Analysis of Variance

Chap 11-1
 Learning Objectives
In this chapter, you learn:
 The basic concepts of experimental design
 How to use one-way analysis of variance to test for differences
among the means of several groups
 How to use two-way analysis of variance and interpret the
interaction effect
 How to perform multiple comparisons in a one-way analysis of
variance and a two-way analysis of variance

Chap 11-2
 Chapter Overview
DCOVA
Analysis of Variance (ANOVA)

One-Way Two-Way
ANOVA ANOVA
F-test
Interaction
Tukey- Effects
Kramer
Multiple Tukey Multiple
Comparisons Comparisons
Levene Test
For
Homogeneity
of Variance
Chap 11-3
 General ANOVA Setting
DCOVA

 Investigator controls one or more factors of interest

 Each factor contains two or more levels

 Levels can be numerical or categorical

 Different levels produce different groups

 Think of each group as a sample from a different

population
 Observe effects on the dependent variable
 Are the groups the same?

 Experimental design: the plan used to collect the data

Chap 11-4
 Completely Randomized Design
DCOVA

 Experimental units (subjects) are assigned

randomly to groups
 Subjects are assumed homogeneous
 Only one factor or independent variable
 With two or more levels
 Analyzed by one-factor analysis of variance
(ANOVA)

Chap 11-5
 One-Way Analysis of Variance
DCOVA

 Evaluate the difference among the means of three

or more groups
Examples: Number of accidents for 1st, 2nd, and 3rd shift
Expected mileage for five brands of tires

 Assumptions
 Populations are normally distributed

 Populations have equal variances

 Samples are randomly and independently drawn

Chap 11-6
 Hypotheses of One-Way ANOVA
DCOVA

 H0 : μ1  μ2  μ3    μc
 All population means are equal
 i.e., no factor effect (no variation in means among
groups)

 H1 : Not all of the population means are equal

 At least one population mean is different
 i.e., there is a factor effect
 Does not mean that all population means are
different (some pairs may be the same)

Chap 11-7
 One-Way ANOVA
DCOVA
H0 : μ1  μ2  μ3    μc
H1 : Not all μ j are equal
The Null Hypothesis is True
All Means are the same:
(No Factor Effect)

μ1  μ 2  μ 3
Chap 11-8
 One-Way ANOVA DCOVA
(continued)
H0 : μ1  μ2  μ3    μc
H1 : Not all μ j are equal
The Null Hypothesis is NOT true
At least one of the means is different
(Factor Effect is present)

or

μ1  μ2  μ3 μ1  μ2  μ3
Chap 11-9
 Partitioning the Variation
DCOVA

 Total variation can be split into two parts:

SST = SSA + SSW

SST = Total Sum of Squares

(Total variation)
SSA = Sum of Squares Among Groups
(Among-group variation)
SSW = Sum of Squares Within Groups
(Within-group variation)

Chap 11-10
 Partitioning the Variation
(continued)

DCOVA
SST = SSA + SSW

Total Variation = the aggregate variation of the individual

data values across the various factor levels (SST)

Among-Group Variation = variation among the factor

sample means (SSA)

Within-Group Variation = variation that exists among

the data values within a particular factor level (SSW)

Chap 11-11
 Partition of Total Variation
DCOVA
Total Variation (SST)

Variation Due to Variation Due to Random

= Factor (SSA) + Error (SSW)

Chap 11-12
 Total Sum of Squares
DCOVA

SST = SSA + SSW

c nj

SST   ( Xij  X) 2

j1 i1
Where:
SST = Total sum of squares
c = number of groups or levels
nj = number of observations in group j
Xij = ith observation from group j
X = grand mean (mean of all data values)
Chap 11-13
 Total Variation DCOVA
(continued)

2 2 2
SST  ( X 11  X )  ( X 12  X )      ( X cn  X )
c

Response, X

Group 1 Group 2 Group 3

Chap 11-14
 Among-Group Variation
DCOVA

SST = SSA + SSW

c
SSA   n j ( X j  X) 2

j1
Where:
SSA = Sum of squares among groups
c = number of groups
nj = sample size from group j
Xj = sample mean from group j
X = grand mean (mean of all data values)
Chap 11-15
 Among-Group Variation
(continued)
c
DCOVA
SSA   n j ( X j  X) 2

j1

SSA
Variation Due to
MSA 
Differences Among Groups
c 1
Mean Square Among =
SSA/degrees of freedom

i j

Chap 11-16
 Among-Group Variation
DCOVA
(continued)

SSA  n1 (X1  X)  n 2 (X 2  X)      n c (X c  X)
2 2 2

Response, X

X3
X2 X
X1

Group 1 Group 2 Group 3

Chap 11-17
 Within-Group Variation
DCOVA

SST = SSA + SSW

c nj
SSW    ( Xij  X j ) 2

j1 i1
Where:
SSW = Sum of squares within groups
c = number of groups
nj = sample size from group j
Xj = sample mean from group j
Xij = ith observation in group j
Chap 11-18
 Within-Group Variation
(continued)
DCOVA
c nj
SSW    ( Xij  X j ) 2

j1 i1
SSW
Summing the variation
MSW 
within each group and then
adding over all groups nc
Mean Square Within =
SSW/degrees of freedom

μj
Chap 11-19
 Within-Group Variation
DCOVA
(continued)

SSW  (X11  X1 )  (X12  X 2 )      (X cnc  X c )

2 2 2

Response, X

X3
X2
X1

Group 1 Group 2 Group 3

Chap 11-20
 Obtaining the Mean Squares
DCOVA
The Mean Squares are obtained by dividing the various
sum of squares by their associated degrees of freedom

SSA Mean Square Among

MSA  (d.f. = c-1)
c 1
SSW
MSW  Mean Square Within
nc (d.f. = n-c)

SST
MST  Mean Square Total
n 1 (d.f. = n-1)
Chap 11-21
 One-Way ANOVA Table
DCOVA

Source of Degrees of Sum Of Mean Square F

Variation Freedom Squares (Variance)

Among SSA FSTAT =

c-1 SSA MSA =
Groups c-1
MSA
Within SSW
n-c SSW MSW = MSW
Groups n-c

Total n–1 SST

c = number of groups
n = sum of the sample sizes from all groups
df = degrees of freedom
Chap 11-22
 One-Way ANOVA
F Test Statistic DCOVA

H0: μ1= μ2 = … = μc
H1: At least two population means are different

 Test statistic MSA

FSTAT 
MSW
MSA is mean squares among groups
MSW is mean squares within groups
 Degrees of freedom
 df1 = c – 1 (c = number of groups)
 df2 = n – c (n = sum of sample sizes from all populations)

Chap 11-23
 Interpreting One-Way ANOVA
F Statistic DCOVA

 The F statistic is the ratio of the among

estimate of variance and the within estimate
of variance
 The ratio must always be positive
 df1 = c -1 will typically be small
 df2 = n - c will typically be large

Decision Rule:
 Reject H0 if FSTAT > Fα, 
otherwise do not reject
H0 0 Do not Reject H0
reject H0
Fα
Chap 11-24
 One-Way ANOVA
F Test Example DCOVA

You want to see if three Club 1 Club 2 Club 3

different golf clubs yield 254 234 200
different distances. You 263 218 222
randomly select five 241 235 197
measurements from trials on 237 227 206
an automated driving 251 216 204
machine for each club. At the
0.05 significance level, is
there a difference in mean
distance?

Chap 11-25
 One-Way ANOVA Example:
Scatter Plot DCOVA
Distance
Club 1 Club 2 Club 3 270
254 234 200 260 •
263 218 222 ••
241 235 197
250 X1
240 •
237 227 206 • ••
251 216 204 230
• X
220
••
X2 •
210
x1  249.2 x 2  226.0 x 3  205.8
•• X3
200 ••
x  227.0 190

1 2 3
ClubChap 11-26
 One-Way ANOVA Example
Computations DCOVA

Club 1 Club 2 Club 3 X1 = 249.2 n1 = 5

254 234 200 X2 = 226.0 n2 = 5
263 218 222
X3 = 205.8 n3 = 5
241 235 197
237 227 206 n = 15
X = 227.0
251 216 204 c=3
SSA = 5 (249.2 – 227)2 + 5 (226 – 227)2 + 5 (205.8 – 227)2 = 4716.4
SSW = (254 – 249.2)2 + (263 – 249.2)2 +…+ (204 – 205.8)2 = 1119.6

MSA = 4716.4 / (3-1) = 2358.2 2358.2

FSTAT   25.275
MSW = 1119.6 / (15-3) = 93.3 93.3
Chap 11-27
 One-Way ANOVA Example
Solution DCOVA

H0: μ1 = μ2 = μ3 Test Statistic:

H1: μj not all equal
 = 0.05
MSA 2358.2
FSTAT    25.275
df1= 2 df2 = 12 MSW 93.3

Critical Decision:
Value:
Reject H0 at  = 0.05
Fα = 3.89
 = .05 Conclusion:
There is evidence that
0 Do not Reject H
0
at least one μj differs
reject H0
FSTAT = 25.275 from the rest
Fα = 3.89
Chap 11-28
 One-Way ANOVA
Excel Output DCOVA

SUMMARY
Groups Count Sum Average Variance
Club 1 5 1246 249.2 108.2
Club 2 5 1130 226 77.5
Club 3 5 1029 205.8 94.2
ANOVA
Source of
SS df MS F P-value F crit
Variation
Between
4716.4 2 2358.2 25.275 0.0000 3.89
Groups
Within
1119.6 12 93.3
Groups
Total 5836.0 14

Chap 11-29
 The Tukey-Kramer Procedure
DCOVA

 Tells which population means are significantly

different
 e.g.: μ1 = μ2  μ3
 Done after rejection of equal means in ANOVA
 Allows paired comparisons
 Compare absolute mean differences with critical
range

μ1= μ2 μ3 x

Chap 11-30
 Tukey-Kramer Critical Range
DCOVA

MSW  1 1 
Critical Range  Q α 
2  n j n j' 

where:
Qα = Upper Tail Critical Value from Studentized
Range Distribution with c and n - c degrees
of freedom (see appendix E.7 table)
MSW = Mean Square Within
nj and nj’ = Sample sizes from groups j and j’

Chap 11-31
 The Tukey-Kramer Procedure:
Example DCOVA
1. Compute absolute mean
Club 1 Club 2 Club 3 differences:
254 234 200
263 218 222 x1  x 2  249.2  226.0  23.2
241 235 197 x1  x 3  249.2  205.8  43.4
237 227 206
251 216 204 x 2  x 3  226.0  205.8  20.2

2. Find the Qα value from the table in appendix E.7 with

c = 3 and (n – c) = (15 – 3) = 12 degrees of freedom:

Q α  3.77
Chap 11-32
 The Tukey-Kramer Procedure:
Example
(continued)
3. Compute Critical Range: DCOVA

MSW  1 1  93.3  1 1 
Critical Range  Q α   3.77     16.285
 
2  n j n j'  2 5 5

4. Compare:
5. All of the absolute mean differences
x1  x 2  23.2
are greater than critical range.
Therefore there is a significant x1  x 3  43.4
difference between each pair of
means at 5% level of significance. x 2  x 3  20.2
Thus, with 95% confidence we can conclude
that the mean distance for club 1 is greater
than club 2 and 3, and club 2 is greater than
club 3.
Chap 11-33
 ANOVA Assumptions
DCOVA

 Randomness and Independence

 Select random samples from the c groups (or
randomly assign the levels)
 Normality
 The sample values for each group are from a normal
population
 Homogeneity of Variance
 All populations sampled from have the same
variance
 Can be tested with Levene’s Test
Chap 11-34
 ANOVA Assumptions
Levene’s Test
DCOVA

 Tests the assumption that the variances of each

population are equal.
 First, define the null and alternative hypotheses:
 H0: σ21 = σ22 = …=σ2c
 H1: Not all σ2j are equal
 Second, compute the absolute value of the difference
between each value and the median of each group.
 Third, perform a one-way ANOVA on these absolute
differences.

Chap 11-35
 Levene Homogeneity Of Variance
Test Example
DCOVA

H0: σ21 = σ22 = σ23

H1: Not all σ2j are equal

Calculate Medians Calculate Absolute Differences

Club 1 Club 2 Club 3 Club 1 Club 2 Club 3

237 216 197 14 11 7
241 218 200 10 9 4
251 227 204 Median 0 0 0
254 234 206 3 7 2
263 235 222 12 8 18

Chap 11-36
 Levene Homogeneity Of Variance
Test Example (continued)
Anova: Single Factor DCOVA
SUMMARY
Groups Count Sum Average Variance
Since the
Club 1 5 39 7.8 36.2 p-value is
Club 2 5 35 7 17.5 greater
Club 3 5 31 6.2 50.2 than 0.05
there is
P- insufficient
Source of Variation SS df MS F value F crit
evidence
Between Groups 6.4 2 3.2 0.092 0.912 3.885
of a
Within Groups 415.6 12 34.6 difference
in the
Total 422 14 variances

Chap 11-37
 Factorial Design:
Two-Way ANOVA DCOVA

 Examines the effect of

 Two factors of interest on the dependent
variable
 e.g., Percent carbonation and line speed on soft drink
bottling process
 Interaction between the different levels of these
two factors
 e.g., Does the effect of one particular carbonation
level depend on which level the line speed is set?

Chap 11-38
 Two-Way ANOVA
(continued)
DCOVA
 Assumptions

 Populations are normally distributed

 Populations have equal variances
 Independent random samples are
drawn

Chap 11-39
 Two-Way ANOVA
Sources of Variation DCOVA

Two Factors of interest: A and B

r = number of levels of factor A
c = number of levels of factor B
n’ = number of replications for each cell
n = total number of observations in all cells
n = (r)(c)(n’)
Xijk = value of the kth observation of level i of
factor A and level j of factor B
Chap 11-40
 Two-Way ANOVA DCOVA
Sources of Variation (continued)

SST = SSA + SSB + SSAB + SSE Degrees of

Freedom:
SSA r–1
Factor A Variation

SST SSB c–1

Factor B Variation
Total Variation
SSAB
Variation due to interaction (r – 1)(c – 1)
between A and B
n-1
SSE rc(n’ – 1)
Random variation (Error)

Chap 11-41
 Two-Way ANOVA Equations
DCOVA

Total Variation: r c n
SST   ( Xijk  X) 2

i1 j1 k 1

Factor A Variation: r
SSA  cn  ( Xi..  X)
 2

i1

Factor B Variation:
c
SSB  rn  ( X. j.  X)
 2

j1

Chap 11-42
 Two-Way ANOVA Equations
(continued)
DCOVA

Interaction Variation:
r c
SSAB  n ( Xij.  Xi..  X.j.  X)2
i1 j1

Sum of Squares Error:

r c n
SSE   ( Xijk  Xij. )2
i1 j1 k 1

Chap 11-43
 Two-Way ANOVA Equations
(continued)
r c n

where:  X
i1 j1 k 1
ijk DCOVA
X  Grand Mean
c n
rcn
 X
j1 k 1
ijk

Xi..   Mean of ith level of factor A (i  1, 2, ..., r)

cn
r n

 X ijk
X. j.  i1 k 1
 Mean of jth level of factor B (j  1, 2, ..., c)
rn
n
Xijk
Xij.  
r = number of levels of factor A
 Mean of cell ij
k 1 n
c = number of levels of factor B
n’ = number of replications in each cell
Chap 11-44
 Mean Square Calculations
DCOVA
SSA
MSA  Mean square factor A 
r 1

SSB
MSB  Mean square factor B 
c 1

SSAB
MSAB  Mean square interactio n 
(r  1)(c  1)

SSE
MSE  Mean square error 
rc(n'1)
Chap 11-45
 Two-Way ANOVA:
The F Test Statistics DCOVA

F Test for Factor A Effect

H0: μ1..= μ2.. = μ3..= • • = µr..
MSA Reject H0 if
H1: Not all μi.. are equal FSTAT 
MSE FSTAT > Fα

F Test for Factor B Effect

H0: μ.1. = μ.2. = μ.3.= • • = µ.c.
MSB Reject H0 if
H1: Not all μ.j. are equal FSTAT 
MSE FSTAT > Fα

F Test for Interaction Effect

H0: the interaction of A and B is
equal to zero
MSAB
H1: interaction of A and B is not FSTAT  Reject H0 if
MSE FSTAT > Fα
zero
Chap 11-46
 Two-Way ANOVA
Summary Table DCOVA

Source of Sum of Degrees of Mean

F
Variation Squares Freedom Squares

MSA MSA
Factor A SSA r–1
= SSA /(r – 1) MSE
MSB MSB
Factor B SSB c–1
= SSB /(c – 1) MSE

AB MSAB MSAB
SSAB (r – 1)(c – 1)
(Interaction) = SSAB / (r – 1)(c – 1) MSE

MSE =
Error SSE rc(n’ – 1)
SSE/rc(n’ – 1)
Total SST n–1

Chap 11-47
 Features of Two-Way ANOVA
F Test DCOVA

 Degrees of freedom always add up

 n-1 = rc(n’-1) + (r-1) + (c-1) + (r-1)(c-1)
 Total = error + factor A + factor B + interaction

 The denominators of the F Test are always the

same but the numerators are different
 The sums of squares always add up
 SST = SSE + SSA + SSB + SSAB
 Total = error + factor A + factor B + interaction

Chap 11-48
 Examples:
Interaction vs. No Interaction
DCOVA
 Interaction is present:
 No interaction: line some line segments
segments are parallel not parallel

Factor B Level 1
Mean Response

Mean Response
Factor B Level 1
Factor B Level 3

Factor B Level 2
Factor B Level 2
Factor B Level 3

Factor A Levels Factor A Levels

Chap 11-49
 Do ACT Prep Course Type & Length
Impact Average ACT Scores DCOVA
ACT Scores for Different Types and Lengths of Courses

LENGTH OF COURSE

TYPE OF COURSE Condensed Regular

Traditional 26 18 34 28

Traditional 27 24 24 21

Traditional 25 19 35 23

Traditional 21 20 31 29

Traditional 21 18 28 26

Online 27 21 24 21

Online 29 32 16 19

Online 30 20 22 19

Online 24 28 20 24

Online 30 29 23 25
Chap 11-50
Plotting Cell Means Shows A
Strong Interaction DCOVA

Nonparallel lines indicate

the effect of condensing
the course depends on
whether the course is
taught in the traditional
classroom or by online
distance learning

The online course yields

higher scores when
condensed while the
traditional course yields
higher scores when not
condensed (regular).

Chap 11-51
Excel Analysis Of ACT Prep
Course Data DCOVA

The interaction between course

length & type is significant
because its p-value is 0.0000.

While the p-values associated

with both course length &
course type are not significant,
because the interaction is
significant you cannot directly
conclude they have no effect.

Chap 11-52
 With The Significant Interaction
Collapse The Data Into Four Groups
DCOVA

 After collapsing into four groups do a one way

ANOVA

 The four groups are

1. Traditional course condensed
2. Traditional course regular length
3. Online course condensed
4. Online course regular length

Chap 11-53
 Excel Analysis Of Collapsed
Data
DCOVA

1. Traditional regular > Traditional condensed

2. Online condensed > Traditional condensed
Group is a significant effect. 3. Traditional regular > Online regular
4. Online condensed > Online regular
p-value of 0.0003 < 0.05
If the course is take online should use the
condensed version and if the course is taken
by traditional method should use the regular.
Chap 11-54
 The Randomized Block Design
Is Often Useful

 The randomized block design is an on-line topic

Chap 11-55
 Chapter Summary
In this chapter we discussed
 The one-way analysis of variance
 The logic of ANOVA
 ANOVA assumptions
 F test for difference in c means
 The Tukey-Kramer procedure for multiple comparisons
 The Levene test for homogeneity of variance
 The two-way analysis of variance
 Examined effects of multiple factors
 Examined interaction between factors

Chap 11-56
Statistics for Managers Using
Microsoft Excel
7th Edition

Online Topic

The Randomized Block Design

RBD - 1
 Learning Objective

 To learn the basic structure and use of a randomized block design

RBD - 2
 The Randomized Block Design
DCOVA

 Like One-Way ANOVA, we test for equal

population means (for different factor levels, for
example)...

 ...but we want to control for possible variation

from a second factor (with two or more levels)

 Levels of the secondary factor are called blocks

RBD - 3
 Partitioning the Variation
DCOVA

 Total variation can now be split into three parts:

SST = SSA + SSBL + SSE

SST = Total variation

SSA = Among-Group variation
SSBL = Among-Block variation
SSE = Random variation

RBD - 4
 Sum of Squares for Blocks
DCOVA

SST = SSA + SSBL + SSE

r
SSBL  c  ( Xi.  X) 2

i1
Where:
c = number of groups
r = number of blocks
Xi. = mean of all values in block i
X = grand mean (mean of all data values)
RBD - 5
 Partitioning the Variation
DCOVA

 Total variation can now be split into three parts:

SST = SSA + SSBL + SSE

SST and SSA are SSE = SST – (SSA + SSBL)

computed as they were
in One-Way ANOVA

RBD - 6
 Mean Squares
DCOVA

SSBL
MSBL  Mean square blocking 
r 1

SSA
MSA  Mean square among groups 
c 1

SSE
MSE  Mean square error 
(r  1)(c  1)

RBD - 7
 Randomized Block ANOVA Table
DCOVA
Source of SS df MS F
Variation
Among MSBL
Blocks SSBL r-1 MSBL
MSE
Among SSA c-1 MSA MSA
Groups
MSE
Error SSE (r–1)(c-1) MSE

Total SST rc - 1
c = number of populations rc = total number of observations
r = number of blocks df = degrees of freedom
RBD - 8
 Testing For Factor Effect
DCOVA
H 0 : μ .1  μ .2  μ .3      μ .c
H1 : Not all population means are equal

MSA
FSTAT =
MSE  Main Factor test: df1 = c – 1
df2 = (r – 1)(c – 1)

Reject H0 if FSTAT > Fα

RBD - 9
 Test For Block Effect
DCOVA

H 0 : μ1.  μ2.  μ3.  ...  μr.

H1 : Not all block means are equal

MSBL
FSTAT =
MSE  Blocking test: df1 = r – 1
df2 = (r – 1)(c – 1)

Reject H0 if FSTAT > Fα

RBD - 10
 Topic Summary

In this topic we discussed

 The basic structure and use of a randomized block

design

RBD - 11