Chapter 12

Multisample inference:
Analysis of Variance

EPI809/Spring 2008 1
 Learning Objectives

1.Describe Analysis of Variance (ANOVA)

2.Explain the Rationale of ANOVA
3.Compare Experimental Designs
4.Test the Equality of 2 or More Means
 Completely Randomized Design
 Randomized Block Design
 Factorial Design

EPI809/Spring 2008 2
 Analysis of Variance
A analysis of variance is a technique that
partitions the total sum of squares of
deviations of the observations about their
mean into portions associated with
independent variables in the experiment
and a portion associated with error

EPI809/Spring 2008 3
 Analysis of Variance
The ANOVA table was previously
discussed in the context of regression
models with quantitative independent
variables, in this chapter the focus will be
on nominal independent variables (factors)

EPI809/Spring 2008 4
Analysis of Variance

A factor refers to a categorical

quantity under examination in an
experiment as a possible cause of
variation in the response variable.

EPI809/Spring 2008 5
 Analysis of Variance

Levels refer to the categories,

measurements, or strata of a factor of
interest in the experiment.

EPI809/Spring 2008 6
Types of Experimental Designs

Experimental
Designs

Completely Randomized Factorial

Randomized Block

One-Way Two-Way
Anova Anova

EPI809/Spring 2008 7
Completely Randomized
Design

EPI809/Spring 2008 8
Completely Randomized Design
1. Experimental Units (Subjects) Are
Assigned Randomly to Treatments
 Subjects are Assumed Homogeneous

2. One Factor or Independent Variable

 2 or More Treatment Levels or groups

3. Analyzed by One-Way ANOVA

EPI809/Spring 2008 9
 One-Way ANOVA F-Test

1. Tests the Equality of 2 or More (p)

Population Means

2. Variables
 One Nominal Independent Variable
 One Continuous Dependent Variable

EPI809/Spring 2008 10
 One-Way ANOVA F-Test
Assumptions
1. Randomness & Independence of Errors
2. Normality
 Populations (for each condition) are
Normally Distributed
3.Homogeneity of Variance
 Populations (for each condition) have Equal
Variances

EPI809/Spring 2008 11
 One-Way ANOVA F-Test
Hypotheses
H0:  1 =  2 =  3 = ... =  p
 All Population Means
are Equal
 No Treatment Effect

Ha: Not All  j Are Equal

 At Least 1 Pop. Mean
is Different
 Treatment Effect
   1   2  ...   p

EPI809/Spring 2008 12
 One-Way ANOVA F-Test
Hypotheses
H0:  1 =  2 =  3 = ... =  p
f(X)
 All Population Means
are Equal
 No Treatment Effect
X
1 = 2 = 3
Ha: Not All  j Are Equal
 At Least 1 Pop. Mean is
Different f(X)
 Treatment Effect
   1 =  2 = ... =  p
X
 Or  i ≠  j for some i, j. 1 = 2 3
EPI809/Spring 2008 13
 One-Way ANOVA
Basic Idea
1. Compares 2 Types of Variation to Test
Equality of Means
2. If Treatment Variation Is Significantly
Greater Than Random Variation then
Means Are Not Equal
3.Variation Measures Are Obtained by
‘Partitioning’ Total Variation

EPI809/Spring 2008 14
 One-Way ANOVA
Partitions Total Variation

EPI809/Spring 2008 15
 One-Way ANOVA
Partitions Total Variation

Total variation

EPI809/Spring 2008 16
 One-Way ANOVA
Partitions Total Variation

Total variation

Variation due to
treatment

EPI809/Spring 2008 17
 One-Way ANOVA
Partitions Total Variation

Total variation

Variation due to Variation due to

treatment random sampling

EPI809/Spring 2008 18
 One-Way ANOVA
Partitions Total Variation

Total variation

Variation due to Variation due to

treatment random sampling

Sum of Squares Among

Sum of Squares Between
Sum of Squares Treatment
Among Groups Variation
EPI809/Spring 2008 19
 One-Way ANOVA
Partitions Total Variation

Total variation

Variation due to Variation due to

treatment random sampling

Sum of Squares Among Sum of Squares Within

Sum of Squares Between Sum of Squares Error
Sum of Squares Treatment (SSE)
(SST) Within Groups Variation
Among Groups Variation
EPI809/Spring 2008 20
 Total Variation
2 2

SS  Total    Y11  Y    Y21  Y     Yij  Y  2

Response, Y

Y

Group 1 Group 2 Group 3

EPI809/Spring 2008 21
 Treatment Variation

SST  n1  Y1  Y   n2  Y2  Y     n p Y p  Y 
2 2 2

Response, Y
 Y3
Y
 Y2
 Y1

Group 1 Group 2 Group 3

EPI809/Spring 2008 22
 Random (Error) Variation

SSE   Y11  Y1   Y21  Y1     Y pj  Y p 

2 2 2

Response, Y

 Y3
 Y2
 Y1

Group 1 Group 2 Group 3

EPI809/Spring 2008 23
 One-Way ANOVA F-Test
Test Statistic
STT /  p  1
 1. Test Statistic
 F = MST / MSE 
SSE /  n  p 
• MST Is Mean Square for Treatment
• MSE Is Mean Square for Error
 2. Degrees of Freedom
 1 = p -1
 2 = n - p
• p = # Populations, Groups, or Levels
• n = Total Sample Size

EPI809/Spring 2008 24
 One-Way ANOVA
Summary Table
Source of Degrees Sum of Mean F
Variation of Squares Square
Freedom (Variance)
Treatment p-1 SST MST = MST
SST/(p - 1) MSE
Error n-p SSE MSE =
SSE/(n - p)
Total n-1 SS(Total) =
SST+SSE

EPI809/Spring 2008 25
One-Way ANOVA F-Test Critical
Value
If means are equal,
F = MST / MSE  1.
Only reject large F! Reject H0

Do Not 
Reject H0

0 F
Fa ( p1, n p)

Always One-Tail!
© 1984-1994 T/Maker Co.

EPI809/Spring 2008 26
 One-Way ANOVA F-Test
Example
As a vet epidemiologist you Food1 Food2 Food3
want to see if 3 food 25.40 23.40 20.00
supplements have different 26.31 21.80 22.20
mean milk yields. You 24.10 23.50 19.75
assign 15 cows, 5 per food 23.74 22.75 20.60
supplement. 25.10 21.60 20.40
Question: At the .05 level, is
there a difference in mean
yields?

EPI809/Spring 2008 27
 One-Way ANOVA F-Test
Solution
 H0:  1 =  2 =  3
Test Statistic:
 Ha: Not All Equal
MST 23.5820
  = .05 F   25.6
MSE .9211
  1 = 2  2 = 12
 Critical Value(s):
Decision:
Reject at  = .05
 = .05
Conclusion:
There Is Evidence Pop.
0 3.89 F Means Are Different
EPI809/Spring 2008 28
 Summary Table
Solution
Source of Degrees of Sum of Mean F
Variation Freedom Squares Square
(Variance)
Food 3-1=2 47.1640 23.5820 25.60

Error 15 - 3 = 12 11.0532 .9211

Total 15 - 1 = 14 58.2172

EPI809/Spring 2008 29
 SAS CODES FOR ANOVA
 Data Anova;
 input group$ milk @@;
 cards;
 food1 25.40 food2 23.40 food3 20.00
 food1 26.31 food2 21.80 food3 22.20
 food1 24.10 food2 23.50 food3 19.75
 food1 23.74 food2 22.75 food3 20.60
 food1 25.10 food2 21.60 food3 20.40
 ;
 run;

 proc anova; /* or PROC GLM */

 class group;
 model milk=group;
 run;

EPI809/Spring 2008 30
 SAS OUTPUT - ANOVA
Sum of
Source DF Squares Mean Square F Value Pr > F

Model 2 47.16400000 23.58200000 25.60 <.0001

Error 12 11.05320000 0.92110000

Corrected Total 14 58.21720000

EPI809/Spring 2008 31
 Pair-wise comparisons
 Needed when the overall F test is rejected

 Can be done without adjustment of type I error if

other comparisons were planned in advance
(least significant difference - LSD method)

 Type I error needs to be adjusted if other

comparisons were not planned in advance
(Bonferroni’s and scheffe’s methods)

EPI809/Spring 2008 32
 Fisher’s Least Significant
Difference (LSD) Test
To compare level 1 and level 2
1 1
t   y1  y2  MSE   
 n1 n2 
Compare this to t/2 = Upper-tailed value or - t/2
lower-tailed from Student’s t-distribution for /2 and
(n - p) degrees of freedom

MSE = Mean square within from ANOVA table

n = Number of subjects
p = Number of levels
EPI809/Spring 2008 33
 Bonferroni’s method
To compare level 1 and level 2
1 1
t   y1  y2  MSE   
 n1 n2 

Adjust the significance level α by

taking the new significance level α*
�p �
  /� �
*

�2 �
EPI809/Spring 2008 34
 SAS CODES FOR multiple
comparisons

proc anova;
class group;
model milk=group;
means group/ lsd bon;
run;
EPI809/Spring 2008 35
 SAS OUTPUT - LSD
t Tests (LSD) for milk

NOTE: This test controls the Type I comparisonwise error rate,

not the experimentwise error rate.

Alpha 0.05
Error Degrees of Freedom 12
Error Mean Square 0.9211
Critical Value of t 2.17881 = t.975,12
Least Significant Difference 1.3225

Means with the same letter are not significantly different.

t Grouping Mean N group

A 24.9300 5 food1
B 22.6100 5 food2
C 20.5900 5 food3

EPI809/Spring 2008 36
 SAS OUTPUT - Bonferroni
Bonferroni (Dunn) t Tests for milk

NOTE: This test controls the Type I experimentwise error rate

Alpha 0.05
Error Degrees of Freedom 12
Error Mean Square 0.9211
Critical Value of t 2.77947=t1-0.05/3/2,12
Minimum Significant Difference 1.6871

Means with the same letter are not significantly different.

Bon Grouping Mean N group

A 24.9300 5 food1
B 22.6100 5 food2
C 20.5900 5 food3

EPI809/Spring 2008 37
Randomized Block
Design

EPI809/Spring 2008 38
 Randomized Block Design

1.Experimental Units (Subjects) Are Assigned

Randomly within Blocks
 Blocks are Assumed Homogeneous
2.One Factor or Independent Variable of
Interest
 2 or More Treatment Levels or Classifications
3. One Blocking Factor

EPI809/Spring 2008 39
 Randomized Block Design
Factor Levels:
(Treatments) A, B, C, D
Treatments are randomly
Experimental Units assigned within blocks
Block 1 A C D B
Block 2 C D B A
Block 3 B A D C
. . . . .
. . . . .
. . . . .

Block b D C A B
EPI809/Spring 2008 40
 Randomized Block F-Test

1.Tests the Equality of 2 or More (p)

Population Means

2.Variables
 One Nominal Independent Variable
 One Nominal Blocking Variable
 One Continuous Dependent Variable

EPI809/Spring 2008 41
 Randomized Block F-Test
Assumptions
1.Normality
 Probability Distribution of each Block-
Treatment combination is Normal

2.Homogeneity of Variance
 Probability Distributions of all Block-
Treatment combinations have Equal
Variances

EPI809/Spring 2008 42
 Randomized Block F-Test
Hypotheses
 H0:  1 =  2 =  3 = ... =  p
 All Population Means are
Equal
 No Treatment Effect

 Ha: Not All  j Are Equal

 At Least 1 Pop. Mean is
Different
 Treatment Effect
  1   2  ...   p Is wrong

EPI809/Spring 2008 43
 Randomized Block F-Test
Hypotheses
H0:  1 =  2 = ... =  p
 All Population Means f(X)
are Equal
 No Treatment Effect
X
Ha: Not All  j Are Equal 1 = 2 = 3
 At Least 1 Pop. Mean
is Different f(X)
 Treatment Effect
  1   2  ...   p Is
wrong X
1 = 2 3
EPI809/Spring 2008 44
 The F Ratio for Randomized
Block Designs
 SS=SSE+SSB+SST

MST SST /  p  1
F 
MSE SSE /  n  1  p  1  b  1
SST /  p  1

SSE /  n  p  b  1
 Randomized Block F-Test
Test Statistic
 1. Test Statistic
 F = MST / MSE
• MST Is Mean Square for Treatment
• MSE Is Mean Square for Error

 2. Degrees of Freedom
 1 = p -1
 2 = n – b – p +1
• p = # Treatments, b = # Blocks, n = Total Sample Size

EPI809/Spring 2008 46
 Randomized Block F-Test
Critical Value
If means are equal,
F = MST / MSE  1.
Only reject large F! Reject H0

Do Not 
Reject H0

0 F
Fa ( p1, n p)

Always One-Tail!
© 1984-1994 T/Maker Co.

EPI809/Spring 2008 47
 Randomized Block F-Test
Example
 You wish to determine which of four brands of tires has
the longest tread life. You randomly assign one of each
brand (A, B, C, and D) to a tire location on each of 5
cars. At the .05 level, is there a difference in mean tread
life? Tire Location
Block Left Front Right Front Left Rear Right Rear
Car 1 A: 42,000 C: 58,000 B: 38,000 D: 44,000
Car 2 B: 40,000 D: 48,000 A: 39,000 C: 50,000
Car 3 C: 48,000 D: 39,000 B: 36,000 A: 39,000
Car 4 A: 41,000 B: 38,000 D: 42,000 C: 43,000
Car 5 D: 51,000 A: 44,000 C: 52,000 B: 35,000
EPI809/Spring 2008 48
 Randomized Block F-Test
Solution
 H0:  1 =  2 =  3=  4
Test Statistic:
 Ha: Not All Equal
  = .05 F = 11.9933
  1 = 3  2 = 12
Decision:
 Critical Value(s): Reject at  = .05
Conclusion:
 = .05
There Is Evidence Pop.
Means Are Different
0 3.49 F
EPI809/Spring 2008 49
 SAS CODES FOR ANOVA
data block;
input Block$ trt$ resp @@;
cards;
Car1 A: 42000 Car1 C: 58000 Car1 B: 38000 Car1 D: 44000
Car2 B: 40000 Car2 D: 48000 Car2 A: 39000 Car2 C: 50000
Car3 C: 48000 Car3 D: 39000 Car3 B: 36000 Car3 A: 39000
Car4 A: 41000 Car4 B: 38000 Car4 D: 42000 Car4 C: 43000
Car5 D: 51000 Car5 A: 44000 Car5 C: 52000 Car5 B: 35000
;
run;

proc anova;
class trt block;
model resp=trt block;
Means trt /lsd bon;
run;

EPI809/Spring 2008 50
 SAS OUTPUT - ANOVA
Dependent Variable: resp

Sum of
Source DF Squares Mean Square F Value Pr > F

Model 7 544550000.0 77792857.1 6.22 0.0030

Error 12 150000000.0 12500000.0
Corrected Total 19 694550000.0

R-Square Coeff Var Root MSE resp Mean

0.784033 8.155788 3535.534 43350.00

Source DF Anova SS Mean Square F Value Pr > F

trt 3 449750000.0 149916666.7 11.99 0.0006

Block 4 94800000.0 23700000.0 1.90 0.1759

EPI809/Spring 2008 51
 SAS OUTPUT - LSD

Means with the same letter are not significantly different.

t Grouping Mean N trt

A 50200 5 C:

B 44800 5 D:
B
C B 41000 5 A:
C
C 37400 5 B:
EPI809/Spring 2008 52
 SAS OUTPUT - Bonferroni
Means with the same letter are not significantly different.

Bon Grouping Mean N trt

A 50200 5 C:
A
B A 44800 5 D:
B
B C 41000 5 A:
C
C 37400 5 B:

EPI809/Spring 2008 53
Factorial Experiments

EPI809/Spring 2008 54
 Factorial Design
 1.
Experimental Units (Subjects) Are
Assigned Randomly to Treatments
 Subjects are Assumed Homogeneous
 2.Two or More Factors or Independent
Variables
 Each Has 2 or More Treatments (Levels)
 3. Analyzed by Two-Way ANOVA

EPI809/Spring 2008 55
 Advantages
of Factorial Designs
1.Saves Time & Effort
 e.g., Could Use Separate Completely
Randomized Designs for Each Variable
2.Controls Confounding Effects by Putting
Other Variables into Model
3.Can Explore Interaction Between Variables

EPI809/Spring 2008 56
 Two-Way ANOVA

1. Tests the Equality of 2 or More

Population Means When Several
Independent Variables Are Used

2. Same Results as Separate One-Way

ANOVA on Each Variable
- But Interaction Can Be Tested

EPI809/Spring 2008 57
 Two-Way ANOVA
Assumptions
1.Normality
 Populations are Normally Distributed
2.Homogeneity of Variance
 Populations have Equal Variances
3.Independence of Errors
 Independent Random Samples are Drawn

EPI809/Spring 2008 58
 Two-Way ANOVA
Data Table
Factor Factor B
A 1 2 ... b Observation k
1 Y 111 Y121 ... Y1b1
Y112 Y122 ... Y1b2 Yijk
2 Y211 Y221 ... Y2b1
Y212 Y222 ... YX
2b2
2b2
Level i Level j
: : : : : Factor Factor
a Ya11 Ya21 ... Yab1 A B

Ya12 Ya22 ... Yab2

EPI809/Spring 2008 59
 Two-Way ANOVA
Null Hypotheses
1.No Difference in Means Due to Factor A
 H0: 1. = 2. =... = a.
2.No Difference in Means Due to Factor B
 H0: .1 = .2 =... = .b
3.No Interaction of Factors A & B
 H0: ABij = 0

EPI809/Spring 2008 60
 Two-Way ANOVA
Total Variation Partitioning
Total
Total Variation
Variation
SS(Total)

Variation
VariationDue
Dueto
to Variation
Variation Due
Dueto
to
Treatment
TreatmentAA Treatment
TreatmentBB
 SSA SSB

Variation
VariationDue
Duetoto Variation
VariationDue
Dueto
to
Interaction
Interaction Random
Random Sampling
Sampling
SS(AB)  SSE
EPI809/Spring 2008 61
 Two-Way ANOVA
Summary Table
Source of Degrees of Sum of Mean F
Variation Freedom Squares Square
A a-1 SS(A) MS(A) MS(A)
(Row) MSE
B b-1 SS(B) MS(B) MS(B)
(Column) MSE
AB (a-1)(b-1) SS(AB) MS(AB) MS(AB)
(Interaction) MSE
Error n - ab SSE MSE
Total n-1 SS(Total) Same as
Other
EPI809/Spring 2008 Designs 62
 Interaction
1.Occurs When Effects of One Factor Vary
According to Levels of Other Factor
2.When Significant, Interpretation of Main
Effects (A & B) Is Complicated
3.Can Be Detected
 In Data Table, Pattern of Cell Means in One
Row Differs From Another Row
 In Graph of Cell Means, Lines Cross

EPI809/Spring 2008 63
 Graphs of Interaction

Effects of Gender (male or female) & dietary

group (sv, lv, nor) on systolic blood pressure
Interaction No Interaction
Average Average
Response male Response male

female female

sv lv nor sv lv nor

EPI809/Spring 2008 64
 Two-Way ANOVA F-Test
Example
Effect of diet (sv-strict vegetarians, lv-
lactovegetarians, nor-normal) and gender (female,
male) on systolic blood pressure.

Question: Test for interaction and main effects at

the .05 level.

EPI809/Spring 2008 65
 SAS CODES FOR ANOVA
data factorial;
input dietary$ sex$ sbp; lv male 116.5
cards; lv male 118.5
lv male 119.5
sv male 109.9 lv male 110.5
sv male 101.9 lv male 115.5
sv male 100.9 lv male 105.2
sv male 119.9 nor male 128.3
sv male 104.9 nor male 129.3
sv male 189.9 nor male 126.3
nor male 127.3
nor male 126.3
sv female 102.6 nor male 125.3
sv female 99
nor female 119.1
sv female 83 .6 nor female 119.2
sv female 99.6 nor female 115.6
sv female 102.6 nor female 119.9
nor female 119.8
sv female 112.6 nor female 119.7
;
run;
EPI809/Spring 2008 66
 SAS CODES FOR ANOVA

proc glm;
class dietary sex;
model sbp=dietary sex dietary*sex;
run;

proc glm;
class dietary sex;
model sbp=dietary sex;
run;

EPI809/Spring 2008 67
 SAS OUTPUT - ANOVA
Dependent Variable: sbp

Sum of
Source DF Squares Mean Square F Value Pr > F
Model 5 2627.399667 525.479933 1.96 0.1215
Error 24 6435.215000 268.133958
Corrected Total 29 9062.614667

R-Square Coeff Var Root MSE sbp Mean

0.289916 14.08140 16.37480 116.2867

Source DF Type I SS Mean Square F Value Pr > F

dietary 2 958.870500 479.435250 1.79 0.1889
sex 1 1400.686992 1400.686992 5.22 0.0314
dietary*sex 2 267.842175 133.921087 0.50 0.6130

Source DF Type III SS Mean Square F Value Pr > F

dietary 2 1039.020874 519.510437 1.94 0.1659
sex 1 877.982292 877.982292 3.27 0.0829
dietary*sex 2 267.842175 133.921087 0.50 0.6130

EPI809/Spring 2008 68
 Linear Contrast
 Linear Contrast is a linear combination of the means
of populations

 Purpose: to test relationship among different group

means

L  �c j  j with �c j 0

Example: 4 populations on treatments T1, T2, T3 and T4.

Contrast T1 T2 T3 T4 relation to test
L1 1 0 -1 0 μ1 - μ3 = 0
L2 1 -1/2 -1/2 0 μ1 – μ2/2 – μ3/2 = 0
EPI809/Spring 2008 69
T-test for Linear Contrast (LSD)
 Construct a t statistic involving k group means.
Degrees of freedom of t - test: df = n-k.

k L
To test H0: L  �c j  j  0 Construct t
2
j 1
k c
s 2
�n
j 1
j

j
Compare with critical value t1-α/2,, n-k.
Reject H0 if |t| ≥ t1-α/2,, n-k.
SAS uses contrast statement and performs an F – test df (1, n-k);
Or estimate statement and perform a t-test df (n-k).
EPI809/Spring 2008 70
T-test for Linear Contrast (Scheffe)
 Construct multiple contrasts involving k group
means. Trying to search for significant contrast

k L
To test H0: L  �c j  j  0 Construct t
2
j 1
k c
Compare with critical value.
s 2
�n
j 1
j

j
a  (k  1) Fk 1,n  k ,1

Reject H0 if |t| ≥ a

EPI809/Spring 2008 71
 SAS Code for contrast testing
 proc glm;
 class trt block;
 model resp=trt block;
 Means trt /lsd bon scheffe;
 contrast 'A - B = 0' trt 1 -1 0 0 ;
 contrast 'A - B/2 - C/2 = 0' trt 1 -.5 -.5 0 ;
 contrast 'A - B/3 - C/3 -D/3 = 0' trt 3 -1 -1 -1 ;
 contrast 'A + B - C - D = 0' trt 1 1 -1 -1 ;
 lsmeans trt/stderr pdiff;
 lsmeans trt/stderr pdiff adjust=scheffe; /* Scheffe's test */
 lsmeans trt/stderr pdiff adjust=bon; /* Boneferoni's test
*/
 estimate ‘A - B' trt 1 -1 0 0 0;

 run;

EPI809/Spring 2008 72
Regression representation of Anova

EPI809/Spring 2008 73
 Regression representation of
Anova
yij  i  eij     i  eij
 One-way anova: p

�
i 1
i 0

 Two-way anova:
yijk  ij  eijk     i  b j  g ij  eijk
a b b a

�
i 1
i  0, �b j  0,
j 1
�g
j 1
ij  0 for all i and �g
i 1
ij  0 for all j

 SAS uses a different constraint

EPI809/Spring 2008 74
 Regression representation of
Anova
 One-way anova: Dummy variables of factor
with p levels

y  b 0  b1 x1  b 2 x2  ...  b p 1 x p 1  e
1
� if level i
where xi  �
0
� if otherwise

 This is the parameterization used by SAS

EPI809/Spring 2008 75
 Conclusion: should be able to
1. Recognize the applications that uses ANOVA

2. Understand the logic of analysis of variance.

3. Be aware of several different analysis of

variance designs and understand when to use
each one.

4. Perform a single factor hypothesis test using

analysis of variance manually and with the aid of
SAS or any statistical software.
EPI809/Spring 2008 76
Conclusion: should be able to
5. Conduct and interpret post-analysis of
variance pairwise comparisons procedures.

6. Recognize when randomized block

analysis of variance is useful and be able to
perform the randomized block analysis.

7. Perform two factor analysis of variance

tests with replications using SAS and
interpret the output.
EPI809/Spring 2008 77
 Key Terms

 Between-Sample  Levels
Variation  One-Way Analysis
 Completely of Variance
Randomized Design  Total Variation
 Experiment-Wide
 Treatment
Error Rate
 Within-Sample
 Factor
Variation

EPI809/Spring 2008 78