Scribd
Upload
59 views4 pages

Analysis of Variance (ANOVA) : Completely Randomized Design (CRD)

This document discusses analysis of variance (ANOVA) and provides examples of its use. It introduces ANOVA as a technique to partition total variation in a data set into components associated with different sources of variation. One-way ANOVA allows testing if three or more population means are equal. Examples are provided of using one-way ANOVA to analyze data from completely randomized experiments with different treatments and experimental units. The document outlines the hypotheses, calculations, and statistical tests involved in one-way ANOVA.

Uploaded by

Robert Kier Tanquerido Tomaro
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
59 views4 pages

Analysis of Variance (ANOVA) : Completely Randomized Design (CRD)

This document discusses analysis of variance (ANOVA) and provides examples of its use. It introduces ANOVA as a technique to partition total variation in a data set into components associated with different sources of variation. One-way ANOVA allows testing if three or more population means are equal. Examples are provided of using one-way ANOVA to analyze data from completely randomized experiments with different treatments and experimental units. The document outlines the hypotheses, calculations, and statistical tests involved in one-way ANOVA.

Uploaded by

Robert Kier Tanquerido Tomaro
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 4

Analysis of Variance (ANOVA)

Analysis of Variance is a technique whereby the total variation present in a set of data
is partitioned into several components. Associated with each of these components is a
specific source of variation, so that, in the analysis, it is possible to ascertain the magnitude of
the contribution of each of these sources to the total variation.
The techniques and concepts of analysis of variance are used most frequently to test
hypotheses about the equality of three or more population means. The introduction and
development of the ANOVA techniques are due to R.A. Fisher, whose contributions over the
years 1912 to 1962 had a tremendous influence on modern statistical thought.
A treatment is any factor that the experimenter controls. It may refer, for example, to
type of drug, one several concentrations of a single drug, a new type of house paint, an
advertising technique, or a particular training program.
The entity that receives a treatment is called an experimental unit. An experimental
unit may be an individual, a single white mouse, a group of white mice, plot of ground, a
segment of a consuming public, a group of trainees, or an item of production.
In this lesson, the expression mean square = variance.
We will discuss ANOVA in the context of three different experimental designs:
completely randomized, the randomized complete bock, and the Latin square.

Completely Randomized Design (CRD)


When we use the CRD, we assign the treatments at random to the experimental units.
Suppose for example, that we want to road test four brands of tires, A, B, C, and D, to
determine whether there are any differences among the brands with respect to expected
mileage. We can assign 10 tires of each brand at random to the 40 rear wheels of 20 cars.
We can then drive the cars until a predetermined amount of tread wear occurs. At that time,
we record the number of miles driven. We then use ANOVA to decide whether the brands
differ with respect to expected tire mileage.
We analyze data from an experiment using the CRD by what is known as one-way
analysis of variance. The term one-way refers to the fact that we classify the experimental
units (and consequently the measurements obtained) according to only one criterion- the
treatment group to which they belong. There are two sources of variations here that together
constitute the total variation, namely: (1) variability due to treatment; and (2) variability due
to all other sources, called the residual or error variability.

The One-way ANOVA Hypotheses


We are testing the hypothesis that all treatment, or population, means are equal (H 0)
against the alternative that there is at least one inequality among them. In general, we may
state the hypotheses symbolically as follows:
Ho : μ1 = μ2 = μ3 = ….= μk Ha: not all μj are equal

One-way ANOVA Calculations


We calculate the following:
2
Correction Term: C= ¿
n
n
2
Total Sum of Squares (SST): SST = ∑ x i – C.
i=1
k
P2j
Treatment Sum of Squares: SSTr = ∑ – C.
j=1 nj
Error Sum of Squares: SSE = SST – SSTr
SSTr
Treatment Mean Square: MSTr =
k−1
SSE
Error Mean Square: MSE =
n −k
MSTr
Fcomputed: F=
MSE

Sample (CRD)
JaDine Plastic wants to know what effect three formula ingredients have on the
elasticity of its plastic products. Each of the ingredients is randomly assigned to
batches experimental material. The table below shows the results of elasticity tests
made on each specimen of the product. The manufacturer wishes to know whether
the formula ingredients have a differential effect on the elasticity of the plastic. Set α
= 0.05.

Elasticity of plastic product produced with three different formula ingredients (A, B,
and C)

A 5 6 5 8 6 7 6 5 6 7
B 8 9 8 7 9 9 10 8
C 10 10 9 8 8 9 10 9 8 9 10 8

In this problem our treatment (our groups or independent variable) is the formula
ingredient (A,B, and C), while our dependent variable [the one that we measure or get from
each group (our basis in order to determine if the groups differ or not)] ] is the elasticity of
plastic.

Hypothesis testing

In the construction of our hypothesis, we will take the form where we specify the groups (we
refer here as treatment/independent variable) and the dependent variable.

1. Ho: The formula ingredients do not have a differential effect on the elasticity of plastic.

(μA = μB= μC).


In our Ho there are three equalities (μA = μB; μB= μC; and μA= μC)

Ha: The formula ingredients have differential effect on the elasticity of plastic. (at least one
equality in Ho does not hold).

2. α = 0.05

3. F-statistic

4. Fcomputed

Sum
A 5 6 5 8 6 7 6 5 6 7 61
B 8 9 8 7 9 9 10 8 68
C 10 10 9 8 8 9 10 9 8 9 10 8 108
GT 237

nA = 10 (number of observations/experimental units in A)


nB = 8 (number of observations/experimental units in B)
nC = 12 (number of observations/experimental units in C)
n = 10 + 8 +12 = 30
k= 3 (we have 3 treatments/groups)
Correction term:
2
C= ¿
n
GT2 = the summation of the sum in each group ;
n = total number of experimental units (values)

2372
C= = 1872.3
30

52+ 62 +52 +...+92 +102 +82


C=
30
Total sum of squares
n
2
SST = ∑ x i – C = 52 +62 +52 +...+9 2+10 2+ 82 - 1872.3 = 1945 - 1872.3 = 72.7
i=1

Treatment sum of squares


k
P2j 612 682 1082
SSTr = ∑ –C= + + - 1872.7 = 49.80
j=1 nj 10 8 12
Error sum of squares
SSE = SST – SSTr = 72.7 - 49.8 = 22.9
Treatment Mean square
SSTr 49.80
MSTr = = = 24.9
k−1 3 −1
Error mean square
SSE 22.9
MSE = = = 0.8481
n −k 30− 3
MSTr 24.9
Fcomputed = = = 29.35
MSE 0.8481
5. Ftabular [refer to table A.3; F (5% significance level) since our α = 0.05 ]
Degrees of freedom numerator = k -1 = 3-1 = 2 [see top row of the table(v1)]
Degrees of freedom denominator = n - k = 30-3 = 27 [see first column of the table(v 2)]
Ftabular = 3.35

6. Fcomputed vs Ftabular
29.35 > 3.35 reject Ho

7. Conclusion: The formula ingredients have differential effect on the elasticity of plastic.

====

Task
The table below shows the results, in miles per gallon, of an experiment conducted to
compare three brands of gasoline. Each brand was used with seven different cars of
the same weight and engine size, driven under similar conditions. Do these data
provide sufficient evidence at the 0.01 level of significance to indicate that the three
gasoline brands differ?

Brand A 14 19 19 16 15 17 20
Brand B 20 21 18 20 19 19 18
Brand C 20 26 23 24 23 25 23

Note: for Ftabular; use table A.3 Critical Values of F (1% significance level) since our α = 0.05

You might also like