




下載本文檔
版權說明:本文檔由用戶提供并上傳,收益歸屬內容提供方,若內容存在侵權,請進行舉報或認領
文檔簡介
Business
Statistics:
ACourse6th
EditionChapter
10Two-Sample
TestsandOne-Way
ANOVA兩樣本檢驗和單邊方差分析Copyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallChap
10-1Copyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallChap
10-2Learning
ObjectivesIn
this
chapter,
you
learnHow
to
use
hypothesis
testing
for
comparing
thedifferencebetween:如何利用統計假設來比較兩個不一樣的:The
means
of
two
independent
populations
兩個獨立的總體的均值The
means
of
two
related
populations
兩個相關的總體的均值The
proportions
of
two
independent
populations兩個獨立的總體的比例The
variances
of
two
independent
populations兩個獨立的總體的方差The
means
of
more
than
two
populations
兩個以上總體的均值Two-Sample
TestsPopulationMeans,IndependentSamplesPopulationMeans,RelatedSamplesPopulationVariancesGroup
1
vs.Group
2Samegroupbefore
vs.
aftertreatmentVariance
1
vs.Variance
2Examples:PopulationProportionsProportion
1
vs.Proportion
2DCOVATwo-Sample
TestsCopyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallChap
10-3Difference
Between
Two
MeansDCOVAPopulation
means,independent
samples獨立樣本總體均值檢驗Goal:Test
hypothesis
or
form
檢驗的假設或形式:a
confidence
interval
for
thedifference
between
two
populationmeans兩總體均值差異的置信區間,μ1
–μ2The
point
estimate
for
thedifference
isX1
–
X2*σ1
and
σ2
unknown,assumed
equalσ1
and
σ2
unknown,not
assumed
equalCopyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallChap
10-4Difference
Between
Two
Means:Independent
SamplesPopulation
means,independentsamples*Use
Sp
to
estimate
unknownσ.Use
a
Pooled-Variance
ttest.混合方差t檢驗σ1
and
σ2
unknown,assumed
equalσ1
and
σ2
unknown,not
assumed
equalUse S1
and
S2
to
estimate
unknownσ1
and
σ2.Use
a
Separate-Variancet
test.不同方差t檢驗Different
data
sourcesUnrelatedIndependent
兩總體間不相關Sample
selected
from
oneCopyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallChap
10-5population
has
no
effect
on
thesample
selected
from
the
otherpopulationDCOVACopyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallChap
10-6Hypothesis
Tests
forTwo
Population
Means兩總體均值假設檢驗Lower-tail
test:左側檢驗
H0:μ1
μ2H1:μ1
<μ2i.e.,H0:
μ1
–
μ2
0H1:
μ1
–
μ2
<
0Upper-tail
test:右側檢驗
H0:μ1
≤μ2H1:μ1
>μ2i.e.,H0:
μ1
–
μ2
≤
0H1:
μ1
–
μ2
>
0Two-tail
test:雙側檢驗
H0:μ1
=μ2H1:
μ1
≠
μ2i.e.,H0:
μ1
–
μ2
=
0H1:
μ1
–
μ2
≠
0Two
Population
Means,
Independent
SamplesDCOVATwo
Population
Means,
Independent
SamplesLower-tail
test:H0:
μ1
–
μ2
0H1:
μ1
–
μ2
<
0Upper-tail
test:H0:
μ1
–
μ2
≤
0H1:
μ1
–
μ2
>
0Two-tail
test:H0:
μ1
–
μ2
=
0H1:
μ1
–
μ2
≠
0-ttReject
H0
if
tSTAT
<
-tReject
H0
if
tSTAT
>
t
/2
/2Copyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallChap
10-7-t/2
t/2Reject
H0
if
tSTAT
<
-t/2or
tSTAT
>
t/2Hypothesis
tests
for
μ1
–
μ2DCOVAPopulation
means,independentsamplesAssumptions:Samples
are
randomlyandindependently
drawn樣品隨機,獨立抽取Populations
are
normallydistributed
or
both
samplesizes
are
at
least30人群呈正態分布或兩個樣本量均至少為30Population
variances
areunknown
but
assumed
equal總體方差未知,但假設他們相等Hypothesis
tests
for
μ1
-
μ2
with
σ1and
σ2
unknown
and
assumed
equalDCOVA*σ1
and
σ2
unknown,assumed
equalσ1
and
σ2
unknown,not
assumed
equalCopyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallChap
10-8Population
means,independentsamples混合方差(continued)The
pooled
variance
is:Where
tSTAThas
d.f.
=
(n1
+
n2
–
2)1
22221
1
2p(n
1)
(n
1)n
1S
n
1
SS2*
?
The
test
statistic
is:σ1
and
σ2
unknown,assumed
equalσ1
and
σ2
unknown,not
assumed
equalHypothesis
tests
for
μ1
-
μ2
withσ1and
σ2
unknown
and
assumed
equal2Copyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallChap
10-911
2STAT1 2
p
n
nS2
1
1
X
X
μ
μ
t
Population
means,independentsamples
2p
n1
n2
1
1/2X1
X2
tS置信區間The
confidence
intervalforμ1
–μ2
is:Where
tα/2
has
d.f.
=
n1
+n2
–
2*Confidence
interval
for
μ1
-
μ2
with
σ1and
σ2
unknown
and
assumed
equalDCOVAσ1
and
σ2
unknown,assumed
equalσ1
and
σ2
unknown,not
assumed
equalCopyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallChap
10-10Pooled-Variance
t
Test
Example東北游客游客Sample
Size2125Sample
mean3.272.53Sample
stddev1.301.16是否存在差異?DCOVA據說來自和東北的游客,在家界游玩的滿意度存在差異,旅游學院老師希望通過問卷檢驗這一結論的真實性,收集了以下數據:Assuming
both
populations
areapproxima y
normal
with
equal
variances,
istherea
difference
in
mean
yield
(
=
0.05)?假設兩總體呈正態分布,方差相同,請問兩總體均值Copyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallChap
10-11Pooled-Variancet
Test
Example:Calculating
the
Test
Statistic(continued)1.50212122211
2(n
1)
(n
1) (21
-
1)
(25
1)n
1S
n
1S
21
11.302
25
11.162
S
2P3.27
2.53
02.040t
X1
X2
μ1
μ2
Copyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallChap
10-12STAT1.5021
1
21
25
1
S
n1
n2
2p1
1
The
test
statistic
is:H0:
μ1
-
μ2
=
0 i.e.
(μ1
=
μ2)H1:
μ1
-
μ2
≠
0 i.e.
(μ1
≠μ2)DCOVAPooled-Variancet
Test
Example:Hypothesis
Test
SolutionH0:
μ1
-
μ2
=
0 i.e.
(μ1
=
μ2)H1:
μ1
-
μ2
≠0 i.e.
(μ1
≠μ2)
=
0.05df
=
21
+
25
-
2
=
44Critical
Values:
t
=
±
2.0154Test
Statistic:Decision:Reject
H0
at
=
0.05Conclusion:There
is
evidence
of
adifference
in
means.t02.0154-2.0154.025Reject
H0.0252.0402.040Copyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallChap
10-13DCOVAReject
H03.27
2.53STATt
21
25
1.5021
1
1
Pooled-Variancet
Test
Example:Confidence
Interval
for
μ1
-
μ2Copyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallChap
10-14DCOVASince
we
rejected
H0
can
webe
95%
confident
that
μDB
>μGD?95%
Confidence
Interval
for
μDB
-μGDSince
0
is
less
than
the
entire
interval,
wecan
be
95%confident
that
μNYSED
>
μGDSp
n1
n2
2
1 1
0.74
2.0154
0.3628
(0.009,
1.471)X1
X2
t/2Population
means,independentsamplesHypothesis
tests
for
μ1
-
μ2
withσ1and
σ2
unknown,
not
assumed
equal*σ1
and
σ2
unknown,assumed
equalσ1
and
σ2
unknown,not
assumed
equalDCOVACopyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallChap
10-15Assumptions:Samples
arerandomly
andindependently
drawn樣品隨機,獨立抽取Populations
arenormallydistributed
or
both
samplesizes
are
at
least
30人群呈正態分布或兩個樣本量均至少為30Population
variances
areunknownand cannot
beassumed
tobe
equal總體方差未知,不能假設他們相等Population
means,independentsamples*σ1
and
σ2
unknown,assumed
equalσ1
and
σ2
unknown,Hypothesis
tests
for
μ1
-
μ2
with
σ1
andnot
assumed
equalCopyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallChap
10-16σ2
unknown
and
not
assumed
equal(continued)DCOVAThe
formulae
for
this
test
arenot
covered
in
this
book.See
reference
8
from
thischapter
for
moredetails.This
test
utilizes
twoseparatesample
variances
to
estimatethe
degrees
of
freedom
for
thet
test
Or,
if
not
Normal,use
large
samples
大樣本Copyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallChap
10-17Related
Populations
相關總體The
Paired
Difference
TestDCOVATests
Means
of
2
Related
PopulationsPaired
or
matched
samples
配對樣本,期望與滿意度Repeated
measures(before/after)重復檢驗,事前事后
Use
difference
between
paired
values:兩個配對值之間的差異Eliminates
Variation
Among
Subjects消除兩主體之間的變異Assumptions:Both
Populations
Are
Normally
Distributed
正態分布RelatedsamplesDi
=
X1i
-X2iRelated
PopulationsThe
Paired
Difference
TestThe
ith
paired
difference
is Di
,
whereRelatedsamplesDi
=
X1i
-
X2iThe
point
estimate
for
the
paireddifference
population
mean
μD
is匹配均值差異的點估計D:n
D
iD
i
1
nn
is
the
number
of
pairs
in
the
paired
sampleN是匹配樣本的成對的數量n2(D
D)SD
i
i1
n
1The
sample
standarddeviation樣本標準差is
SDCopyright
?2013
Pearson
Education,
Inc.
publishing
asPrentice
HallChap
10-18DCOVA(continued)The
test
statistic
for
μD
is:Copyright
?2013
Pearson
Education,
Inc.
publishing
asPrentice
HallChap
10-19PairedsamplestSTAT
D
μ
DSDnWhere
tSTAT
has n
-
1
d.f.The
Paired
Difference
Test:Finding
tSTATDCOVALower-tail
test:H0:
μD
0H1:
μD
<
0Upper-tail
test:H0:
μD
≤
0H1:
μD
>
0Two-tail
test:H0:
μD
=
0H1:
μD
≠
0PairedSamplesThe
Paired
Difference
Test:Possible
Hypotheses-ttReject
H0
if
tSTAT
<
-tReject
H0
if
tSTAT
>
t
/2/2-t/2
t/2Reject
H0
if
tSTAT
<
-tor
tSTAT
>
tWhere
tSTAT
has n
-
1
d.f.Copyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallChap
10-20DCOVAThe
confidence
interval
for
μD
isPairedsamplesSD
Copyright
?2013
Pearson
Education,
Inc.
publishing
asPrentice
HallChap
10-21n
2i(D
D)
i1
n
1SDn
/
2D
twhereThe
Paired
DifferenceConfidence
IntervalDCOVA旅游有助于提高員工的服務質量,降量。我們收集了員工旅游前和旅游后一通常認為低顧客的個月的顧客量,數據如下,請問 旅游對員工服務質量的是否存在顯著影響:Paired
Difference
Test匹配t檢驗:ExampleSalesNumber
of
Complaints:Before
(1)
After
(2)(2)
-
(1)Difference,
DiC.B.T.F.M.H.R.K.M.O.62030446200-
2-14-
10-
4-21D
=
Din
5.672(D
D)
in
1DS
=
-4.2DCOVACopyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallChap
10-22Has
the
employees’
travel
experience
made
a
difference
inthe
number
of
complaints
(at
the
0.01
level)?1.66SD/
n
5.67/
5
D
μD
4.2
0
tSTATH0:
μD
=0H1: μD
0/2-
4.604
4.604Decision:
Do
not
rejectH0(tstat
is
not
in
the
reject
region)Conclusion:
There
isinsufficient
evidence
thereis
significant
change
in
thenumber
ofcomplaints.
=
.01 D
=
-
4.2t0.005
=
±
4.604d.f.
=
n
-
1
=
4Test
Statistic:Paired
Difference
Test:SolutionReject
Reject/2-
1.66DCOVACopyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallChap
10-23Two
Population
Proportions兩總體比例Copyright
?2013
Pearson
Education,
Inc.
publishing
asPrentice
HallChap
10-24difference
between
two
populationproportions,
兩總體比例的差異DCOVAGoal:
test
a
hypothesis
or
form
檢驗的假設:a
confidence
interval
for
theπ1
–
π2The
point
estimate
for
thedifference
這個差異的點估計isPopulationproportionsp1
p2Two
Population
Proportions兩總體比例Populationproportionsp
X1
X2n1
n2Copyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallChap
10-25In
the
null
hypothesis
we
assume
thenull
hypothesis
is
true,
so
we
assume
π1=
π2
and
pool
the
two
sampleestimates零假設:假設兩總體比例相等The
pooled
estimate
for
theoverallproportion總的比例的混合估計is:where
X1
and
X2
are
the
number
of
items
ofinterest
in
samples
1
and
2DCOVATwo
Population
ProportionsPopulationproportions
n1
n2
p
(1
p)1
1
p
p
π
π
12
1
2ZSTATThe
test
statistic
forπ1
–
π2is
a
Z
statistic:(continued)2Copyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallChap
10-262111n
n2p
X1
X2n,
p
X
2n,
p
X1whereDCOVACopyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallChap
10-27Hypothesis
Tests
forTwo
Population
ProportionsPopulation
proportionsLower-tail
test:H0:
π1
π2H1:
π1
<
π2i.e.,H0:
π1
–
π2
0H1:
π1
–
π2
<
0Upper-tail
test:H0:
π1
≤
π2H1:
π1
>
π2i.e.,H0:
π1
–
π2
≤0H1:
π1
–
π2
>
0Two-tail
test:H0:
π1
=
π2H1:
π1
≠
π2i.e.,H0:
π1
–
π2
=
0H1:
π1
–
π2
≠
0DCOVAHypothesis
Tests
forTwo
Population
ProportionsPopulation
proportionsLower-tail
test:H0:
π1
–
π2
0H1:
π1
–
π2
<
0Upper-tail
test:H0:
π1
–
π2
≤
0H1:
π1
–
π2
>
0Two-tail
test:H0:
π1
–
π2
=
0H1:
π1
–
π2
≠
0-zzReject
H0
if
ZSTAT
<
-ZReject
H0
if
ZSTAT
>
Z
/2
/2Copyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallChap
10-28-z/2
z/2Reject
H0
if
ZSTAT
<
-Zor
ZSTAT
>
Z(continued)DCOVAHypothesis
Test
Example:Two
population
Proportions游客和女性游客,購物的比例是否存在顯著差異?In
a
random
sample,
36
of
72
menand35
of
50
women
indicated
they
wouldpurchase.Test
at
the
.05
levelofsignificanceDCOVACopyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallChap
10-29The
hypothesis
:H0:
π1
–
π2
=
0H1:
π1
–
π2
≠
0(the
two
proportions
are
equal)(there
is
a
significant
difference
between
proportions)The
sample
proportions
are:Men:Women:p1
=
36/72
=
0.50p2
=
35/50
=
0.70
36
35
71
0.58272
50
122p
X1
X2n1
n2The
pooled
estimate
for
the
overall
proportion
is:Hypothesis
Test
Example:Two
population
Proportions(continued)DCOVACopyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallChap
10-30The
test
statistic
for
π1
–
π2
is:(continued).025961.96.025-1.-2.20Conclusion:
There
isevidence
of
a
difference
inproportions
who
willpurchase
between
men
andwomen.
和女性的購物的比例存在顯著差異zSTAT
72
50
.582(
1
.582
)
1
1
n1
n2
.50
.70
0
2.20
Decision:
Reject
H0
1 1
p(
1
p)
p
p
π
π
1
2
1
2Reject
H0Reject
H0Hypothesis
Test
Example:Two
population
Proportions
DCOVACritical
Values
=
±1.96For
=
.05Copyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallChap
10-31Confidence
IntervalforTwo
Population
ProportionsDCOVAPopulationproportions1
2Copyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallChap
10-32n
np1
(1
p1
)
p2
(1
p2
)p
p
Z1
2/2The
confidence
interval
forπ1
–
π2
is:Difference
Between
Two
Means:Independent
SamplesPopulation
means,independentsamples*Use
Sp
to
estimate
unknownσ.Use
a
Pooled-Variance
ttest.混合方差t檢驗σ1
and
σ2
unknown,assumed
equalσ1
and
σ2
unknown,not
assumed
equalUse S1
and
S2
to
estimate
unknownσ1
and
σ2.Use
a
Separate-Variancet
test.不同方差t檢驗Different
data
sourcesUnrelatedIndependent
兩總體間不相關Sample
selected
from
oneCopyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallChap
10-33population
has
no
effect
on
thesample
selected
from
the
otherpopulationDCOVATesting
for
the
Ratio
Of
TwoPopulation
Variances兩總體方差齊性檢驗Tests
for
TwoPopulationVariancesF test
statistic1
2H0:
σ
2
=
σ
2H1:
σ12
≠
σ221
2H0:
σ
2
≤
σ
212H1:
σ
2
>
σ
2*HypothesesFSTAT1n1
=
sample
size
of
sample
1S
2
=
Variance
of
sample
1(the
larger
sample
variance)S
2
=
Variance
of
sample
2
(the
smaller
sample
variance)2n2
=
sample
size
of
sample
2n1
–1
=
numerator
degrees
offreedomn2
–
1
=
denominator
degrees
of
freedomCopyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallWhere:DCOVAChap
10-342S22S1The
F
critical
value
is
found
from
the
F
tableF臨界值可以從F表中查找P483There
are
two
degrees
of
freedom
required:
numeratorand
denominator
度由分子和分母決定The
larger
sample
variance
isalways
thenumerator大的是分子WhenIn
the
F
table,numerator
degrees
of
freedom
determine
the
column分子是列denominator
degrees
of
freedom
determine
the
row分母是行The
F
Distribution1
1df =
n –
1
;df2
=n2
–12S
2S
2FSTAT
1DCOVACopyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallChap
10-35Finding
the
Rejection
Region12H1:
σ
2
≠
σ
21
2
1
2H0:
σ
2
=
σ
2
H0:
σ
2
≤
σ
212H1:
σ
2
>
σ
2F0FαReject
H0Do
notreject
H0Reject
H0
if
FSTAT
>
FαF0/2Reject
H0Do
notreject
H0
Fα/2Reject
H0
if
FSTAT
>
Fα/2Copyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallChap
10-36DCOVAF
Test:
An
ExampleDCOVA假設我們現在正在對來自不同地區的游客對家界旅游的滿意度.以下數據是華北游客滿意度和華南游客滿意度的數據。華北游客華南游客Number2125Mean3.272.53Std
dev1.301.16華北游客和華南游客滿意度的方差在
=0.05
的水平上是否存在差異呢?Copyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallChap
10-37F
Test:
Example
SolutionCopyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallChap
10-38(there
is
no
difference
between
variances)(there
is
a
difference
between
variances)DCOVAForm
the
hypothesis
test:Find
the
F
critical
value
for
=
0.05:Numerator
d.f.
=
n1
–
1
=
21
–1
=
20Denominator
d.f.
=
n2
–
1
=
25
–1
=
24Fα/2
=
F.025,
20,
24
=
2.33H
:
σ
2
σ
21
1
2H
:
σ
2
σ
20
1
2Chap
10-39The
test
statistic
is:1.2561.1622S
2S
21.302FSTAT
1
/2
=
.025Reject
H0F0.025=2.33Do
notreject
H0H0:
σ
2
=
σ
21
2H1:
σ12
≠
σ22F
Test:
ExampleSolution0FSTAT
=
1.256
is
not
in
the
rejectionregion,
so
we
do
not
reject
H0DCOVA(continued)Conclusion:
There
is
insufficient
evidence
of
adifference
in
variances
at
=
.05在0.05的顯著性水平下,沒有充分的
表明,華北游客的滿意度和華南游客的滿意度方差存在差異FOne-Way ysis
Of
Variance(ANOVA)Setting單向方差分D析COVAWant
to
examine
differences
among
more
than
two
groups檢驗上個以上總體間差異The
groups
involvedare
classified
according
to
levels
of
afactor
of
interest(numerical
or
categorical)根據因素的層級進行的分組Different
levels
produce
different
groups不同的層級就是不同的組Think
of
each
group
as
asample
from
a
differentpopulation講每個組假想成一個樣本Observe
effects
on
the
dependentvariableAre
the
groups
thesame?他們是一樣的嗎?
When
there
is
only
1
factor
the
design
is
called
a
comple
yrandomizeddesign當只有一個因素的時候,稱為完全隨機設計Copyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallChap
10-40One-Way ysis
of
Variance單向方差分析Evaluate
the
difference
among
the
means
of
threeor
more
groups評估三個或以上的組之間的差異Examples:
不同出游方式的游客的購物花費AssumptionsPopulations
are
normally
distributed
正態分布Populations
have
equal
variances
方差齊(相等)Samples
are
randomly
and
independently
drawn樣本隨機獨立抽取DCOVACopyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallChap
10-41All
population
means
are
equal
所有總體均值相等i.e.,
no
factor
effect
(no
variation
in
means
among
groups)各組不相互影響(各組均值方差不存在差異)At
least
one
population
mean
isdifferent
至少有一個總體均值是不一樣的i.e.,there
is
a
factor
effect
存在因素影響Does
not
mean
that
all
population
means
are
different(some
pairs
may
be
the
same)并不意味著所有組的均值都不一樣H0
:
μ1
μ2
μ3
μcCopyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallChap
10-42Hypotheses
of
One-Way
ANOVADCOVAH1
:
Not
all
of
the
population
means
are
the
sameOne-Way
ANOVAThe
Null
Hypothesis
is
TrueAll
Means
are
thesame:(No
Factor
Effect)H0
:
μ1
μ2
μ3
μcH1
:
Not
all
μj
are
the
sameCopyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallChap
10-43μ1
μ
2
μ3DCOVAOne-Way
ANOVAμ1
μ2
μ3H0
:
μ1
μ2
μ3
μcH1
:
Not
all
μj
are
the
sameThe
Null
Hypothesisis
NOT
trueAt
least
one
of
the
means
is
different(Factor
Effect
is
present)orμ1
μ2
μ3Copyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallChap
10-44DCOVA(continued)Copyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallChap
10-45Partitioning
the
Variation離差的構成Total
variation
can
be
split
into
twoparts:SST
=Total
Sum
of
Squares總離差
(Total
variation)SSA
=Sum
of
Squares
Among
Groups組間離差
(Among-group
variation)SSW
=Sumof
Squares
Within
Groups組內離差
(Within-group
variation)SST
=
SSA
+
SSWDCOVAPartitioning
the
VariationTotal
Variation
=the
aggregate
variationof
the
individualdata
values
across
the
various
factor
levels(SST)不同因素層級間所有數據的總變異Among-Group
Variation
=
variation
among
the
factorsample
means(SSA)因素各層級均值的變異Within-Group
Variation
=
variation
that
exists
amongthe
data
values
withina
particular
factor
level(SSW)各因素層 所有數據值之間的變異/離差SST
=
SSA
+
SSW(continued)DCOVACopyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallChap
10-46Variation
Due
toFactor
(SSA)組間離差Copyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallChap
10-47Variation
Due
to
RandomError
(SSW)隨機誤差Total
Variation
(SST)Partition
of
Total
VariationDCOVA=+Total
Sum
of
Squares離差平方和cnjSST
(
X
ij
X
)2j
1
i
1Where:SST
=
Total
sum
of
squaresc
=
number
of
groups
or
levelsnj
=
number
of
observations
in
group
jXij
=
ith
observation
from
group
jX
=
grand
mean
(mean
of
all
datavalues)SSTCopyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallChap
10-48=
SSA
+
SSWDCOVATotal
Variation
總離差(總變異)(continued)Group
1Group
2Group
3Response,
XXCopyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallChap
10-49Among-Group
Variation
組間離差Where:SSA
=
Sum
of
squares
among
groupsc
=
number
ofgroupsnj
=
sample
size
from
group
jXj
=
sample
mean
from
group
jX
=
grand
mean
(meanof
alldata
values)2jc
n
(
X
X
)SSA
jj
1SST
=
SSA
+
SSWCopyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallChap
10-50Among-Group
VariationVariation
Due
toDifferences
Among
Groupsi
j2jc
jn
(
X
X
)SSA
j
1MSA
SSAc
1Mean
Square
Among
間均方=
SSA/degrees
of
freedom(continued)DCOVACopyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallChap
10-51Among-Group
VariationGroup
1Group
2Group
31XX2DCOVA(continued)SSA
n
(
X
1
X
)2
n
(
X
2
X
)2
n
(
X
c
X
)21
2
cResponse,
XX
3
XCopyright
?2013
Pearson
Education,
Inc.
publishing
as
Prentice
HallChap
10-52Within-Group
Variation組內離差Where:SSW
=
Sum
of
squareswithin
groupsc
溫馨提示
- 1. 本站所有資源如無特殊說明,都需要本地電腦安裝OFFICE2007和PDF閱讀器。圖紙軟件為CAD,CAXA,PROE,UG,SolidWorks等.壓縮文件請下載最新的WinRAR軟件解壓。
- 2. 本站的文檔不包含任何第三方提供的附件圖紙等,如果需要附件,請聯系上傳者。文件的所有權益歸上傳用戶所有。
- 3. 本站RAR壓縮包中若帶圖紙,網頁內容里面會有圖紙預覽,若沒有圖紙預覽就沒有圖紙。
- 4. 未經權益所有人同意不得將文件中的內容挪作商業或盈利用途。
- 5. 人人文庫網僅提供信息存儲空間,僅對用戶上傳內容的表現方式做保護處理,對用戶上傳分享的文檔內容本身不做任何修改或編輯,并不能對任何下載內容負責。
- 6. 下載文件中如有侵權或不適當內容,請與我們聯系,我們立即糾正。
- 7. 本站不保證下載資源的準確性、安全性和完整性, 同時也不承擔用戶因使用這些下載資源對自己和他人造成任何形式的傷害或損失。
最新文檔
- 家鄉環保企業設備管理制度
- 小型培訓機構老師管理制度
- 工會驛站智能化設備管理制度
- 幼兒園學??倓展芾碇贫?/a>
- 幼兒園帶藥入園管理制度
- 幼兒園消防責任管理制度
- 幼兒園用火消防管理制度
- 2025-2030年中國系泊檢查行業市場現狀供需分析及投資評估規劃分析研究報告
- 2025-2030年中國空氣消除器行業市場現狀供需分析及投資評估規劃分析研究報告
- 2025-2030年中國硫酸鈣板活動地板行業市場現狀供需分析及投資評估規劃分析研究報告
- GB/T 18024.6-2010煤礦機械技術文件用圖形符號第6部分:露天礦機械圖形符號
- iso-7010-safety-signpdf原版標準文件
- 灌砂法壓實度檢測記錄表(自動計算表)
- 江蘇省泰州市2022年中考生物試題真題(含答案+解析)
- 中國慢性髓性白血病診療指南更新
- 《民法典》合同編實務培訓課件
- 第7章食品原料的采購與貯存管理ppt課件
- 食品安全承諾書
- 湘教版高中美術選修:美術鑒賞 第一單元 第二課 圖像與眼睛 (教案)
- 《政治學原理(二)》課程教學大綱
- 石膏板A1級燃燒性能報告
評論
0/150
提交評論