  with   fire  retardant;  and  then  guide  them  to  the  drop  zone,  in  support  of  ground  fire   operations.  In  many  ways,  this  world  seemed  similar  to  that  of  the  Coast  Guard,  yet  I   would  learn  that  it  was  also  strikingly  different.     I  was  hired  in  May  with  a  report  date  of  August,  along  with  another  pilot.  I  would  learn   later  that  the  original  solicitation  asked  for  one  pilot,  but  there  had  been  a  mid-­‐air   collision  that  took  the  life  of  a  lead  plane  pilot,  and  thus  the  hiring  official  selected  a   second  applicant.  Clearly  one  of  us  was  replacing  this  fallen  comrade.       The  national  average  for  aviation  fatalities  for  the  Forest  Service  was  2.5  human  losses   per  year.  The  wildland  fire  statistics  were  telling  a  very  different  story  than  the  Coast   Guard  statistics.  The  investigation  reports  that  resulted  from  each  fatality  unilaterally   pointed  to  errors  on  the  part  of  the  flight  crews.  As  my  experience  grew,  I  began  to   realize  that  it  had  to  be  more  than  just  pilot  error—something  did  not  fit.     Each  aviation  fatality  represented  a  friend  to  me;  these  were  people  with  whom  I  had   dinner  one  night  and  were  gone  the  next  day.  Each  accident  report  listed  error  as  a  cause,   I v a n   P u p u l i d y     2 TRANSFORMATION  OF  ACCIDENT  INVESTIGATION  FROM  FINDING  CAUSE  TO  SENSEMAKING   often  implying  or  openly  stating  that  it  was  human  caused  and  citing  a  friend  at  fault.  I   knew  these  people  and  at  a  deep  level,  I  knew  it  wasn't  as  simple  as  error  on  their  part.   After  all,  I  had  made  mistakes,  and  I  was  often  doing  the  same  things  they  were.  This   fueled  a  fire  within  me  to  learn  more,  which  ultimately  grew  into  a  desire  to  change  the   system.     People  were  being  blamed  for  accidents  as  though  they  had  intended  to  crash,  and  in  that   way  the  process  was  creating  second  victims  (Dekker,  2013).  I  saw  good  people— suddenly  by  the  virtue  of  a  report—transformed  into  flawed,  error  prone,  risk-­‐takers  that   clearly  didn't  have  the  right  stuff.  One  day  they  were  upheld  as  heroes  for  successful   outcomes  like  saving  a  section  of  fireline,  or  a  house,  or  in  one  case  an  entire  town.  Days   later,  these  same  people  could  find  themselves  labeled  as  “rogue  pilots”  (Kern,  1999)   simply  because  they  were  involved  in  an  accident.       My  interest  in  safety  became  more  intense  with  each  fatality  and  after  a  few  years  I   became  a  regional  aviation  safety  manager  (RASM)  and  began  to  pursue  accident   investigation  as  a  collateral  duty.  I  completed  several  civilian  courses,  which  augmented   my  military  training,  and  following  a  seemingly  short  apprenticeship;  I  was  assigned  as  the   chief  investigator  to  the  Norcross  helicopter  fatality  accident.  The  incident  would  become   pivotal  to  my  own  growth,  as  well  as  that  of  the  Forest  Service.     I  went  to  the  incident  armed  with  all  the  latest  techniques,  tools,  and  the  most  recent   interagency  SAIG.  I  was  nervous  about  the  new  responsibility  and  carefully  reviewed  the   guide  contents  to  ensure  that  I  could  deliver  the  product  that  the  organization  desired.   What  I  found  was  that  the  guide  offered  too  much  help—step-­‐by-­‐step  instructions  that,  in   some  cases,  provided  conclusions  before  any  information  had  been  gathered.  The  guide   asked  me  to  view  the  incident  from  the  perspective  that  everything  is  knowable,   discoverable,  or  observable  and  all  I  had  to  do  was  to  look  harder,  deeper,  or  more   carefully  to  find  the  single  truth,  the  error.  The  SAIG  specifically  recommended  that   investigators  judge  human  actions  and  decisions  as  bad  or  good,  largely  based  on  the   assumption  that  there  had  to  be  a  violation  or  error  for  an  accident  to  occur  (Wiegmann   &  Shappell,  2003).       The  Norcross  accident  investigation,  as  will  be  explained  later,  inspired  me  to  inquire—as   I  began  to  inquire,  the  thin  veil  of  realism  began  to  rapidly  fall  away.  What  remained   challenged  almost  everything  I  had  been  taught  in  accident  investigation  training,   uprooted  the  principles  of  the  interagency  SAIG  and  shattered  my  belief  in  causality.  The   very  nature  of  these  reports  was  based  on  factual  accounts,  and  I  found  myself   challenging  the  very  existence  of  facts.   Three  realizations  paralleled  the  recognition  that  there  was  more  to  accidents  and   incidents  than  simply  finding  facts.  First,  I  realized  that  humans  are  naturally  biased  and   that  our  biases  influence  what  facts  we  find  or  create  (Kahneman,  Slovic,  &  Tversky,  1982;   I v a n   P u p u l i d y     3 TRANSFORMATION  OF  ACCIDENT  INVESTIGATION  FROM  FINDING  CAUSE  TO  SENSEMAKING   Kahneman  &  Klein,  2009).  For  example,  if  I  enter  into  an  investigation  to  find  error  I  will   find  it  (Hollnagel,  2008).  This  guidance  can  be  explicit  as  it  is  in  much  of  the  SAIG.  Or  it  can   be  implicit,  embedded  in  the  language  or  in  investigative  process  itself.  The  classic   example  of  this  is  root-­‐cause  analysis,  which  implies  that  there  is  a  single  or  root  cause,  a   truth  that  can  be  discovered  (Hollnagel,  2008;  Dekker,  2006).     My  second  realization  and  break  from  the  established  norm  was  an  understanding  that   time  is  a  significant  construction,  and  it  can  influence  judgments  in  ways  that  can  be   harmful  to  learning.  For  example,  time  is  easily  accepted  as  a  fact,  which  is  reinforced  by   the  way  time  is  incorporated  into  modern  society  as  a  measurable  entity.  However,  its   role  in  the  review  of  accidents  can  point  to  individual  human  failures  and  omit  important   context.  Statements  like  ‘it  took  five  minutes’  can  be  interpreted  in  a  number  of  ways— they  had  five  minutes;  they  only  had  five  minutes;  or  they  had  five  minutes!  Simple  time   references,  without  context,  can  be  meaningless  and  yet  can  result  in  judgments  that   affect  the  creation  and  interpretation  of  an  accident  report.   The  third  realization  was  that  the  same  adaptations  that  result  in  success  could  also  result   in  failure.  Our  culture  often  rewards  outside-­‐the-­‐box  thinking,  which  encourages   innovation  and  independence.  Our  heroes  are  often  those  who  buck  the  system  and   stand  as  outliers,  seemingly  ignoring  organizational  guidance  or  even  laws.  We  often   uphold  these  individuals  as  the  change  makers  in  our  society,  and  their  success  is   heralded.  However,  following  an  accident  or  failure  we  commonly  overlook  that  the  same   innovation  and  adaptation  can  also  lead  to  failure.  If  cause-­‐effect  exists,  then  each  action   should  deliver  the  same  effect—clearly  actions  delivered  a  myriad  of  outcomes  ranging   from  success  to  failure.  To  me  this  challenged  the  basis  of  the  cause-­‐effect  relationship,  a   central  principle  of  traditional  accident  investigation  processes.   My  accident-­‐investigation  role  evolved  rapidly,  as  I  began  to  recognize  and  capture  these   three  concepts  in  three  particular  fatality  investigations,  starting  with  Norcross,  then   Panther,  and  culminating  in  Saddleback.  Through  the  reports  and  dialogue  that  emerged   from  these  investigations,  Forest  Service  leaders  realized  the  importance  of  learning  from   events  and  began  to  tie  learning  to  prevention.  The  most  significant  shift  in  my  role   occurred  when  I  was  asked  to  develop  a  guide  to  replace  the  accepted  interagency  SAIG.   The  creation  of  what  became  known  as  the  Learning  Review  required  deep  personal   introspection  and  challenged  deep  assumptions  within  me.   Purpose   The  purpose  of  this  dissertation  is  to  demonstrate  how  the  USDA  Forest  Service  accident   investigation  process  was  transformed  from  finding  cause  to  sensemaking  and  learning.     I v a n   P u p u l i d y     4 TRANSFORMATION  OF  ACCIDENT  INVESTIGATION  FROM  FINDING  CAUSE  TO  SENSEMAKING   Case  Study  Format     When  I  consider  the  personal  and  organizational  transformation  described  in  this   dissertation,  I  only  see  it  as  a  story.  Human  beings  are  story-­‐telling  creatures—a  point   Fisher  (1987)  makes  when  he  bestows  the  title  homo  narans.  Stories  knit  together   settings,  actors,  events,  pressures,  conditions,  and  ethical  considerations.  As  such  they   can  be  an  intense  medium  to  help  people  make  sense  of  seemingly  related  or  unrelated   factors  (Schrader,  2004).  Understanding  the  importance  of  story  to  the  evolution  of  the   Learning  Review  and  presenting  it  in  writing  was  challenging.  The  linear  medium  of   writing  by  its  nature  makes  it  difficult  to  describe  non-­‐linear  events.  It  became  apparent   that  the  only  way  to  describe  this  story  was  to  explore  the  stories  that  contributed  to  the   transformation.   The  research  methodology  needed  to  explore  inter-­‐related  and  embedded  stories  must   be  capable  of  integrating  event,  activity,  progress,  and  influences  for  a  wide  variety  of   individuals.  Case  studies  emerged  as  a  qualitative  method  to  achieve  this  goal.  “Case   studies  are  a  strategy  of  inquiry  in  which  the  researcher  explores  in  depth  a  program,   event,  activity,  process  or  one  or  more  individuals”  (Creswell,  2009,  p.  13).  This  is   strengthened  by  the  idea  that  the  objects  of  a  case  study  must  be  “similar  enough  and   separate  enough  to  permit  treating  them  as  comparable  instances  of  the  same  general   phenomenon”  (Ragin  &  Becker,  1992,  p.2).     The  structure  of  the  case  study  method  also  allowed  for  the  emergence  of  concepts  that   would  result  from  the  recognition  of  connections  during  the  study  and  writing.  This   happened  on  several  occasions  during  the  creation  of  this  dissertation.  As  Ragin  &  Becker   (1992)  state,  “What  is  this  case  of  will  coalesce  gradually,  sometimes  catalytically,  and  the   final  realization  of  the  case’s  nature  may  be  the  most  important  part  of  the  interaction   between  ideas  and  evidence”  (p.  6).     Within  the  methodology  of  case  study  research  there  are  provisions  for  the  type  of   research  conducted.  “In  case  studies,  sampling  is  purposive.  They  will  be  most  instructive   when  they  are  methodologically  based  on  open  case-­‐sensitive  approaches  like  the   narrative  interview  and  ethnography”  (Flick,  2009,  p.  134).  Each  of  the  cases  used  in  the   dissertation  fit  this  description.  The  selection  of  case  study  format  for  this  dissertation   also  meets  the  intense  guidelines  for  case  study  research  described  by  George  and     Bennett  (2005).  These  criteria  are  described  in  three  parts  (2005,  p.  69):   First,  the  cases  must  all  be  instances  of…  only  one  phenomenon.  Second,  a   well-­‐defined  research  objective  and  appropriate  research  strategy  to   achieve  that  objective  should  guide  the  selection  and  analysis  of  the…cases   under  investigation.  Third,  case  studies  should  employ  variables  of   theoretic  interest  for  the  purpose  of  explanation.   I v a n   P u p u l i d y     5