Completely

Randomized Design
 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 Classifications
3. Analyzed by One-Way ANOVA
 Cage positions
Completely randomized design
(3 treatments x 3 replications)

A
 Cage positions
Completely randomized design
(3 treatments x 3 replications)

A B
 Cage positions
Completely randomized design
(3 treatments x 3 replications)

A B B
 Cage positions
Completely randomized design
(3 treatments x 3 replications)

A B B
 Cage positions
Completely randomized design
(3 treatments x 3 replications)

A B B
C
 Cage positions
Completely randomized design
(3 treatments x 3 replications)

A B B
C A
 Cage positions
Completely randomized design
(3 treatments x 3 replications)

A B B
C A C
 Cage positions
Completely randomized design
(3 treatments x 3 replications)

A B B
C A C
B
 Cage positions
Completely randomized design
(3 treatments x 3 replications)

A B B
C A C
B C
 Cage positions
Completely randomized design
(3 treatments x 3 replications)

A B B
C A C
B C A
 The Linier Model
X =μ + t +ε
ij i ij

i = 1,2,…, t j = 1,2,…, r
Xij = the observation in ith treatment and the jth
replication
m = overall mean
t = the effect of the ith treatment
i

eij = random error

 One-Way ANOVA F-Test

1. Tests the Equality of 2 or More (t) population means

2. Variables
• One Nominal Scaled Independent Variable
 2 or More (t) Treatment Levels or Classifications

• One Interval or Ratio Scaled Dependent Variable

3. Used to analyze Completely Randomized
Experimental Designs
 Assumptions

1. Randomness & Independence of Errors

• Independent Random Samples are Drawn for each
condition
2. Normality
• Populations (for each condition) are Normally Distributed
3. Homogeneity of Variance
• Populations (for each condition) have Equal Variances
 Hypotheses

H0: m1 = m2 = m3 = ... = mt


• All Population Means are Equal

• No Treatment Effect

Ha: Not All mi Are Equal



• At Least 1 Pop. Mean is Different

• Treatment Effect
NOT m1  m2  ...  mt
 Hypotheses

 H0: m1 = m2 = m3 = ... = mt
• All Population Means are f(X)
Equal
• No Treatment Effect
X
m1 = m2 = m3
 Ha: Not All mi Are Equal
• At Least 1 Pop. Mean is
Different f(X)
• Treatment Effect
NOT m1  m2  ...  mt X
m1 = m 2 m 3
 One-Way ANOVA
Basic Idea

1. Compares 2 Types of Variation to test equality

of means
2. Comparison basis is ratio of variances
3. If treatment variation is significantly greater
than random variation then means are not
equal
4. Variation measures are obtained by
‘Partitioning’ Total Variation
 One-Way ANOVA
Partitions Total Variation
 One-Way ANOVA
Partitions Total Variation

Total variation
 One-Way ANOVA
Partitions Total Variation

Total variation

Variation due to
treatment
 One-Way ANOVA
Partitions Total Variation

Total variation

Variation due to Variation due to

treatment random sampling
 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
 One-Way ANOVA
Partitions Total Variation
Total variation
Variation due to
treatment
 Sum of Squares Among
 Sum of Squares Between
 Sum of Squares
Treatment (SST)
 Among Groups Variation
Variation due to
random sampling
 Sum of Squares Within
 Sum of Squares Error
(SSE)
 Within Groups Variation
 Total Variation

SS Total   X 11  X   X 21  X     X ij  X 
2 2 2

Response, X

X

Group 1 Group 2 Group 3

 Treatment Variation

 2  2 
SST  n X  X  n X  X    n X  X
1 1 2 2 t t
2  
Response, X
X3
X
X2
X1

Group 1 Group 2 Group 3

Random (Error) Variation

SSE  X11  X1   X 21  X1     X tj  X t 
2 2 2

Response, X

X3
X2
X1

Group 1 Group 2 Group 3

 SStotal=SSE+SST

SS    X ij  X .. 
n1 ni

i 1 j 1

X ij  X i.    X i.  X .. 
n1 ni 2
  
i 1 j 1

 
n1 ni n ni
   X ij  X i.     X i.  X .. 
2 1 2
i 1 j 1 i 1 j 1

 2   X ij  X i.  X i.  X .. 
n1 ni

i 1 j 1
 But

  X ij  X i.  X i.  X .. 
n1 ni

i 1 j 1

   X i.  X ..   X ij  X i. 
n1 ni

i 1 j 1
n1
   X i.  X .. n X i.  n X i. 
i 1
0
Thus, SStotal=SSE+SST

SS    X ij  X i.      X i.  X .. 
n1 ni 2 n1 ni 2
i 1 j 1 i 1 j 1

   X ij  X i.    ni  X i.  X .. 
n1 ni 2 n1 2
i 1 j 1 i 1
 SSE  SST
 One-Way ANOVA 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 = t -1
 2 = tr - t
 t = # Populations, Groups, or Levels

 tr = Total Sample Size

 One-Way ANOVA
Summary Table

Source of Degrees Sum of Mean F

Variation of Squares Square
Freedom (Variance)
Treatment t-1 SST MST = MST
SST/(t - 1) MSE
Error tr - t SSE MSE =
SSE/(tr - t)
Total tr - 1 SS(Total) =
SST+SSE
 ANOVA Table for a
Completely Randomized Design

Source of Sum of Degrees of Mean

Variation Squares Freedom Squares F

Treatments SST t-1 SST/t-1 MST/MSE

Error SSE tr - t SSE/tr-t

Total SSTot tr - 1
 The F distribution
 Two parameters
• increasing either one decreases F-alpha (except for
v2<3)
• I.e., the distribution gets smashed to the left

0 F
F ( v1 , v2 )

 One-Way ANOVA F-Test Critical
Value
If means are equal, F =
MST / MSE  1. Only
reject large F! Reject H0

Do Not 
Reject H 0

0 F
F ( t 1, tr -t)

Always One-Tail!
© 1984-1994 T/Maker Co.
 Example: Protein sources

Completely Randomized Design

Experiment is conducted to evaluate the

effect of protein sources (catfish, tilapia, tuna)
to the final catfish weight. Three ponds are
used as replications. Are three protein
sources equally effective?
Randomized Design Example

Factor (Protein source)

Protein Catfish Tilapia Tuna


source
(Treatments)
Experimental
units

 
 

Dependent 4 9 14
variable 5 7 18
(Response) 3 14 16
 Example: Protein source
Observation Catfish Tilapia Tuna

1 4 9 14
2 5 7 18
3 3 14 16

Sample Mean 4 10 16
Sample Variance 1 13 4
 Example: Protein sources

 Hypotheses
H0: m1 = m2 = m3
Ha: Not all the means are equal
where:
m1 = mean number of catfish
m2 = mean number of tilapia
m3 = mean number of tuna
 Example: Protein sources
 Mean Square among Treatments
Since the
_ sample
_ _ sizes are all equal:
μ (grand mean)= (4 + 10 + 16)/3 = 10
SSTR= 3(4–10)2+ 3(10–10)2+ 3(16–10)2= 216
MSTR = 216/(3 - 1) = 108
 Mean Square Error
SSE = 2(1) + 2(13) + 2(4) = 36
MSE = 36/(9 - 3) = 6
 Example: Protein sources
 Rejection Rule
Using test statistic: Reject H0 if Ftable > 5.14

Using p-value: Reject H0 if p-value < .05

where F.05 = 5.14 is based on an F distribution with 2

numerator degrees of freedom and 6 denominator degrees of
freedom
 Example: Protein Sources

 Test Statistic
Fcalc. = MST/MSE = 108/6= 18.00
 Conclusion
Since Fcalc. = 18.00 > F.05 = 5.14, we reject H0.
There is sufficient evidence to conclude that the
weight of harvested catfish are affected by protein
source.
 Example: protein source
 ANOVA Table
Source of Sum of Degrees of Mean
Variation Squares Freedom Squares Fcalc.

Treatments 216 2 108 18.00

Error 36 6 6
Total 252 8
 Using Excel’s ANOVA:
Single Factor Tool

 Value Worksheet (top portion)

10 SUMMARY
11 Groups Count Sum Average Variance
12 Observation 9 90 10 31,5
13 Catfish 3 12 4 1
14 Tilapia 3 30 10 13
15 Tuna 3 48 16 4
 Using Excel’s ANOVA:
Single Factor Tool
 Value Worksheet (bottom portion)
12 Catfish 3 12 4 1
13 Tilapia 3 30 10 13
14 Tuna 3 48 16 4
15
16 ANOVA
17 Source of Variation SS df MS Fcalculate Ftable
18 Among Groups 216 2 108 18 5,14
19 Within Groups 36 6 6
20
21 Total 252 8
 RCBD
(Randomized Complete Block Design)
 Randomized Complete Block
Design
 An experimental design in which there is one independent
variable, and a second variable known as a blocking
variable, that is used to control for confounding or
concomitant variables.
 It is used when the experimental unit or material are
heterogeneous
 There is a way to block the experimental units or materials
to keep the variability among within a block as small as
possible and to maximize differences among block
 The block (group) should consists units or materials which
are as uniform as possible
 Randomized Complete Block Design

 Confounding or concomitant variable are not being

controlled by the analyst but can have an effect on the
outcome of the treatment being studied
 Blocking variable is a variable that the analyst wants
to control but is not the treatment variable of interest.
 Repeated measures design is a randomized block
design in which each block level is an individual item or
person, and that person or item is measured across all
treatments.
 The Blocking Principle
 Blocking is a technique for dealing with nuisance factors
 A nuisance factor is a factor that probably has some
effect on the response, but it is of no interest to the
experimenter. However, the variability it transmits to the
response needs to be minimized
 Typical nuisance factors include batches of raw material,
operators, pieces of test equipment, time (shifts, days, etc.),
different experimental units
 Many industrial experiments involve blocking (or should)
 Failure to block is a common flaw in designing an
experiment
 The Blocking Principle
 If the nuisance variable is known and controllable, we
use blocking
 If the nuisance factor is known and uncontrollable,
sometimes we can use the analysis of covariance to
statistically remove the effect of the nuisance factor from
the analysis
 If the nuisance factor is unknown and uncontrollable
(a “lurking” variable), we hope that randomization
balances out its impact across the experiment
 Sometimes several sources of variability are combined
in a block, so the block becomes an aggregate variable
 Class positions
Randomized Complete Block Design
(3 treatments x 3 Blocks)

0,75 x 0,75
1,00 x 1,00
1,25 x 1,25
 Class positions
Randomized Complete Block Design
(3 treatments x 3 Blocks)

A 0,75 x 0,75
1,00 x 1,00
1,25 x 1,25
 Class positions
Randomized Complete Block Design
(3 treatments x 3 Blocks)

A C 0,75 x 0,75
1,00 x 1,00
1,25 x 1,25
 Class positions
Randomized Complete Block Design
(3 treatments x 3 Blocks)

A C B 0,75 x 0,75
1,00 x 1,00
1,25 x 1,25
 Class positions
Randomized Complete Block Design
(3 treatments x 3 Blocks)

A C B 0,75 x 0,75
B 1,00 x 1,00
1,25 x 1,25
 Class positions
Randomized Complete Block Design
(3 treatments x 3 Blocks)

A C B 0,75 x 0,75
B A 1,00 x 1,00
1,25 x 1,25
 Class positions
Randomized Complete Block Design
(3 treatments x 3 Blocks)

A C B 0,75 x 0,75
B A C 1,00 x 1,00
1,25 x 1,25
 Class positions
Randomized Complete Block Design
(3 treatments x 3 Blocks)

A C B 0,75 x 0,75
B A C 1,00 x 1,00
C 1,25 x 1,25
 Class positions
Randomized Complete Block Design
(3 treatments x 3 Blocks)

A C B 0,75 x 0,75
B A C 1,00 x 1,00
C A 1,25 x 1,25
 Class positions
Randomized Complete Block Design
(3 treatments x 3 Blocks)

A C B 0,75 x 0,75
B A C 1,00 x 1,00
C A B 1,25 x 1,25
 Partitioning the Total Sum of Squares
in the Randomized Block Design

SStotal
(total sum of squares)

SSE
(error sum of squares)

SST SSB SSE’

(treatment (sum of squares (sum of squares
sum of squares) blocks) error)
 A Randomized Block Design

Single Independent Variable

.
MST
MSE Individual
.
observations
Blocking
Variable .
. . . .
. . . . .
. . . .
.
 The Linier Model
y  μ  τ ρ ε
ij i j ij
i = 1,2,…, t j = 1,2,…,r
yij = the observation in ith treatment in the jth block

m = overall mean
ti = the effect of the ith treatment
No interaction
rj = the effect of the jth block between blocks
and treatments
eij = random error
 Extension of the ANOVA to the RCBD

ANOVA partitioning of total variability:

 
t r t r

 (yij  y.. )  (yi.  y.. )  (y.j  y.. )  (yij  yi.  y.j  y.. )
2 2

i 1 j1 i 1 j1
t r t r
 r  (yi.  y.. )  t  (y.j  y.. )   (y ij  yi.  y.j  y.. ) 2
2 2

i 1 j1 i 1 j1

SST  SSTreatments  SS Blocks  SS E

Extension of the ANOVA to the RCBD

The degrees of freedom for the sums of squares in

SST  SSTreatments  SS Blocks  SS E

are as follows:
tr  1  (t  1)  (r  1)  [(t  1)( r  1)]
 Ratios of sums of squares to their degrees of freedom result
in mean squares, and
 The ratio of the mean square for treatments to the error
mean square is an F statistic  used to test the hypothesis of
equal treatment means
 ANOVA Procedure
 The ANOVA procedure for the randomized block design
requires us to partition the sum of squares total (SST) into
three groups: sum of squares due to treatments, sum of
squares due to blocks, and sum of squares due to error.
 The formula for this partitioning is

SSTot = SSTreatment + SSBlock + SSE

 The total degrees of freedom, tr - 1, are partitioned such that

t - 1 degrees of freedom go to treatments, r - 1 go to blocks,
and (t - 1)(r - 1) go to the error term.
 ANOVA Table for a
Randomized Block Design
Source of Sum of Degrees of Mean
Variation Squares Freedom Squares F

Treatments SST t–1 SST/t-1 MST/MSE

Blocks SSB r-1

Error SSE (t - 1)(r - 1) SSE/(t-1)(r-1)

Total SSTot tr - 1
 Example: Protein sources

Randomized Block Design

There are three protein sources (A: Catfish; B:
Tilapia; C: Tuna) that will be evaluated. To study the
effect of protein sources to the final catfish weight,
three classes of ponds will used for the experiment
(0,75 x 0,75; 1 x 1 and 1,25 x 1,25). One catfish will be
taken randomly as a sample for each protein source.
Is there any different final catfish weight from each
protein sources in the feed?
 Example: Protein sources

Class Protein source Block

(Block) A B C Means
1 4 9 14 9
2 5 7 18 10
3 3 14 16 11

Treatment
Means 4 10 16 10
 Example: Protein sources
 Mean Square Due to Treatments
The overall sample mean is 10. Thus,
SSTreatment = 3[(4-10)2+ (10-10)2+ (16-10)2]= 216.00
MSTreatment = 216/(3 - 1) = 108
 Mean Square Due to Blocks
SSBlock = 3[(9-10)2 + (10-10)² + (11-10)2] = 6
MSBlock= 6/(3 - 1) = 2
 Mean Square Due to Error
CF = (90)2 /(3x3) = 900
SStotal = (4)2+ …+ (3)2 – 900 = 252
SSE = 252 - 216 – 6 = 30
MSE = 30 /[(3 - 1)(3- 1)] = 7.5
 Example: protein source
 ANOVA Table
Source of Sum of Degrees of Mean
Variation Squares Freedom Squares Fcalc.
Classes 6 2 3 0.03
Treatments 216 2 108 14.40
Error 30 4 7.5
Total 252 8
 Example: Protein sources
 Rejection Rule
Using test statistic: Reject H0 if Ftable > 6.94

Assuming α = .05, F.05 = 6.94 (2 d.f. numerator

and 4 d.f. denominator)
 Example: Protein sources
 Test Statistic
F = MSTreatment/MSE = 108/7.5 = 14.40
 Conclusion
Since 14.40> 6.94, we reject H0. There is sufficient
evidence to conclude that protein sources affects the final
catfish weight
 Using Excel’s Anova

 Step 1 Select the Tools pull-down menu

 Step 2 Choose the Data Analysis option
 Step 3 Choose Anova: Two Factor Without
Replication from the list of Analysis Tools
 Using Excel’s Anova

 Step 4 When the Anova: Two Factor Without

Replication dialog box appears:
Enter A1:D6 in the Input Range box
Select Labels
Enter .05 in the Alpha box
Select Output Range
Enter A8 (your choice) in the Output Range box
Click OK
 Using Excel’s Anova:
Two-Factor Without Replication Tool

 Value Worksheet (top portion)

1
2
Class
1
A
4
B
9
C
14
3 2 5 7 18
4 3 3 14 16
5
6
7
 Using Excel’s Anova:
Two-Factor Without Replication Tool

Value
9
 10
Class
Worksheet
Count
(middle portion)
Sum Average
11 0,75 x 0,75 3 27 9
12 1,00 x 1,00 3 30 10
13 1,25 x 1,25 3 33 11
16
17 Catfish 3 12 4
18 Tilapia 3 30 10
19 Tuna 3 48 16
20
21
22
23
 Using Excel’s Anova

A B C D E G
8 ANOVA
9 Variation SS df MS Fcalc F table
10 Classes 6 2 3 0,02778 6,94
11 Protein source 216 2 108 14,4 6,94
12 Error 30 4 7,5
13
14 Total 252 8
 Using Excel’s Anova:
Two-Factor Without Replication Tool

 Conclusion Using F table

• The value worksheet shows that F table for protein
sources is 6.94
• The rejection rule is “Reject H0 if F calculated > F
Table”
• Thus, we reject H0 because F calculated > F Table
for  = .05
• There is sufficient evidence to conclude that protein
sources affect final catfish weight
 Similarities and differences between
CRD and RCBD: Procedures
 RCBD: Every level of “treatment” encountered by each
experimental unit; CRD: Just one level each
 Descriptive statistics and graphical display: the same as
CRD
 Model adequacy checking procedure: the same except:
specifically, NO Block x Treatment Interaction
 ANOVA:
Inclusion of the Block effect;
dferror change from t(r – 1) to (t – 1)(r – 1)
Latin Square Design
 Definition
 A Latin square is a square array of objects (letters A, B,
C, …) such that each object appears once and only
once in each row and each column.

 Example - 4 x 4 Latin Square.

ABCD
BCDA
CDAB
DABC
 The Latin Square Design
 This design is used to simultaneously control (or eliminate) two
sources of nuisance variability (confounding variables)
 It is called “Latin” because we usually specify the treatment by the
Latin letters
 “Square” because it always has the same number of levels (t) for the
row and column nuisance factors
 A significant assumption is that the three factors (treatments and two
nuisance factors) do not interact
 More restrictive than the RCBD
 Each treatment appears once and only once in each row and column
 If you can block on two (perpendicular) sources of variation
A B (rowsC xD
columns) you can reduce experimental error when comparedB C to D theA
RCBD
C D A B
D A B C
 Useful in Animal Nutrition
Studies
 Suppose you had four feeds you wanted to test on dairy
cows. The feeds would be tested over time during the
lactation period
 This experiment would require 4 animals (think of
these as the rows)
 There would be 4 feeding periods at even intervals
during the lactation period beginning early in lactation
(these would be the columns)
 The treatments would be the four feeds. Each animal
receives each treatment one time only.
The “Latin Square” Cow

Mid Mid
Early Late
Early Late
Uses in Field Experiments
 When two sources of variation must be controlled
• Slope and fertility
• Furrow irrigation and shading
• If you must plant your plots perpendicular to a linear
gradient
‘Row’
1 2 3 4

A B C D B C D A C D A B D A B C
1 2 3 4
‘Column’

 Practically speaking, use only when you have more

than four but fewer than ten treatments
• a minimum of 12 df for error
 Latin Square Designs
Randomization → Used selected latin squares
Selected Latin Squares
3x3 4x4
ABC ABCD ABCD ABCD ABCD
BCA BADC BCDA BDAC BADC
CAB CDBA CDAB CADB CDAB
DCAB DABC DCBA DCBA

5x5 6x6
ABCDE ABCDEF
BAECD BFDCAE
CDAEB CDEFBA
DEBAC DAFECB
ECDBA FEBADC
Randomization
Selected Latin Square
A B C

B C A

C A B
 Randomization
Randomize the order of rows: 2
B C A
 Randomization
Randomize the order of rows: 1
B C A

A B C
 Randomization
Randomize the order of rows: 3
B C A

A B C

C A B
 Randomization
Randomize the order of columns: 3
A

B
 Randomization
Randomize the order of columns: 1
A B

C A

B C
 Randomization
Randomize the order of columns: 2
A B C

C A B

B C A
 Latin Square Designs
There are three factors:
 Treatments (t) (letters A, B, C, …)

 Rows (t)

 Columns (t)

The number of treatments = the number of rows =

the number of columns = t.
The row-column treatments are represented by cells in
a t x t array.
The treatments are assigned to row-column
combinations using a Latin-square arrangement
 Advantages and Disadvantages

 Advantage:
• Allows the experimenter to control two sources of variation
 Disadvantages:
• Error degree of freedom (df) is small if there are only a few
treatments
• The experiment becomes very large if the number of
treatments is large
• The statistical analysis is complicated by missing plots and
mis-assigned treatments
 Analysis
 Set up a two-way table and compute the row and column
totals
 Compute a table of treatment totals and means
 Set up an ANOVA table divided into sources of variation
• Rows
• Columns
• Treatments
• Error
 Significance tests
• FT tests difference among treatment means
• FR and FC test if row and column groupings are effective
 The Linier Model
yij k   m  t k  r i   j  e ij k 
i = 1,2,…, t j = 1,2,…, t k = 1,2,…, t
yij(k) = the observation in ith row and the jth column
receiving the kth treatment
m = overall mean
tk = the effect of the kth treatment
No interaction
ri = the effect of the i row
th
between rows,
j = the effect of the jth column columns and
treatments
eij(k) = random error
 Latin Square

 A Latin Square experiment is assumed to be a

three-factor experiment.
 The factors are rows, columns and treatments.
 It is assumed that there is no interaction between
rows, columns and treatments.
 The degrees of freedom for the interactions is used
to estimate error.
 The Anova for a Latin Square

Source S.S. d.f. M.S. Fcal Ftable

Treat SST t-1 MST MST /MSE Fα, (t-1);
(t-1)(t-2)

Rows SSRow t-1 MSRow

Cols SSCol t-1 MSCol
Error SSE (t-1)(t-2) MSE
Total SST t2 - 1
 Example
The effect of protein sources on the final catfish
weight.
Ponds as an experimental unit can be grouped based
on size of pond and original catfish

Treatment:
A: Catfish
B: Tilapia
C: Tuna
 Protein sources
_
3-4 5-6 7-8 Xi. Xi
0,75 x 0,75 A (4) B(9) C(14) 27 9
1x1 C(18) A(5) B(7) 30 10
1,25 x 1,25 B(14) C(16) A(3) 33 11
Xj 36 30 24 90
_
Xj 12 10 8 10
A= 12 B=
_
30 C=48
_
Ā= 4 B= 10 C= 12
 Final Exam Notes

CF= (90)²/9= 900

SStotal= (4)²+ …+(3)² - 900= 252
SStreatment = (12)²/3 + …+ (48)²/3 – 900= 216
SSrows = (27)²/3 + …+ (33)²/3 – 900= 6
SScolumn = (36)²/3 + …+ (24)²/3 – 900= 24
SSerror = SStotal – SStreatment – SSrows – SScolumn
= 252 – 216 – 06 – 24 = 6
 The Anova Table
Source S.S. d.f. M.S. Fcalc Ftable
Catfish original 24 2 12 4 19
size
Pond size 6 2 3 1 19
Protein source 216 2 108 36 19
Error 6 2 3
Total 252 8
Graeco-Latin Square
Designs
Mutually orthogonal Squares
 Definition
A Greaco-Latin square consists of two latin squares (one using
the letters A, B, C, … the other using greek letters a, b, c, …)
such that when the two latin square are supper imposed on each
other the letters of one square appear once and only once with
the letters of the other square. The two Latin squares are called
mutually orthogonal.

Example: a 7 x 7 Greaco-Latin Square

A Be Cb Df Ec F Gd
Bb Cf Dc E Fd G Ae
Cc D Ed F Ge Ab Bf
Dd E Fe Gb Af Bc C
Ee Fb Gf Ac B Cd D
Ff Gc A Bd C De Eb
G Ad B Ce Db Ef Fc
The Graeco-Latin Square Design

√ This design is used to simultaneously control (or

eliminate) three sources of nuisance variability
√ It is called “Graeco-Latin” because we usually specify
the third nuisance factor, represented by the Greek
letters, orthogonal to the Latin letters
√ A significant assumption is that the four factors
(treatments, nuisance factors) do not interact
√ If this assumption is violated, as with the Latin square
design, it will not produce valid results
√ Graeco-Latin squares exist for all t ≥ 3 except t = 6
Note:

At most (t –1) t x t Latin squares L1, L2, …, Lt-1 such

that any pair are mutually orthogonal.

It is possible that there exists a set of six 7 x 7

mutually orthogonal Latin squares L1, L2, L3, L4, L5,
L6 .
 The Linear Model
yijkl   m  t k  l  ri   j  e ijkl 

i = 1,2,…, t j = 1,2,…, t

k = 1,2,…, t l = 1,2,…, t

yij(kl) = the observation in ith row and the jth column receiving the
kth Latin treatment and the lth Greek treatment
m = overall mean
tk = the effect of the kth Latin treatment
l = the effect of the lth Greek treatment
ri = the effect of the ith row

j = the effect of the jth column

eij(k) = random error

No interaction between rows, columns, Latin

treatments and Greek treatments
 A Greaco-Latin Square experiment is assumed to
be a four-factor experiment.
 The factors are rows, columns, Latin treatments
and Greek treatments.
 It is assumed that there is no interaction between
rows, columns, Latin treatments and Greek
treatments.
 The degrees of freedom for the interactions is used
to estimate error.