anova dasar
DESCRIPTION
SatistikTRANSCRIPT
-
7/21/2019 ANOVA Dasar
1/37
1
Analysis of VarianceAnalysis of Variance
Introduction to Analysis of VarianceIntroduction to Analysis of Variance
Analysis of Variance: Testing for the Equality ofAnalysis of Variance: Testing for the Equality of
kk Population MeansPopulation Means
Multiple Comparison ProceduresMultiple Comparison Procedures
-
7/21/2019 ANOVA Dasar
2/37
A!A"ISIS VA#IA!S $A!%VA&A!A"ISIS VA#IA!S $A!%VA&
ISI:ISI:
' Men(elas)an )onsep dasar A!%VAMen(elas)an )onsep dasar A!%VA
$*(i +&$*(i +&
' #umus dan cara penghitungan ,er,agai#umus dan cara penghitungan ,er,agai
item dalam A!%VAitem dalam A!%VA' Contoh penggunaan A!%VAContoh penggunaan A!%VA
' +isher test untu) u(i per,edaan diantara+isher test untu) u(i per,edaan diantara
rata-rata sampel yang adarata-rata sampel yang ada
-
7/21/2019 ANOVA Dasar
3/37
.
Introduction to Analysis of VarianceIntroduction to Analysis of Variance
Analysis of VarianceAnal
ysis of Variance$A!%VA& can ,e used to test$A!%VA& can ,e used to testfor the equality of three or more population means/for the equality of three or more population means/
Analysis of VarianceAnal
ysis of Variance$A!%VA& can ,e used to test$A!%VA& can ,e used to testfor the equality of three or more population means/for the equality of three or more population means/
0ata o,tained from o,serational or e2perimental0ata o,tained from o,serational or e2perimental
studies can ,e used for the analysis/studies can ,e used for the analysis/
0ata o,tained from o,serational or e2perimental0ata o,tained from o,serational or e2perimentalstudies can ,e used for the analysis/studies can ,e used for the analysis/
3e 4ant to use the sample results to test the3e 4ant to use the sample results to test thefollo4ing hypotheses:follo4ing hypotheses:
3e 4ant to use the sample results to test the3e 4ant to use the sample results to test thefollo4ing hypotheses:follo4ing hypotheses:
HH55:: 116666..66/ / // / / 66 ))
HHaa: !ot all population means are equal: !ot all population means are equal
-
7/21/2019 ANOVA Dasar
4/37
7
Introduction to Analysis of VarianceIntroduction to Analysis of Variance
HH55:: 116666..66/ / // / / 66 ))
HHaa: !ot all population means are equal: !ot all population means are equal
IfIf HH55is re(ected8 4e cannot conclude thatis re(ected8 4e cannot conclude that allallpopulation means are di9erent/population means are di9erent/
IfIf HH55is re(ected8 4e cannot conclude thatis re(ected8 4e cannot conclude that allallpopulation means are di9erent/population means are di9erent/
#e(ecting#e(ecting HH55means that at least t4o populationmeans that at least t4o population
means hae di9erent alues/means hae di9erent alues/
#e(ecting#e(ecting HH55means that at least t4o populationmeans that at least t4o population
means hae di9erent alues/means hae di9erent alues/
-
7/21/2019 ANOVA Dasar
5/37
Sampling 0istri,ution of ;ienSampling 0istri,ution of ;ien HH55is Trueis Truexx
Introduction to Analysis of VarianceIntroduction to Analysis of Variance
1x1x .x.xxx
Sample means are close togetherSample means are close together,ecause there is only,ecause there is only
one sampling distri,utionone sampling distri,ution4hen4hen HH55is true/is true/
xn
=
xn
=
-
7/21/2019 ANOVA Dasar
6/37
-
7/21/2019 ANOVA Dasar
7/37 =
+or each population8 the response aria,le is+or each population8 the response aria,le isnormally distri,uted/normally distri,uted/+or each population8 the response aria,le is+or each population8 the response aria,le isnormally distri,uted/normally distri,uted/
Assumptions for Analysis of VarianceAssumptions for Analysis of Variance
The ariance of the response aria,le8 denotedThe ariance of the response aria,le8 denoted 88
is the same for all of the populations/is the same for all of the populations/
The ariance of the response aria,le8 denotedThe ariance of the response aria,le8 denoted 88is the same for all of the populations/is the same for all of the populations/
The o,serations must ,e independent/The o,serations must ,e independent/The o,serations must ,e independent/The o,serations must ,e independent/
-
7/21/2019 ANOVA Dasar
8/37 >
:Testing for the Equality ofTesting for the Equality of kkPopulationPopulation
MeansMeans ?et4een-Treatments Estimate of Population Variance?et4een-Treatments Estimate of Population Variance
3ithin-Treatments Estimate of Population Variance3ithin-Treatments Estimate of Population Variance
Comparing the Variance Estimates: TheComparing the Variance Estimates: The FF TestTest
A!%VA Ta,leA!%VA Ta,le
-
7/21/2019 ANOVA Dasar
9/37 @
?et4een-Treatments Estimate?et4een-Treatments Estimateof Population Varianceof Population Variance
A ,et4een-treatment estimate ofA ,et4een-treatment estimate of is called theis called the
mean square treatmentmeansquare treatmentand is denoted MST#/and is denoted MST#/
1
$ &
MST# 1
k
j jj
n x x
k
=
=
1
$ &
MST# 1
k
j j
j
n x x
k
=
=
0enominator represents0enominator representsthethe degrees of freedomdegrees of freedomassociated 4ith SST#associated 4ith SST#
!umerator is the!umerator is the
sum of squaressum ofsquaresdue to treatmentsdue to treatmentsand is denoted SST#and is denoted SST#
-
7/21/2019 ANOVA Dasar
10/37 15
The estimate ofThe estimate of ,ased on the ariation of the,ased on the ariation of the
sample o,serations 4ithin each sample is calledsample o,serations 4ithin each sample is calledthethe mean square errormean square errorand is denoted ,y MSE/and is denoted ,y MSE/
3ithin-Samples Estimate3ithin-Samples Estimateof Population Varianceof Population Variance
kn
sn
T
k
j
jj
=
=1
2)1(
MSEkn
sn
T
k
j
jj
=
=1
2)1(
MSE
0enominator represents0enominator representsthethe degrees of freedomdegrees of freedom
associated 4ith SSEassociated 4ith SSE
!umerator is the!umerator is the
sum of squaressum of squaresdue to errordue to error
and is denoted SSEand is denoted SSE
-
7/21/2019 ANOVA Dasar
11/37 11
Comparing the Variance Estimates: TheComparing the Variance Estimates: The FFTestTest
If the null hypothesis is true and the A!%VAIf the null hypothesis is true and the A!%VA
assumptions are alid8 the sampling distri,ution oassumptions are alid8 the sampling distri,ution o MST#MSE is anMST#MSE is an FFdistri,ution 4ith MST# d/f/distri,ution 4ith MST# d/f/ equal toequal to kk- 1 and MSE d/f/ equal to- 1 and MSE d/f/ equal to nnTT-- kk//
If the means of theIf the means of the kkpopulations are not equal8 thpopulations are not equal8 th
alue of MST#MSE 4ill ,e inBated ,ecause MST#alue of MST#MSE 4ill ,e inBated ,ecause MST# oerestimatesoerestimates //
ence8 4e 4ill re(ectence8 4e 4ill re(ect HH55if the resulting alue ofif the resulting alue of
MST#MSE appears to ,e too large to hae ,eenMST#MSE appears to ,e too large to hae ,een
selected at random from the appropriateselected at random from the appropriate FF distri,ution/distri,ution/
-
7/21/2019 ANOVA Dasar
12/37 1
Test for the Equality ofTest for the Equality of kk PopulationPopulationMeansMeans
FF6 MST#MSE6 MST#MSE
HH55:: 116666..66/ / // / / 66 ))
HHaa: !ot all population means are equal: !ot all population means are equal
ypothesesypotheses
Test StatisticTest Statistic
-
7/21/2019 ANOVA Dasar
13/37 1.
Test for the Equality ofTest for the Equality of kk PopulationPopulationMeansMeans
#e(ection #ule#e(ection #ule
4here the alue of4here the alue of FF
is ,ased on anis ,ased on an
FFdistri,ution 4ithdistri,ution 4ith kk- 1 numerator d/f/- 1 numerator d/f/andand nnTT-- kkdenominator d/f/denominator d/f/
#e(ect#e(ect HH55ififpp-alue-alue DDpp-alue Approach:-alue Approach:
Critical Value Approach:Critical Value Approach: #e(ect#e(ect HH55ifif FFFF
-
7/21/2019 ANOVA Dasar
14/37 17
Sampling 0istri,ution of MST#MSESampling 0istri,ution of MST#MSE
#e(ection #egion#e(ection #egion
0o !ot #e(ect H5
#e(ect H5
MST#MSE
Critical ValueF
Sampling 0istri,utionSampling 0istri,utionof MST#MSEof MST#MSE
-
7/21/2019 ANOVA Dasar
15/37
1
A!%VA Ta,leA!%VA Ta,le
SST isSST is
partitionedpartitionedinto SST# andinto SST# and
SSE/SSE/
SSTFs degrees ofSSTFs degrees offreedomfreedom
$d/f/& are partitioned$d/f/& are partitionedintointo
SST#Fs d/f/ and SSEFsSST#Fs d/f/ and SSEFsd/f/d/f/
TreatmentTreatment
ErrorError
TotalTotal
SST#SST#
SSESSE
SSTSST
kkG 1G 1
nnTT GG kk
nnTT- 1- 1
MST#MST#
MSEMSE
Source ofSource ofVariationVariation Sum ofSum ofSquaresSquares 0egrees of0egrees of+reedom+reedom MeanMeanSquaresSquares
MST#MSEMST#MSE
FF
-
7/21/2019 ANOVA Dasar
16/37
1
-
7/21/2019 ANOVA Dasar
17/37
1=
A!%VA Ta,leA!%VA Ta,le
A!%VA can ,e ie4ed as the process of partitioningA!%VA can ,e ie4ed as the process of partitioningthe total sum of squares and the degrees of freedomthe total sum of squares and the degrees of freedominto their corresponding sources: treatments and errinto their corresponding sources: treatments and err
A!%VA can ,e ie4ed as the process of partitioningA!%VA can ,e ie4ed as the process of partitioningthe total sum of squares and the degrees of freedomthe total sum of squares and the degrees of freedominto their corresponding sources: treatments and errointo their corresponding sources: treatments and erro
0iiding the sum of squares ,y the appropriate0iiding the sum of squares ,y the appropriatedegrees of freedom proides the ariance estimatesdegrees of freedom proides the ariance estimatesand theand the FFalue used to test the hypothesis of equalalue used to test the hypothesis of equalpopulation means/population means/
0iiding the sum of squares ,y the appropriate0iiding the sum of squares ,y the appropriatedegrees of freedom proides the ariance estimatesdegrees of freedom proides the ariance estimatesand theand the FFalue used to test the hypothesis of equalalue used to test the hypothesis of equalpopulation means/population means/
f h li fT t f th E lit f kk l iP l ti
-
7/21/2019 ANOVA Dasar
18/37
1>
E2ample: #eed ManufacturingE2ample: #eed Manufacturing
Test for the Equality ofTest for the Equality of kk PopulationPopulationMeansMeans
Hanet #eed 4ould li)e to )no4 ifHanet #eed 4ould li)e to )no4 if
there is any signicant di9erence inthere is any signicant di9erence in
the mean num,er of hours 4or)ed perthe mean num,er of hours 4or)ed per
4ee) for the department managers4ee) for the department managers
at her three manufacturing plantsat her three manufacturing plants
$in ?u9alo8 Pitts,urgh8 and 0etroit&/$in ?u9alo8 Pitts,urgh8 and 0etroit&/
T f h E li fT t f th E lit f kk P l iP l ti
-
7/21/2019 ANOVA Dasar
19/37
1@
E2ample: #eed ManufacturingE2ample: #eed Manufacturing
Test for the Equality ofTest for the Equality of kk PopulationPopulationMeansMeans
A simple random sample of eA simple random sample of e
managers from each of the three plantsmanagers from each of the three plants
4as ta)en and the num,er of hours4as ta)en and the num,er of hours
4or)ed ,y each manager for the4or)ed ,y each manager for the
preious 4ee) is sho4n on the ne2tpreious 4ee) is sho4n on the ne2t
slide/slide/
Conduct anConduct an FFtest usingtest using 6 /5/6 /5/
T t f th E lit fT t f th E lit f kk P l tiP l ti
-
7/21/2019 ANOVA Dasar
20/37
5
11
..
77
7>7>
77
==
77
-
7/21/2019 ANOVA Dasar
21/37
1
Test for the Equality ofTest for the Equality of kk PopulationPopulationMeansMeans
HH55:: 1166 66 ..
HHaa: !ot all the means are equal: !ot all the means are equal
4here:4here: 11 6 mean num,er of hours 4or)ed per6 mean num,er of hours 4or)ed per
4ee) ,y the managers at Plant 14ee) ,y the managers at Plant 1 6 mean num,er of hours 4or)ed per6 mean num,er of hours 4or)ed per
4ee) ,y the managers at Plant 4ee) ,y the managers at Plant .. 6 mean num,er of hours 4or)ed per6 mean num,er of hours 4or)ed per
4ee) ,y the managers at Plant .4ee) ,y the managers at Plant .
1/ 0eelop the hypotheses/1/ 0eelop the hypotheses/
pp-Value and Critical Value Approaches-Value and Critical Value Approaches
T t f th E lit fT t f th E lit f kk P l tiP l ti
-
7/21/2019 ANOVA Dasar
22/37
/ Specify the leel of signicance// Specify the leel of signicance/ 6 /56 /5
Test for the Equality ofTest for the Equality of kk PopulationPopulationMeansMeans
pp-Value and Critical Value Approaches-Value and Critical Value Approaches
./ Compute the alue of the test statistic/./ Compute the alue of the test statistic/
MST# 6 7@5$. - 1& 6 7MST# 6 7@5$. - 1& 6 7SST# 6 $ -
-
7/21/2019 ANOVA Dasar
23/37
.
./ Compute the alue of the test statistic/./ Compute the alue of the test statistic/
Test for the Equality ofTest for the Equality of kk PopulationPopulationMeansMeans
MSE 6 .5>$1 - .& 6 /
-
7/21/2019 ANOVA Dasar
24/37
7
TreatmentTreatment
ErrorError
TotalTotal
7@57@5
.5>.5>
=@>=@>
11
1717
77
/
-
7/21/2019 ANOVA Dasar
25/37
Test for the Equality ofTest for the Equality of kk PopulationPopulationMeansMeans
/ 0etermine 4hether to re(ect/ 0etermine 4hether to re(ect HH55//
3e hae suLcient eidence to conclude that3e hae suLcient eidence to conclude thatthe mean num,er of hours 4or)ed per 4ee)the mean num,er of hours 4or)ed per 4ee),y department managers is not the same at,y department managers is not the same atall . plant/all . plant/
TheThepp-alue-alue DD/58/58so 4e re(ectso 4e re(ect HH55//
3ith numerator d/f/ and 13ith numerator d/f/ and 1denominator d/f/8denominator d/f/8
thethepp-alue is /51 for-alue is /51 for FF6
-
7/21/2019 ANOVA Dasar
26/37
@8 4e re(ect./>@8 4e re(ect HH55//
Critical Value ApproachCritical Value Approach
7/ 0etermine the critical alue and re(ection rule/7/ 0etermine the critical alue and re(ection rule/
#e(ect#e(ect HH55ifif FF./>@./>@
Test for the Equality ofTest for the Equality of kk PopulationPopulationMeansMeans
3e hae suLcient eidence to conclude that3e hae suLcient eidence to conclude thatthe mean num,er of hours 4or)ed per 4ee)the mean num,er of hours 4or)ed per 4ee),y department managers is not the same at,y department managers is not the same atall . plant/all . plant/
?ased on an?ased on an FFdistri,ution 4ith numeratordistri,ution 4ith numeratord/f/ and 1 denominator d/f/8d/f/ and 1 denominator d/f/8 FF/5/56 ./>@/6 ./>@/
-
7/21/2019 ANOVA Dasar
27/37
=
Multiple Comparison ProceduresMultiple Comparison Procedures
Suppose that analysis of ariance hasSuppose that analysis of ariance has
proided statistical eidence to re(ect the nullproided statistical eidence to re(ect the nullhypothesis of equal population means/hypothesis of equal population means/
+isherFs least signicant di9erence $"S0& procedure can+isherFs least signicant di9erence $"S0& procedure can,e used to determine 4here the di9erences occur/,e used to determine 4here the di9erences occur/
-
7/21/2019 ANOVA Dasar
28/37
>
+isherFs "S0 Procedure+isherFs "S0 Procedure
1 1MSE$ &
i j
i j
x xt
n n
=
+1 1MSE$ &
i j
i j
x xt
n n
=
+
Test StatisticTest Statistic
ypothesesypotheses
5 : i jH 5 : i jH
:a i jH :a i jH
-
7/21/2019 ANOVA Dasar
29/37
@
+isherFs "S0 Procedure+isherFs "S0 Procedure
4here the alue of4here the alue of ttaa is ,ased on ais ,ased on a
ttdistri,ution 4ithdistri,ution 4ith nnTT-- kkdegrees of freedom/degrees of freedom/
#e(ection #ule#e(ection #ule
#e(ect#e(ect HH55ififpp-alue-alue DD
pp-alue Approach:-alue Approach:
Critical Value Approach:Critical Value Approach:
#e(ect#e(ect HH55ifif ttD -D -ttaa oror tt ttaa
+isherFs "S0 Procedure+isherFs "S0 Procedure
-
7/21/2019 ANOVA Dasar
30/37
.5
Test StatisticTest Statistic
+isher s "S0 Procedure+isher s "S0 Procedure?ased on the Test Statistic?ased on the Test Statisticxxii--xxjj
A 1 1"S0 MSE$ &i j
t n n= +A 1 1"S0 MSE$ &i jt n n= +
4here4here
i jx xi jx x
#e(ect#e(ect HH55if "S0if "S0i jx xi jx x
ypothesesypotheses
#e(ection #ule#e(ection #ule
5 : i jH 5 : i jH
:a i jH :a i jH
+isherFs "S0 Procedure+isherFs "S0 Procedure
-
7/21/2019 ANOVA Dasar
31/37
.1
+isher s "S0 Procedure+isher s "S0 Procedure?ased on the Test Statistic?ased on the Test Statisticxxii--xxjj
E2ample: #eed ManufacturingE2ample: #eed Manufacturing
#ecall that Hanet #eed 4ants to )no4#ecall that Hanet #eed 4ants to )no4
if there is any signicant di9erence inif there is any signicant di9erence in
the mean num,er of hours 4or)ed perthe mean num,er of hours 4or)ed per
4ee) for the department managers4ee) for the department managers
at her three manufacturing plants/at her three manufacturing plants/
Analysis of ariance has proidedAnalysis of ariance has proided
statistical eidence to re(ect the nullstatistical eidence to re(ect the null
hypothesis of equal population means/hypothesis of equal population means/
+isherFs least signicant di9erence $"S0&+isherFs least signicant di9erence $"S0&
procedureprocedure
can ,e used to determine 4here the di9erencescan ,e used to determine 4here the di9erences
occur/occur/
+isherFs "S0 Procedure+isherFs "S0 Procedure
-
7/21/2019 ANOVA Dasar
32/37
.
+or+or 6 /5 and6 /5 and nnTT-- kk6 1 G . 6 16 1 G . 6 1
degrees of freedom8degrees of freedom8 tt..55 6 /1=@6 /1=@
"S0 = + = 1=@ :
-
7/21/2019 ANOVA Dasar
33/37
..
"S0 for Plants 1 and "S0 for Plants 1 and
+isher s "S0 Procedure+isher s "S0 Procedure?ased on the Test Statistic?ased on the Test Statisticxxii--xxjj
' ConclusionConclusion
' Test StatisticTest Statistic
1 x x1 x x 6 N6 N N 6 1.N 6 1.
#e(ect#e(ect HH55ifif 1 x x
1 x x
' #e(ection #ule#e(ection #ule
5 1 :H 5 1 :H
1 :aH 1 :aH
' ypotheses $A&ypotheses $A&
The mean num,er of hours 4or)ed at Plant 1 isThe mean num,er of hours 4or)ed at Plant 1 is
not equalnot equalto the mean num,er 4or)ed at Plant /to the mean num,er 4or)ed at Plant /
+isherFs "S0 Procedure+isherFs "S0 Procedure
-
7/21/2019 ANOVA Dasar
34/37
.7
"S0 for Plants 1 and ."S0 for Plants 1 and .
+isher s "S0 Procedure+isher s "S0 Procedure?ased on the Test Statistic?ased on the Test Statisticxxii--xxjj
' ConclusionConclusion
' Test StatisticTest Statistic
1 .x x1 .x x 6 N6 N =N 6 =N 6
#e(ect#e(ect HH55ifif 1 .x x
1 .x x
' #e(ection #ule#e(ection #ule
5 1 .:H 5 1 .:H
1 .:aH 1 .:aH
' ypotheses $?&ypotheses $?&
There isThere is no signicant di9erenceno signicant di9erence,et4een the me,et4een the me
num,er of hours 4or)ed at Plant 1 andnum,er of hours 4or)ed at Plant 1 and the meathe mea
num,er of hours 4or)ed at Plant ./num,er of hours 4or)ed at Plant ./
+isherFs "S0 Procedure+isherFs "S0 Procedure
-
7/21/2019 ANOVA Dasar
35/37
.
"S0 for Plants and ."S0 for Plants and .
+isher s "S0 Procedure+isher s "S0 Procedure?ased on the Test Statistic?ased on the Test Statisticxxii--xxjj
' ConclusionConclusion
' Test StatisticTest Statistic
.x x .x x 6 N6 N =N 6 11=N 6 11
#e(ect#e(ect HH55ifif
.x x
.x x
' #e(ection #ule#e(ection #ule
5 .:H 5 .:H
.:aH .:aH
' ypotheses $C&ypotheses $C&
The mean num,er of hours 4or)ed at Plant isThe mean num,er of hours 4or)ed at Plant is
not equalnot equalto the mean num,er 4or)ed at Plant ./to the mean num,er 4or)ed at Plant ./
-
7/21/2019 ANOVA Dasar
36/37
.
-
7/21/2019 ANOVA Dasar
37/37
End of ChapterEnd of Chapter