last planner and integrated project delivery
TRANSCRIPT
Seongkyuand IntegJournal 2Delivery S
AbstrResearc
2) DIPD
Purposeper
Researcdetred
Findingandpersignimp
Limitatgen
Implicabetthe
Value foto d
KeyworPaper t
LiteraAccordi(IPD) co
1 PhD C
Berke2 Direct
Profes94720
Lean Cohttp://c
un Cho and Glegrated Project2011 pp 67-78 Special issue w
Last P
ract ch QuestionDoes Integr
D projects ue: The firstrformance. ch Method:termine theduction + timgs: 1) Thered project performance fnificantly dplementatioions: Limit
neralizationations: This tween projee other. or practitiodesign projerds: Integratype: Full p
ature Rng to the A
ontract form
Early invoShared ris
Candidate, Civeley, CA 94720tor, Project Pssor, Civil and
0-1712, USA, Ponstruction Jocreativecommons.
enn Ballard (2t Delivery Lea - Lean and Inwww.leancon
lanner Se
ns: 1) Doesated Projecse LP?
t objective i : survey of e correlatiome reductioe is significaerformancefrom that oifferent imon is near toations in sa
ns. exploratoryect structur
oners: The ect deliveryated Projec
paper.
Review American Insm includes:
lvement of k and rewa
vil and Env. En0-1712, USA, Production Sys
d Env. EngineePhone +1 415/
urnal 2011 .org/licenses/by-
2011) Last Plan Constructio
ntegrated Projstructionjourn
and Ineongkyun C
s the use of ct Delivery
is to figure
‘Lean’ projn between on); and a Tant correlate; 2) IPD prof others noplementatio significan
ample size a
y research res and prac
findings froy systems foct Delivery,
stitute of A
Key particiard;
ngineering. DePhone +1 510/tem Laboratoering. Departm/710-5531, ba
nc-nd/3.0/ p
nner on ject nal.org
ntegrateCho1 and G
Last Plann (IPD) show
out the rel
ects known LP implemeT test betwtion betweeojects do not adopting on of LP fro
nce and data qu
revealed inctices on th
om this papor better p Last Plann
Architects (A
ipants;
epartment, 40/725-7929, seoory, http://p2ment, 215-A [email protected]
page 67
ed Prolenn Ballar
er (LP) imp different p
lationship b
n to adopt Lentation an
ween IPD anen the degrot show signIPD; and 3)om that of
uality reduc
nteresting ahe one hand
per can be uerformance
ner, Lean Co
AIA), the In
07-A [email protected]
McLaughlin Hakeley.edu
www.l
ject Derd2
prove projecproject perf
between IPD
LP, includinnd Project pd non-IPD p
ree of implenificantly d IPD projecothers but
ce the credi
and importad and proje
used by inde. onstruction
ntegrated Pr
lin Hall, [email protected]
du, and Adjunall, Univ. of Ca
eanconstructio
elivery
ct performaformance? 3
D, LP, and p
ng IPD projeperformancprojects. ementation different cts do not shtheir
ibility of
ant relationect perform
ustry pract
, survey.
roject Deliv
of Californiau nct Associate alifornia, Berk
njournal.org
y
ance? 3) Do
project
ects, to e (cost
of LP
how
ships ance on
itioners
very
,
keley, CA
SimilarlOwners AssociatIPD as aparties based oAccordiconstrurework,organiza(Thomse
The construprocess
Many reLiu et aAlarconAgain, tflow to value (B
Whiin orderreducinorganizadesired whethethose q
Hypis a welof Variathree diprojectsquestiomultivadifferen
3 Labou
C
Lean Co http://
Multi partyCollaboratLiability wJointly de
ly, the Natio Associationted Generaa project de selected byon team perng to CMAAction indus, frequent dational, comen et al., 2
Last Plannction projees:
Master schPhase scheLook aheaWeekly Pla
esearchers hal (2008) sug et al (1997the LP has b minimize wBallard, 200
le IPD has tr to improveg plan variaation integr outcomes,r the impleuestions, w
othesis testll establisheance) to invifferent cons. More simnnaire, senriate t-test
nces among
ur Productivity
Cho & Ballar
onstruction Jocreativecommons
y contract;tive decisio
waivers amoveloped an
onal Associn of Americ
al Contractoelivery systey qualificatrformance wA, the purpotry such as disputes, exmmercial, a
2009)
er (LP) is a ects to impr
hedule; edule;
ad Plan; andan (Hamzeh
have provedggesting a r7) showing dbeen createwaste in tim00)
tried to intee project oability for trated by us such as co
ementation we did some
ting regardied methodovestigate if ntract type
milar to our t it to 7600
t, chi squareg three proj
y = 0.530 + 1.0
rd: Last Plan
urnal 2011 s.org/licenses/by-
on making aong key pard validated
ation of Staca, Associators in Ameriem using a ion based pwithout GMose of IPD is low rates oxcessive cosand operati
productionrove plannin
d h, 2009).
d reducing regression ldifference ed to maximme/money i
egrate projoutcomes, Lthe same pusing contracst/time red of LP can ae hypothesis
ing project ology. For e there is sigs, selected research de0 projects, ae test, ANOect deliver
095*Weekly P
nner and Inte
-nc-nd/3.0/ p
and control;ticipants; a
d project go
ate Facility tion of Highica, and Am multi partyprocuremen
MP, and opes to solve cof productivst, and exceional proble
n planning ang and prod
plan variabline3 betwein producti
mize reliabin project p
ect participLP has enforurpose. Ourctual alignmduction. If iachieve thos testing in
performanexample, Chgnificant dif from a govesign, Sanvand got 378
OVA, and rery systems.
lan Percent C
egrated Pro
page 68
; and oals (Cohen
Associationher Educatiomerican Insty contract tnt, managedn book acco
currently acvity, high raessive duraems in curre
and control duction perf
bility helps ien plan relivity beforeility of the processes an
pants’ rolesrced systemr question isment, such ait is not enose outcome this resear
ce based ohoi (2008) ufference in vernment daido et al (18 responsesgression to
Completion
ject Deliver
www.l
et al., 201
n (NASFA), on Facilitiestitute of Arcthat has mod/shared riounting (NAcknowledgeates of inef
ation--all caent project
system impformance.
increase priability and and after iwork/matend to maxim
s and relatiomatic produs if having as IPD, is enough, our nes. To find orch.
n a large nuused one wa schedule patabase of 998) made
s on which t identify pe
ry
eanconstructio
0).
Constructios Officers, chitects de
ore than twosk, compen
ASFA et al 2d problemsfficiency anaused by t delivery sy
plemented It has four
roductivity, d productiviimplementi
erial/informmize custom
ons contracction contrproject nough to mext concernout the ans
umber of pray ANOVA (Aperformancemore than a survey they did erformance
njournal.org
on
fined o
nsation 2010). s in the nd
ystems
on main
such as ty, and ing LP.
mation mer
ctually rol
aximize n is swers to
rojects Analysis e among 1,700
Resea
ReseaOur reseimplemIntegratimplemgeneral
If the ficomparredirecthypotheThe sec
A
This papthird hy
ReseaThe firsof varia(2001).
C
Lean Co http://
arch De
arch Hypoearch assumentation. Bted Projectentation an hypothesis
If a propro
rst hypothering IPD projted to a quaesis was supcond hypoth
If a proje
And the thir
If a proje
per is devotypothesis te
arch Measst thing thatables. We co
Cho & Ballar
onstruction Jocreativecommons
esign
othesis mption: proBased on thit Delivery (Ind IPD projes testing. Th
oject impleoject perfo
esis had notjects with oalitative expported, mahesis is:
ect adopts Idiff
rd hypothes
ect adopts Idiff
ted to the iesting.
surement we have tonceptualiz
rd: Last Plan
urnal 2011 s.org/licenses/by-
oject perforis assumptioPD) projectects’ perforhus, our fir
ements Lastormance be
t been suppothers in texploration saking it me
ntegrated Pferent from
sis is:
IPD, its degferent from
interpretati
nt to do after zed our vari
nner and Inte
-nc-nd/3.0/ p
rmance varion, we diagts to determrmance. Thst research
t Planner (Ltter than th
ported, it werms of LP aseeking whaaningful to
Project Delm those of o
gree of implm those of o
ion of the r
forming hyiables as sh
egrated Pro
page 69
ies with Lasgnosed the mine the cohis assumpti hypothesis
LP) more, ithose emplo
would be meand our reseat caused LP test the se
ivery (IPD),ther projec
lementationther projec
results from
potheses is hown in Figu
ject Deliver
www.l
st Planner ( degree of Lorrelation bion must bes is:
t achieves boying LP les
eaningless tearch woulP to fail. Hoecond and t
, its performcts.
n of Last Plcts.
m the first, t
to specify ure 1, follow
ry
eanconstructio
(LP) LP implemebetween LP e supported
better ss.
to go furthed have beeowever, thethird hypoth
mance is
lanner is
the second
the measurwing Adcoc
njournal.org
ented in
d by
er n e first hesis.
and the
rement ck et al
Figure
We struparts.
The Last Plabased o
C
Lean Co http://
e 1: Concep
uctured the
independeanner (LP). on the follow
1 Pullinnext w(Tomm
2 Looka(prereinappequipin firstasks (Balla
3 Learnanalyzreocc
Cho & Ballar
onstruction Jocreativecommons
ptualizatio
variables i
ent variable To measurewing five el
ng productiworkers (immelein, 199
ahead procequisite wo
propriatenespment, inadst run study are not eligard, 2000)
ning from byzed to rooturrence (Ba
rd: Last Plan
urnal 2011 s.org/licenses/by-
on and mea
n the hypot
e of the firste this abstrlements:
ion: each wmmediate cu98)
cess: each fork, contracss, insuffic
dequate dury, etc) befogible for in
breakdownst causes andallard, 2000
nner and Inte
-nc-nd/3.0/ p
surement:
theses so th
t hypothesiract concep
worker invesustomers) b
front line sctual approient resourration, fundore executionclusion on
s: failures d actions ar0)
egrated Pro
page 70
Levels and
hey could b
s is the degt, we deve
stigates thebefore exec
upervisor rovals, sequerce as well ading probleon of its ta daily or we
to completre taken to
ject Deliver
www.l
d task (Adc
e measured
gree of imploped indic
e readinesscution of ta
removes conential as labour &
em, problemasks. Constreekly work p
te planned t prevent
ry
eanconstructio
ock et al.,
d in the foll
lementatioators to be
s of the asks
nstraints
& ms found rained plans
tasks are
njournal.org
2001)
lowing
n of scored
Th
# Que1 Wha
schewor
2 To wwas plancons
3 To wshouimm
4 Did saidwee100 plan
5 Howweeprev
So far, whypotheperformfinal apapprovegetting
Thvariablethe indesecond the lasthypothewhetheadopts measureto score
C
Lean Co http://
4 Phasecollabhando
5 Districloserwho a
he indicato
estions at percentageduling the pk?
what extent w ready to ben? Bear in mistraints are rwhat extentuld be done
mediate cus
the project d you were goekly work pla tasks on wenned (no parw often wereekly work plavent reoccur
we have spesis. Next, wmance. We dpproved buded schedule good data
he dependee of the firsependent v hypothesis t concept thesis; i.e., tor a project IPD, as the e the extene the degre
Cho & Ballar
onstruction Jocreativecommons
e schedulinboration of off is produ
ibuted conr to executare to do th
rs in the bo
Table 2:
ge of specialtproject phas
was the prine performed ind that worremoved. t was the pe in responsstomer, such
measure theoing to do?’ an tasks comekly work pl
rtial credit), e reasons for an) analyzedrrence?
ecified the we addressdecided to dget) + durae) as a meason other pe
ent variablest hypothesariable of t is the samehat we defio what exte adopts IPD type of thint of impleme of adopti
rd: Last Plan
urnal 2011 s.org/licenses/by-
ng: every ha all relevan
uced (Ballar
trol: Work ion, and plahe work. (Ba
ox above ar
Survey que
ty contractoe(s) in which
nciple followcould be plak is ready to
principle folse to a requh as the nex
e extent to w(The measur
mpleted as pllans and 70 w the percent not complet to root caus
measurem the depend use the sumation reducsure of projerformance
e of the secis. The depthe first hype as the indne is the inent a projecD. We decide
s variable bmentation. ng IPD stru
nner and Inte
-nc-nd/3.0/ p
andoff in a nt specialistrd et al., 20
is planned anning is doallard et al
e transform
estions mea
rs participath they were
ed that onlyaced on a weo be perform
llowed thatuest from axt trade?
which you ‘dre is the percanned. If thewere completage would bting plannedses and actio
ent of the ident variabm of cost retion ratio (ject perform
e dimension
ond hypothpendent varpothesis. Andependent vndependentct adopts Ined to take because we In additionctures if we
egrated Pro
page 71
phase shouts in the ph003)
in greater one collabol., 2003)
med into sur
asuring Las
ted in to do their
y work that eekly work ed when all
work n
id what you centage of ere were eted as be 70%) d tasks (on on taken to
independenble of the saeduction rat%) (actual dmance becas.
hesis is the riable of thend the indevariable of variable of
ntegrated Pthe binary
e could not n, it would e had used
ject Deliver
www.l
uld be definhase before
detail as yooratively by
rvey questio
st Planner Answer tyPercentagNone: 1/625-50%:3/75-100%:5
FrequencyNever: 1/SometimeAnd AlwayFrequencyNever: 1/SometimeAnd Alway
Yes/No ⇒Yes: 1; an
FrequencyNever: 1/SometimeAnd Alway
nt variable ame hypothtio (%) (actduration reause of the
same as thee third hypoependent va the third hf the secon
Project Delivvariable, wget enough be difficult continuous
ry
eanconstructio
ned by e the
ou get y those
ons:
ype & Scoringe ⇒ 6; 0-25%:2/6/6; 50-75%:45/6; and All:y ⇒
/5; Rarely: 2es: 3/5; Ofteys: 1 y ⇒
/5; Rarely: 2es: 3/5; Ofteys: 1
⇒ nd No: 1/6
y ⇒ /5; Rarely: 2es: 3/5; Ofteys: 1
in the first hesis, projeual cost unlative to fin low probab
e dependenothesis is saariable of thypothesis. Td and the tvery (IPD),
whether a prh IPD project for respons variables.
njournal.org
g Rule
6; 4/6; 1
/5; en: 4/5;
/5; en: 4/5;
/5; en: 4/5;
ect der nal bility of
nt ame as he Thus, third or roject cts to ndents
HypotThe hypindepenquantitain Tablea real nhypothetwo varof the sproject a T-testdifferenand the
SamplIn commrandomcontrol implemadvantaparticipLaboratrandomPurposivused effthe winsuch as
Resu
RegreThere isproject have suin Table
Figuline. Evapproprwhose i‘cube o
The cost red
4 http:/5 http:/
C
Lean Co http://
thesis tespothesis tesndent variabatively conte 2 is the tonumber. Theesis is a ratiriables is apsecond and adopts IPDt, to determnt influencee degree of
ling Stratmon sense, ized sampli so that we entation in
age of e-mapants in wortory5. The sly from anyve samplingfectively inner in the p one describ
lts
ession mos a significa performanccessfully se 3 in the A
ure 2 is a grven though wriate in desndependenf X’ or X3
result is enduction and
//finance.dir.//p2sl.berkele
Cho & Ballar
onstruction Jocreativecommons
sting mesting was peble (degreetinuous ordotal degree e dependenio variable
ppropriate f the third h
D’, for whichmine whethe on depend implement
tegy the most aing. Howev need the v their proje
ail lists in rerkshops sucame applieywhere in tg is widely n identifyingpast, or it ibing gangst
odel fromant correlatce—the sumsupported t
Appendix.
raphical repwe used a scribing beh
nt variables
ncouraging.d time reduc
.groups.yahooey.edu/
rd: Last Plan
urnal 2011 s.org/licenses/by-
thodologerformed die of Last Pladinal variab of Last Plant variable representefor testing typothesis ish regressioner the catedent variabtation of La
ppropriate ver, Last Plavery specificects. Thus, elevant groch as those es to selectihe world, vused in studg communits used in se
ter’s lives (B
m testingtion betweem of cost anthe first hyp
presentationstraight linehaviour of v are ‘square
. The regresction) = 0.7
o.com/group/i
nner and Inte
-nc-nd/3.0/ p
gy ifferently aanner implele because
anner imple(cost reduc
ed as a real the hypothes a binary cn analysis isegorization les: projectst Planner i
form of samanner (LP) ic responde we decidedups such assponsored bion of IPD pvery few, if dying unusuties across telecting keyBernard, 20
g the firsten the implnd schedulepothesis. Th
n including e, the scattariables. The of indepe
ssion mode7371101×X2
iglc/message/
egrated Pro
page 72
according toementation the sum ofementation ction + time number. Thesis. Howevcategorical s not appro(IPD or othet performanin the third
mpling to suis a very spents who cand to use a ps general IGby the Projprojects. If any, IPD pual critical the United y informant000)
t hypothementatione reduction his is repres
scatter ploter plot seehus, we trie
endent varia
l with ‘squa2-3.89088’ w
/677
ject Deliver
www.l
o the type o) of the firs
f scores of t of a projece reduction)hus, regresver, the indvariable, ‘w
opriate. In terwise) hasnce in the s
d hypothesis
upport a hyecific tool fn determinepurposive saGLC group inect Productwe were torojects woucases. For e States thatts for ethno
esis n of Last Pla percentagesented as a
otting and aems to showed several lable (X) in F
are of X’ orwith its P<|
ry
eanconstructio
of variable.st hypothesthe five quect, represen) of the samsion betwee
dependent vwhether or this case, ws a significasecond hypos.
ypothesis is for producte the degreampling takn Yahoo4, otion Systemo select prould be incluexample, itt have votedographic stu
anner (LP) aes. That me
a regression
a linear regrw a curve is linear regreFigure 2’ or
r X2 is ‘Y(Sum|t(2.98)| is
njournal.org
The sis is a estions nted as me en the variable not a
we used antly othesis,
tion ee of LP king r
m ojects uded. t can be d for udies
and eans we model
ression more essions, r X2 and
m of 0.005,
which is(P<|t|) Furtherof t is 0does no
The finaproportrepresecurve’, reductio
C
Lean Co http://
s less than 0 means the
rmore, the r0.004, whichot show mor
Figu
al regressiotionate to thented as blu which showon) = 0.148
-20
-10
010
2030
40
.5
Sum
Cho & Ballar
onstruction Jocreativecommons
0.009, the re is greateregression mh is less thare significan
ure 2: Regre
on line with he degree o
ue diamond ws that Proj4254 ×(Imp
1 1.5
m of cost reduct
rd: Last Plan
urnal 2011 s.org/licenses/by-
P<|t(2.71)er significanmodel with an 0.005 in nce than X3
ession of L
X cubed, sof Last Plan type plots ject Performplementatio
5 2Implem
tion and schedu
nner and Inte
-nc-nd/3.0/ p
| in Table 3nce in the c X3 is ‘Y = 0 the regress3.
Last Planne
aying that tnner’s imple in Figure 3mance (sumon of Last P
2.5mentation of Las
ule reduction(Y)
egrated Pro
page 73
3 with merecoefficient o0.1484254×
sion model
r on Projec
the projectementation. We decide
m of cost rePlanner)3-1.
3 3.5st Planner (X)
ject Deliver
www.l
e X. The lesof the regre
×X3-1.61730 with X2. Bu
ct Performa
t performann cubed, is ved to call it
eduction and.617307.
4
Y=4.141356*
ry
eanconstructio
ss p value ofession line.7’ with its
ut, X to the
ance
nce is visually t ‘Cho-Ballad schedule
4.5 5
*X-9.003641
njournal.org
f t p value fourth
ard
SummThe foll
6 For de
-2
-1
1
2
3
4
5Su
m o
f Cost
red
uct
ion a
nd S
ched
ule
red
uct
ion: Y (
%),
"more
red
uct
ion o
f co
st a
nd s
ched
ule
" Bott
om
to T
op
C
Lean Co http://
Figure 3: C
mary of Hlowing box
If a pro
=> Stron(sum
If a proje
If a proje
=> Faimple
etail of hypot
20
10
0
10
20
30
40
50
0
X: S
Cho & Ballar
onstruction Jocreativecommons
Cho-Ballard
Hypothesi summarize
oject implemperforma
ngly supporm of cost red
(Imple
ect adopts Idiff
=> F
ect adopts I
ailed to be smented LP
hesis testing,
1
Score of Last
rd: Last Plan
urnal 2011 s.org/licenses/by-
d Curve b/w
is testings the result
Hy
ments Last ance better
rted by the duction andementation
Hy
ntegrated Pferent from
Failed to b
Hy
IPD, its degfrom thos
supported. to a certai
signific
please see Ap
2
Planner Impl
Cho-
nner and Inte
-nc-nd/3.0/ p
w (Last Plan
g6 ts of hypoth
ypothesis
Planner (Lr than those
regression d schedule of Last Pla
ypothesis 2
Project Delm those of o
e supported
ypothesis 3
gree of implse of other
However, Iin degree evcant statist
ppendix
3
lementation
Ballard C
egrated Pro
page 74
nner)3 and
hesis testing
1
P) more, ite employing
model: Proreduction)
anner)3-1.61
2
ivery (IPD),ther projec
d definitive
3
lementatio projects.
IPD projectven thoughically.
4
(LPI), "more
Curve
Y=0.14
ject Deliver
www.l
Project pe
g so far.
t achieves pg LP less
oject Perfor = 0.14842517307
, its performcts.
ely
n of LP is d
ts in our samh the level i
5
regorous LPI
484254*X^3-1.6
ry
eanconstructio
erformance
project
rmance 54 ×
mance is
different
mple is not
6
I" L->R
617307
njournal.org
e
ConcWe founLast PlaPerform
Thisparticipbetter pcriticiziorganizahow the
Appe
DetailTable 3 projects2’ in TadurationsignificaTable 3model)
Sourc
Mode
Residu
Tota
Y in Figu
X in Figu
Consta
DetailThe secperformthe twoperformshould btest, navarianceprobabialternat
C
Lean Co http://
lusion nd in this reanner. Howemance, and
s research dpants, and Lperformancng forms ofational intee work is ac
endix
l of the f is the resus. Simply, w
able 3. This n reductionance of this). Usually, is significan
ce
el 543
ual 3467al 4010
ure2 Coe
ure2 4.1
ant -9.0
l of the scond hypothmance is difo groups (IPmance becaube performamed as “sde of IPD andility, exprestive hypoth
Cho & Ballar
onstruction Jocreativecommons
esearch thaever, we di Integrated
does not preLP, reducingce. Indeed, f IPD that regration, whctually done
first hypoult producedwe need to is the grad
n’ and X is ‘s coefficienif P>|t| is lnt. In our c
Table 3: ReSS D
.294059
7.23372
0.52778
fficient S
141356 1
003641 5
second hyhesis is < If fferent fromD and Non use generaled under th
dtest” in STd that of Nossed as p (F
hesis, locate
rd: Last Plan
urnal 2011 s.org/licenses/by-
at project pid not find a Project De
event us frog project va this is the cely only on hile neglecte. Future re
othesis ted by STATA see the ‘co
dient of the the degree
nt is determless than 0.ase, the re
esult of ReDF M
1
47
48
Std Errors
1.526046
5.279548
ypothesia project a
m those of oIPD) have sl T test is phe unequal TATA v.10. Ton IPD, whiF<f), p (|F|ed right abo
nner and Inte
-nc-nd/3.0/ p
performancea strong rel
elivery (IPD)
om believinariability, aclaim put fo alignment ting the leaesearch is n
esting v.10, a staoefficient’, regression
e of implemmined by P >.05, we cangression mo
egression foMS
543.294059
73.7709302
83.552662
T
2.71
-1.77
is testingadopts Integother projecignificantly
performed b variance coThe f valueich is expre>|f|), and ove the pro
egrated Pro
page 75
e improves lationship a).
ng that if IPare combineorward by L of commeran ‘operatinneeded to v
atistics pack written on line. Y is ‘sentation of
> |t|, 0.009n say this coodel is Y=4.
or the first N
P
Adj
Root
P>|t|
0.009
0.095
g grated Projects>. Beforey different vbased on eqondition. Ta stands for
essed as ‘Rap (F>f) in T
obability, is
ject Deliver
www.l
with the imamong Last
D, aligning ed, any proLean Constrrcial interesng system’,validate this
kage, using n the right ssum of costf Last Plann9 (red-undeoefficient (t.141356×X-
hypothesisNumber of o
F(1, 47)
Probability >
R-squared
justed R-squ
Mean Squar
95% c
1.071
-19.62
ect Deliverye T test, wevariance in qual variancable 4 is th the ratio batio’ in TablTable 4, is l chosen. In
ry
eanconstructio
mplementat Planner, Pr
goals of oject can acruction adhsts and which adds claim.
data from side of ‘Y int reduction ner’. The erlined numthe regress-9.003641
s object = 49
= 7.36
> F = 0.0093
= 0.1355
uared = 0.117
re Error = 8.
confidence In
347 7.
2472 1.
y (IPD), its e needed to project ce. If not, Te variance
between thele 4. When ess than 0.0 this test, t
njournal.org
tion of roject
chieve herents,
resses
the 49 n Figure and
ber in ion
71
859
nterval
211366
.61744
o see if
T test ratio e the 05, the
the
target ahypotheanother
Grou
Non
IPD
Com
RatiIPD)Null
AlteProb
Table 5 ‘the remwhich is(|T|>|tright abDifferensecond is 0.782second
TaGroNon
IPD
Com
Diffe
Diffe(IPDNull
AlteDiffeProb
As we mthat of bigger t
DetailThe thirPlannersignificaimplemalternat
C
Lean Co http://
alternative esis is 0.084r T test with
Table 4: Vup Ob
IPD 40
9
mbined 49
o = standard)/standard dl hypothesis:
ernative Hypobability: p (F
is the resumainder of s expressedt|), and p (Tbove the pronce!= 0, a d hypothesis.28, much bi hypothesis
able 5: T-teup O
n IPD 40 9
mbined 49erence
erence = MeD) l hypothesis:
ernative Hypoerence < 0 bability: p (T
mentioned, equal variathan 0.05.
l of the trd hypothesr (LP) is diffant differenentation bytive hypoth
Cho & Ballar
onstruction Jocreativecommons
hypothesis 43, which ish unequal v
Variance Rabs. M
5
4
4
d deviation (eviation (IPD Ratio =1
othesis (Ha):F<f)=0.9578
ult of T-test the Performd as ‘DiffereT>t) in Tabobability, isdifferent ex. p (|T| >|tgger than 0
est with eqbs. M0 5
4
9 4
.
an (Non IPD)
Difference
othesis (Ha):
T<t)=0.6086
we did anoance. Unequ
third hypsis is <If a pferent fromnce betweey showing thhesis (standa
rd: Last Plan
urnal 2011 s.org/licenses/by-
is Ha: ratios bigger thavariance for
atio Test onMean
5.105027
4.160776
4.931593
Non D)
: ratio <1
t with equamance meaence’ in Table 5, is lesss chosen. Inxpression but|) right be0.05 so that
qual variancMean 5.105027
4.160776
4.931593
6551067
)-Mean
= 0
:
other T – tesual T test sa
pothesis tproject ado
m those of oen the variahe probabilard deviatio
nner and Inte
-nc-nd/3.0/ p
o!=1. The pan but near r more assu
n performaStd. Err.
1.55351
1.822258
1.305816
Ha: ratio !=2*p (F>f) =
l variance on of Non IPble 5. Whens than 0.05,n our case, ut one havielow the altt we cannot
ce on perfoStd. Err. 1.55351
1.822258
1.305816
3.15756
Ha: Differep (|T|>|t|)
st with uneays the pro
testing pts IPD, its ther projec
ance of the lity, used inons of the t
egrated Pro
page 76
robability r to 0.05 so urance
nce betweStd. Dev.
9.825258
5.466773
9.140715
=1 0.0843
of STATA v.D after subn a probabi, the altern the target ng the sameternative hyt choose the
ormance beStd. Dev. 9.825258
5.466773
9.140715
nce !=0 ) = 0.7828
equal varianbability p (
degree of icts> The var two groupsn determinitwo groups
ject Deliver
www.l
right under that we cam
en IPD and [95% Con
1.962757
-.041357
2.306073
f = 3.230degrees (=40-1),
Ha: RatiP (F > f)
10. The ‘t’ btracted by lity, expresative hypot alternativee meaning ypothesis, He alternativ
etween IPD [95% Co
1.962757
-.041355
2.306073
-5.90614
t = 0.277degrees
Ha: DiffeP (T > t)
nce, whose |T|>|t|) is
implementariance test s (IPD and Ning whether are differe
ry
eanconstructio
the altername to decid
d otherwisenf. Interval]
7 8.2472
77 8.3629
3 7.5571
02 of freedom 8 (=9-1)
o > 1 = 0.0422
value stan the Mean ossed as p (Tthesis, locae hypothesisas that of oHa: Differenve hypothes
D and Non Ionf. Interval7 8.2472
57 8.3629
3 7.5571
44 7.7946
73 of freedom
erence > 0 ) = 0.3914
result is sims 0.6972, m
ation of Las said there Non IPD) in r to choose
ent), is 0.19
njournal.org
ative de to do
e
297
909
13
= 39
ds for of IPD’, T<t), p ated s is Ha: our nce!=0, sis, our
PD ]
297
909
13
646
= 47
milar to uch
st is no LP the
948,
bigger t
TabSimilar (T<t), plocatedmeans othat of clear fothat theat the bimplemLast Pla
TaGrouNon
IPD
Com
Diffe
DiffeNull
AlterProb
ReferAdcock,
anAlarcon
prArbulu,
inPr
Ballard,Ci20
Ballard,ofCh
Ballard,Ballard,
ToCo
Ballard,11
Ballard,Pa
BernardCA
C
Lean Co http://
than 0.05.
le 6 shows to Table 5,p (|T|>|t|) right aboveof IPD and N our third hyor us whethee probabilitborder. In sentation of
anner to a c
able 6: T-teup IPD
mbined
erence
erence = Mea hypothesis:
rnative Hypobability: p (T
rences , R., and Cond quantita, L. F., androject outco R., Tomme
n Constructiroceedings , G. (2000).ivil Eng., Un010) , G. (2006).f the 16th anhile. , G. (2008)., G., Iris Toools and Teonstruction, G., and Ho1th , BlacKsb, G and Grearadigms”, d, R. (2000)A: Sage pub
Cho & Ballar
onstruction Jocreativecommons
the result o if a probab, and p (T>e the probaNon IPD)!=0ypothesis. Eer to discarty, P (T<t), hort, even f Last Planncertain degr
est with eqObs. 40
9
49
an (Non IPD) Difference =
othesis (Ha): <t)=0.0370
ollier, D. (2ative researd Cruz, J. Comes”, Tenelein, I., Waion Supply C of the 2002. “The last niv. of Birm
. “Rethinkinnnual confe
. “LPDS Updommelein, Lechniques”,n building inowell, G. (2burg, VA, Uegory A. How Lean Const. The socia
blication.
rd: Last Plan
urnal 2011 s.org/licenses/by-
of t test witbility right t), is less tability. Our 0’, a differeEven thougrd our third is 0.7828, mthough Inte
ner significaree
qual variancMean
3.266667
3.801471
3.364896
-.5348048
-Mean (IPD) = 0
Difference <
2001). Measch. Americ. (1997). “T
nth World Palsh, K., anChains: Cas2 Winter Simplanner sys
mingham, U
ng project derence, Inte
date”. LeanLauri Koske In Rick Besn Value. Wo2003). “An
USA, July 20well (2004)truction Joual research m
nner and Inte
-nc-nd/3.0/ p
th equal vaunder the ahan 0.05, w alternativeent expressih P (|T| >|
d hypothesismuch biggeegrated Proantly differe
ce on perfoStd. Err.1321312
.1802989
.116053
.2926632
<0 Ha: Diffep (|T|>|
urement vaan PoliticalThe impactProductivitynd Hershauese of Pipe Sumulation Costem of proK. Available
definition inernational G
n constructila, and Grest and Geraoburn, MA: update on
003, pp. 329. “Competiurnal, 1(1)3methods-qu
egrated Pro
page 77
riance in tealternative we can chooe hypothesiion but onet|), 0.074 is. As for secer than 0.05oject Deliveent from ot
ormance be. Std. De2 .835671
9 .540896
.812370
2
erence !=0 |t|) = 0.0740
alidity: A shl Science Re of project
y Congress. er, J. (2002upports useonference. duction cone at www.le
n terms of tGroup for L
ion journal,eg Howel (2ard De Valen Elsevier Sclast planne9-341. ing Construc38-45. ualitative a
ject Deliver
www.l
esting our t hypothesisose the altes is ‘Differe
e having theis bigger thcond hypot5. But, the ery projectstherwise, it
etween IPDev. [95% Co14 2.999406
68 3.385701
07 3.131556
-1.12356
t = -1.82degrees
0 Ha: DiffeP (T > t)
hared standeview, 95(3 planning a Santiago, C
2). “Contribed in power Sandiego, ntrol” PhD. eanconstru
target costiLean Constru
, 2008, pp. 002). “Leannche (Eds),ience. Pp. r.” Proceed
ction Manag
nd quantita
ry
eanconstructio
hird hypoth, representernative hyence (betwee same meahan 0.05, it hesis, it is cthird hypots do not sho seems to e
D and Non Inf. Interval]6 3.53927
1 4.21724
6 3.59823
67 .053957
274 of freedom
erence > 0 = 0.9630
ard for qua3). nd PerformChile, Octo
butors to Ler Plants.” CA, USA, De Diss. Schooction.org (2
ing”. Proceuction, San
1-19. n Construct Design and227-255 dings of the
gement
ative appro
njournal.org
hesis. ted as p pothesis, een
aning as is not clear in thesis is ow employ
PD ] 7
42
36
76
= 47
alitative
mance on ber ad time
ec 8-11 ol of 26 Mar
eedings ntiago,
tion d
e IGLC
aches.
Cho, Sepr
Choi, Kuthan
Chuck TInAv19
Cohen, In
HamzehPlBe
Liebermre
Liu, M.,CoMa
NationaofCoPrhtne
SanvidoBuTe
C
Lean Co http://
eongkyun anroceedings unhee (200hrough a Qund EnvironmThomson, Jontegrated Prvailable at 9-09__2_.pd Jonathan anstitute of Ah, Farook, Rlanning anderkeley, CA
man.E. (200esearch. Am, and Ballarontrol.” Proanchester,
al Associatiof America, Aontractors iroject Delivttp://wwwers.pdf (10 o, Victor, anuild, and Deeam 133, A
Cho & Ballar
onstruction Jocreativecommons
nd Ballard, IGLC-18, Ju8). “A new
uantitative Amental Eng.oel Darringtroject Delivhttp://cmadf (9 Jan 20and FAIA (20Architects CR. (2009). “d Control” PA, USA. 5). Nested
merican Polird, G. (2008oc 16th Ann. UK. on of State Associationin America,very for Pub.agc.org/ga Jan 2010) nd Konchar,esign-Bid-Bustin, Texa
rd: Last Plan
urnal 2011 s.org/licenses/by-
G. (2010), buly 2010, Te Decision SuAnalysis on., Univ. of Cton, Dennisvery. Constaanet.org/f011) 010). IntegrCalifornia C“Improving CPhD. Diss. C
analysis as itical Scien8). “Improvi. Conf. Int’
Facility Ass of Higher E, and Ameriblic and Privalleries/pro
, Mark (199Build. Constas, USA.
nner and Inte
-nc-nd/3.0/ p
building anechnion, Haupport Mod Aspects ofCalifornia, Bs Dunne, antruction Manfiles/shared
rated Projeouncil, SacConstructio
Civil and Env
a mixed mce review, ing Labor Pl Group of
sociation (NEducation Fican Instituvate Owner
ojectd/IPD%
8). Project truction Ind
egrated Pro
page 78
nd argumenaifa, Israel el for Innov
f Project PeBerkeley. C
nd Will Lichtnagement Ad/ng_Integr
ect Deliveryramento, C
on Workflowvironmenta
ethod strat 99(3) Productivity Lean Const
NASFA), ConFacilities Oftes of Archirs” Availabl%20for%20Pu
Delivery Syustry Instit
ject Deliver
www.l
t with hypo vative Conterformance”CA, USA. tig (2009). Association rated_Proje
y: Case studCA, USA. w-the role oal Eng. Univ
tegy for com
y Through Ptr., IGLC 16
nstruction Officers, Assoitects (2010le at ublic%20and
ystems: CMute Design
ry
eanconstructio
othesis test
racting Stra” PhD. Diss
Managing of Americaect_Delivery
dies. Americ
of Productiov. of Califor
mparative
roduction 6, 16-18 July
Owners Assoociated Gen0). “Integra
d%20Private
M at Risk, De Build Resea
njournal.org
ting,
ategies . Civil
a. y__11-
can
on rnia,
y,
ociation neral ated
e%20Ow
esign-arch