STA4C04 - Statistical Inference and Quality Control
Uploaded by
shanSTA4C04 - Statistical Inference and Quality Control
Uploaded by
shanSTATISTICAL INFERENCE AND
QUALITY CONTROL
(STA4C04)
STUDY MATERIAL
IV SEMESTER
B.Sc. MATHEMATICS
(2019 Admission)
UNIVERSITY OF CALICUT
SCHOOL OF DISTANCE EDUCATION
CALICUT UNIVERSITY P.O.
MALAPPURAM - 673 635, KERALA
19562
School of Distance Education
University of Calicut
Study Material
IV Semester
B.Sc. MATHEMATICS
(2019 Admission)
Complementary Course : STA4C04
STATISTICAL INFERENCE AND QUALITY CONTROL
Prepared by:
Dr. APARNA ARAVINDAKSHAN M.,
Assistant Professor,
Department of Statistics,
St. Joseph’s College, Devagiri.
Scrutinized by:
Ms. JEENA M.P.,
Assistant Professor,
Department of Statistics,
Thunjan Memorial Government College, Tirur.
DISCLAIMER
"The author(s) shall be solely responsible
for the content and views
expressed in this book".
STA4C04 : Statistical Inference and Quality Control
CONTENTS
1 Estimation Theory 1
1.1 Statistical Inference . . . . . . . . . . . . . . . . 1
1.2 Estimation of Parameters . . . . . . . . . . . . . 5
1.3 Point Estimation . . . . . . . . . . . . . . . . . . 7
1.4 Interval Estimation . . . . . . . . . . . . . . . . . 30
2 Testing of Hypothesis 49
2.1 Basic Concepts . . . . . . . . . . . . . . . . . . . 50
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
ii
2.2 Large Sample Tests . . . . . . . . . . . . . . . . . 62
2.3 Small Sample Tests . . . . . . . . . . . . . . . . . 76
2.4 Analysis of Variance (ANOVA) . . . . . . . . . . 101
3 Non-parametric Tests 115
3.1 Advantages and Disadvantages . . . . . . . . . . 116
3.2 Applications of Non-Parametric Test . . . . . . . 116
3.3 Test for Randomness . . . . . . . . . . . . . . . . 117
3.4 Problem of Location - Sign Test . . . . . . . . . . 120
3.5 Problem of Equality of two Populations . . . . . 127
3.6 Problem of Equality of Several Population Medians136
4 Quality Control 139
4.1 Basis of S.Q.C. . . . . . . . . . . . . . . . . . . . 141
4.2 Control Charts for Variables . . . . . . . . . . . . 152
4.3 Control Charts for Attributes . . . . . . . . . . . 158
REFERENCES 164
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
MODULE
ONE
Estimation Theory
1.1 Statistical Inference
Statistical inference is the branch of statistics which discuss
about making inferences about the unknown aspects of the dis-
tribution of the population based on a sample drawn from it.
Alternatively, statistical inference deals with
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
1. methods for drawing conclusions about a population
2. and measuring the reliability of conclusions so obtained.
These conclusions and their measure of reliability is acquired
with the help of sample data. The unknown aspects may be the
form of the distribution, or the values of the parameter involved
in it, or both. In contrast with descriptive statistics, inferen-
tial statistics assumes that data comes from a larger popula-
tion. Whereas, descriptive statistics is merely concerned with
the properties of the observed data.
Basically, there are two types of statistical inference:
1. Parametric statistical inference.
2. Non-parametric statistical inference.
Parametric and Non-parametric Statistical In-
ference
In parametric statistical inference, it is assumed that the under-
lying population distribution is known except for the parameters
involved in it. Hence, making inferences on the population pa-
rameter completely specifies the population distribution. On
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
the flip side, non-parametric statistical inference does not make
any such assumptions. The terminology ‘non-parametric’ does
not mean that there is no parameter involved. It only means
that making inferences on the population parameters does not
completely specifies the population distribution.
Properties like central tendency, dispersion etc. of an under-
lying probability distribution are of great interest as it describes
the population distribution. Whenever we talk about certain
distributions we associate with some parameters. For example
N (µ, σ 2 ), B(n, p), etc. These are important parameters for a
distribution because if we looked at them carefully, we can see
that the mean and variance come from these parameters. For
normal, the mean is µ and the variance is σ 2 . They are involved
in the p.d.f. of Normal distribution as parameters. For bino-
mial, the mean is np and variance is np(1 − p). In this case,
we can derive mean and variance from the parameters of the
binomial distribution. If we estimate the parameters of a dis-
tribution, then we get a feel of the overall population. Most
parametric methods assume that:
1. Data is quantitative.
2. Population has a normal distribution.
3. Sample is sufficiently large.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
But in practice, that is not the case always. For example,
1. Data may be categorical or ordinal. To illustrate, consider
the scores of students in an examination. If marks are used
as scores, we have a quantitative data. But, in grading
system students having same grade is somewhat similar
but their marks are not equal, and we have categorical
data. If ranking system is used, we know the position
of students, but the distance between 1st and 2nd rank
won’t be the same as that between 2nd and 3rd . That is,
variables are natural ordered categories and the distance
between the categories are not known.
2. Population size may be too small to be treated as normal.
If the normal distribution is not appropriate, it is common
to consider the following possibilities:
(a) Transform data to normal when it is possible and
use normal methods to the transformed data. By
transformation we mean, if a take a function of the
data, we may satisfy normality assumption.
(b) Use some other known non-normal distribution like
exponential distribution.
(c) Use non-parametric methods. It makes no assump-
tions about the population distribution or the sam-
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
ple size. But, sometimes we may have a very general
assumption like the data is from a continuous distri-
bution.
Basic tasks of statistical inference can be categorised as:
1. Estimation of a parameter
2. Testing of hypothesis
1.2 Estimation of Parameters
Estimation of parameters is a branch of statistical inference
which deals with estimating the values of the parameters based
on sample data. There are parametric and non-parametric
methods for estimating the parameters of a population. We will
be dealing with parametric methods only. Hence, we assume
that we have a random sample of observations generated from
a population whose probability distribution is known, except
for the values of the parameters. The theory of estimation was
put forward by Prof. R.A. Fisher in his research papers around
1930. Parametric estimation is classified into two under the
headings
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
1. Point estimation
2. Interval estimation
Parametric Space
The set of all admissible values of the parameters of probability
distribution is called the parametric space, denoted by Θ.
Example 1.2.1. For P (λ), Θ = {λ > 0}.
Example 1.2.2. For N (µ, σ 2 ), Θ = (−∞, +∞) × (0, +∞).
Estimator and Estimate
A statistic T , used for estimating the population parameter is
called an estimator. This statistic is used as a rule for calculating
an estimate of a given quantity based on observed data.
The numerical value taken by an estimator when a particular
sample is realised is called an estimate.
That is, the term ‘estimator’ represents a rule (or statistic)
used for estimating the unknown parameter and the estimate
represents a specific value produced by the estimator.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
1.3 Point Estimation
If from the observations in a sample, a single value is calculated
as an estimate of the unknown population parameter, the proce-
dure is referred to as point estimation and we refer to the value
of the statistic (estimator) as the point estimate. For example,
if we use a particular value of the statistic X̄ (say, x̄ = 10 ) to
estimate the mean µ of a population, we are using a point esti-
mate of µ. Here X̄ is referred to as an estimator and the specific
value used for estimation, i.e. x̄ = 10 is called the estimate of
µ.
Let X1 , X2 , . . . , Xn be i.i.d. B(1, p) r.v.s. Then the following
are point estimators of p.
T1 = X̄
T2 = X1
X1 + X2
T3 =
2
Xn n
X
T4 = aj Xj , 0 ≤ aj ≤ 1 aj = 1
j=1 j=1
Hence, the problem of point estimation is to pick a statistic that
best estimates the unknown parameter. Clearly, we would like
the estimator T of θ to be as close to θ as possible. Since T is
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
a statistic, the usual measure of closeness |T − θ| is also a r.v.
Examples of such measures of closeness are:
(1) P (|T − θ| < ) for some > 0
(2) E|T − θ|r for some r > 0
we want the first to be large and the second to be small.
Mean Square Error
When r = 2, the quantity defined in (2) is called the mean
square error and we denote it by
M.S.E.θ (T ) = E(T − θ)2
= E[T − E(T ) + E(T ) − θ]2
= E[T − E(T )]2 + E[E(T ) − θ]2
= V (T ) + (Bias)2
An estimator with small M.S.E. has small bias and variance.
So, to control M.S.E. we need to control both variance and bias.
Various statistical properties of estimators can be used to
decide which estimator is most appropriate in a given situation.
That is, which estimator expose us to the smallest risk, which
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
will give us more information at the lowest cost and so on. There
are four desirable characteristics of a good estimator. They are,
i) Unbiasedness
ii) Consistency
iii) Efficiency
iv) Sufficiency
Unbiasedness
An estimator is said to be unbiased, if the expected value of the
statistics is equal to the value of the parameter to be estimated.
An estimator not having this property is called a biased esti-
mator. This property ensures that, on an average the estimator
T has no systematic error; it neither over estimate nor under
estimate θ, on an average.
Definition 1.3.1. Let X be a random variable (defined over a
population) having the probability function f (x) and θ be the pa-
rameter to be estimated. Let X1 , X2 , . . . , Xn be a random sample
taken from the population. Let T = T (X1 , X2 , ...Xn ) be an esti-
mator used to estimate θ. The estimator T is called an unbiased
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
10
estimator if
E(T ) = θ ∀ n.
E(T ) − θ is called the bias of the estimator T . Bias can be
positive, negative or zero.
Problems
1. Let X1 , X2 , . . . , Xn be a r.s. taken from a population with
mean µ and let T = X̄. Check whether T is unbiased for
µ.
2. Examine whether s2 = n1 (Xi − X̄)2 is unbiased for σ 2
P
if the population is N (µ, σ 2 ).
Note: As n → ∞, (n−1) n → 1. Hence, s2 is asymptotically
2
unbiased for σ .
3. Let X ∼ P (λ). A r.s. of size n is taken from X. Let
1
Pn
T1 = X̄, T2 = X1 and T3 = n−1 i=1
(Xi − X̄)2 . Check
whether T1 , T2 and T3 are unbiased for λ. Suggest another
unbiased estimator for λ.
Note:
(a) Unbiased estimator need not be unique.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
11
(b) Convex linear combination of unbiased estimator are
also unbiased estimators.
1
Xi2 . S.T. T is unbiased for
P
4. Let X ∼ N (µ, 1) and T = n
µ2 + 1.
√
5. If T is an unbiased estimator of θ, S.T. T 2 and T are
√
biased estimators of θ2 and θ respectively.
Note: If T is an unbiased estimator of θ, then ψ(T ) is an
unbiased estimator of ψ(θ) iff ψ is a linear function.
T (T −1)
6. If X1 , X2 , . . . , Xn ∼ B(1, p), S.T. n(n−1) is an unbiased
estimator of p2 , where T = Xi .
P
7. S.T the sample mean is an unbiased estimator of θ1 for the
distribution f (x; θ) = θ(1 − θ)x−1 ; x = 1, 2, . . .; 0 < θ < 1.
Consistency
One of the basic properties of a good estimator is that it provides
increasingly more precise information about the parameter θ
with increase in the sample size n. Accordingly we have the
following definition of the consistency of an estimator.
Definition 1.3.2. A sequence of estimators Tn will be called
consistent for the parameter θ, if Tn converges to θ in probabil-
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
12
ity. i.e., for > 0,
lim P (|Tn − θ| ≤ ) = 1
n→∞
or
lim P (|Tn − θ| ≥ ) = 0.
n→∞
P
And we write Tn → θ.
Consistency is a large sample property.
Theorem 1.3.1 (Sufficient condition for consistency).
If Tn is a sequence of estimators such that E(Tn ) → θ and
V (Tn ) → 0 as n → ∞, then Tn is consistent for θ.
Proof. We have, E(Tn ) → θ and V (Tn ) → 0 as n → ∞. By
Chebychev’s inequality
V (Tn )
P [|Tn − E(Tn )| > ] ≤ .
2
Therefore,
P [|Tn − θ| > ] → 0 as n → ∞.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
13
Problems
1. S.T. sample mean X̄ is a consistent estimator of popula-
tion mean µ of an N (µ, σ 2 ) population.
2. S.T. s2 = n1 (Xi − X̄)2 is a consistent estimator of pop-
P
ulation variance in the case of an N (µ, σ 2 ) population.
3. Let n be the number of Bernoulli trials, X the number of
successes in a series of n trials with probability of success
for p for each trial. S.T. X n is a consistent estimator for p.
Note: Let X1 , X2 , . . . , Xn be a r.s. of size n from N (µ, σ 2 )
and the statistic ‘sample median’ be denoted by M . Then,
2
M → N (µ, π2 σn ) as n → ∞.
4. S.T. sample median is a consistent estimator of population
mean of a normal population.
nX̄
5. S.T. n+1 is a consistent estimator of λ in P (λ).
Theorem 1.3.2. If Tn is a consistent estimator of θ and γ(θ)
a continuous function of θ, then γ(Tn ) is a consistent estimator
of γ(θ).
Proof. We have P [|Tn − θ| < ] → 1 as n → ∞.
Since γ is a continuous function, for every > 0 however small,
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
14
there exists a positive number 1 such that |γ(Tn ) − γ(θ)| < 1
whenever |Tn − θ| < . That is,
|Tn − θ| < ⇒ |γ(Tn ) − γ(θ)| < 1 .
For any 2 sets, A and B, if A ⇒ B, then A ⊂ B; which implies
P (A) ≤ P (B). Therefore,
P [|γ(Tn ) − γ(θ)| < 1 ] ≥ P [|Tn − θ| < ] → 1
P
n → ∞. That is, γ(Tn ) → γ(θ). Hence, γ(Tn ) is a consistent
estimator of γ(θ).
Result 1.3.1. If Tn is consistent for θ, then Tn + na , a 6= 0
is consistent for θ. Therefore, consistent estimator need not be
unique.
Efficiency
We know that
M.S.E.θ (T ) = V (T ) + (Bias)2 .
So if we have to choose one among several unbiased estimators
of a given parameter, we use the one whose sampling distribu-
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
15
tion has the smallest variance. Let T1 and T2 be two unbiased
estimators of a given parameter θ. If V (T1 ) is less than V (T2 ),
T1 is said to be more efficient than T2 .
Definition 1.3.3. Let T1 and T2 be two unbiased estimators of
a given parameter θ. Suppose that E(T12 ) < ∞ and E(T22 ) < ∞.
We define the efficiency of T1 relative to T2 by
Vθ (T2 )
R.E.θ (T1 |T2 ) = .
Vθ (T1 )
Note:
1. If R.E.θ (T1 |T2 ) > 1 then T1 is more efficient that T2 .
2. If R.E.θ (T1 |T2 ) = 1 then T1 and T2 are equally efficient.
3. If R.E.θ (T1 |T2 ) < 1 then T2 is more efficient that T1 .
Problems
1. P.T. in estimating the mean of a normal population, sam-
ple mean is more efficient than the sample median.
2. X1 , X2 , X3 are 3 independent observations from a popula-
tion with mean µ and variance σ 2 . If T1 = X1 + X2 − X3
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
16
and T2 = 2X1 + 3X2 − 4X3 are two estimators of µ, check
if T1 and T2 are unbiased estimators of µ and compare
their efficiency.
3. S.T. for a sample of size n drawn from a normal pop-
ulation with mean µ and variance σ 2 the statistic µ̃ =
1
Pn
n+1 i=1
Xi is the most efficient for estimating µ even
though it is biased.
4. Give examples of estimators which are
(a) Unbiased and efficient
(b) Unbiased and inefficient
(c) Biased and inefficient
Sufficiency
An estimator T is said to be sufficient for the parameter θ, if it
provides all the information contained in the sample in respect
of estimating θ. Thus, if we know the value of the sufficient
statistic, then there is no need to know the individual sample
values.
Definition 1.3.4. Let X1 , X2 , . . . , Xn be a r.s. from a popula-
tion whose probability distribution is f (x) and θ be the parameter
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
17
to be estimated. Let T = T (X1 , X2 , ...Xn ) be an estimator used
to estimate θ. The estimator T is said to be sufficient for θ
iff the conditional distribution of (X1 , X2 , . . . , Xn ) given T does
not depend on θ, provided the conditional distribution exists.
Remark 1.3.1. Sample is always sufficient for θ. i.e.,
(X1 , X2 , . . . , Xn ) is always sufficient for θ.
Remark 1.3.2. If T and Θ are vectors, we say that T is jointly
sufficient for Θ.
Note: The sample mean x̄ satisfies all the above four prop-
erties when it is being used as an estimator for estimating the
population mean θ.
Problems
1. A r.s. of size 2 is taken from B(1, θ). S.T. X1 + X2 is
sufficient for θ.
2. If X1 and X2 are two i.i.d. P (λ) r.v.s. S.T. X1 + X2 is
sufficient for λ.
Pn
3. Let X1 , X2 , . . . , Xn be i.i.d. B(1, p) r.v.s. S.T. i=1 Xi
and X̄ are sufficient for p.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
18
Theorem 1.3.3 (Fisher - Neymann Factorisation
Criterion or Theorem). Let X1 , X2 , . . . , Xn be discrete
or continuous r.v.s with joint probability density function
f (x1 , x2 , . . . , xn ; θ), θ ∈ Θ. Then T (X1 , X2 , . . . , Xn ) is
sufficient for θ iff we can write
f (x1 , x2 , . . . , xn ; θ) = h(x1 , x2 , . . . , xn ) gθ (T (X1 , X2 , . . . , Xn ))
where h is a non-negative function of x1 , x2 , . . . , xn only and
does not depend on θ and gθ is a non-negative non-constant
function of θ and T only.
That is, the density f can be factored into a product such
that one factor ‘h’, does not depend on θ and the other factor
‘g’, which does depend on θ, depends on X1 , X2 , . . . , Xn only
through T (X1 , X2 , . . . , Xn ).
Remark 1.3.3. Factorisation theorem leads to a sufficient
statistic, if a sufficient statistic exists.
Remark 1.3.4. The theorem cannot be used to show that a
given statistic T is not sufficient for θ. For this one would have
to use the definition of sufficiency.
Remark 1.3.5. If T is sufficient for θ, any 1-1 function of T
is also sufficient for θ.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
19
Note: The statistic T and the parameter θ may be multidi-
mensional.
Problems
1. A r.s. of size 2 is taken from B(1, θ). S.T. X1 + X2 is
sufficient for θ using factorisation theorem.
2. If X1 and X2 are two i.i.d. P (λ) r.v.s. S.T. X1 + X2 is
sufficient for λ using factorisation theorem.
Pn
3. Let X1 , X2 , . . . , Xn be i.i.d. B(1, p) r.v.s. S.T. i=1 Xi
and X̄ are sufficient for p using factorisation criterion.
Minimal Sufficiency
There may be more than one sufficient statistic for a parameter.
These are all reduced forms of the sample value. The maximum
possible form of reduction of sample values without loss of in-
formation will lead to minimal sufficient statistic. That is, it
cannot be reduced further without loss of information. A statis-
tic T is said to be minimal sufficient, if it can be expressed as a
function of every other sufficient statistic.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
20
Completeness
Definition 1.3.5. Let X1 , X2 , . . . , Xn be a r.s. from a popu-
lation with probability distribution f (x; θ). A statistic T (not
necessarily sufficient) is said to be complete iff
E[g(T )] = 0 ⇒ g(T ) = 0.
It is possible for a complete statistic to provide no informa-
tion at all about θ (ancillary statistic). In order for complete
statistics to be useful, they must also be a sufficient statistic. A
sufficient statistic summarizes all of the information in a sample
about a chosen parameter. Ideally then, a statistic should be
complete and sufficient, which means that: The statistic isn’t
missing any information about θ and doesn’t provide any irrel-
evant information.
Likelihood Function
A random variable X is connected to the possible parameter θ by
means of a function f (x; θ). For any given θ, f (x; θ) is intended
to be the probability (or probability density) of X. For any
given x, on the other hand, f (x; θ) can be viewed as a function
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
21
of θ. This view f (x; θ) is called the ”likelihood function”. It has
a nice property of being continuously differentiable twice. But
the area under curve of likelihood function does not have to add
up to 1.
The likelihood function of n r.v.s X1 , X2 , . . . , Xn is de-
fined to be the joint probability density function of n r.v.s
f (x1 , x2 , . . . , xn ; θ) which is considered to be a function of θ.
Let X be a r.v. and X1 , X2 , . . . , Xn be a r.s. of X having the
density functionf (x; θ). Also x1 , x2 , . . . , xn are observed sample
values. Then the likelihood function is defined by
L(θ; x1 , x2 , . . . , xn ) = f (x1 , x2 , . . . , xn ; θ)
= f (x1 ; θ) f (x2 ; θ) . . . f (xn ; θ)
Yn
= f (xi ; θ)
i=1
Likelihood function is also denoted by L(X; θ) or L(θ). It gives
the likelihood that
Minimum Variance Unbiased Estimator
The unbiased estimator which have the least variance is called
minimum variance unbiased estimator (M.V.U.E.). Let A, B, C
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
22
and D be four estimators of the parameter θ whose distributions
are as in figure below.
We can see that A and C are unbiased and B and D are
biased, V (A) > V (C) and V (D) = 0. Event hough D has the
least variance, it is biased. An estimator which has the least
variance is useless if it is not unbiased. So having just minimum
variance without unbiasedness is not an acceptable choice.
An unbiased estimator with least variance is most efficient.
A natural question arises here is which is the unbiased estima-
tor for a particular parameter with least variance? Is there any
lower limit for the variance of an unbiased estimator for a par-
ticular parameter etc. But it has been proved by Cramer and
Rao that there is a lower limit for the sampling variance of any
unbiased estimator. This is given by the Cramer-Rao inequality.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
23
Cramer-Rao Inequality
Let X be a random variable with density function f (x; θ). Let
T be an unbiased estimator of some function of θ (ψ(θ)). Let
1. the limit of integration be independent of θ
2. differentiation under the integral sign or summation sign,
according as the variable is continuous or discrete, is valid.
Then,
[ψ 0 (θ)]2
V (T ) ≥ ,
I(θ)
where I(θ) is called the Fisher information about θ contained in
the sample given by
2
∂ ln L
I(θ) = E
∂θ
2
∂ ln L
= −E
∂θ2
2
∂ ln f
= nE
∂θ
2
∂ ln f
= −n E
∂θ2
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
24
Minimum Variance Bound Estimator
When Cramer-Rao inequality becomes an equality, i.e.,
[ψ 0 (θ)]2
V (T ) = = C.R.L.B.,
I(θ)
the Cramer-Rao lower bound , then T is called the minimum
variance bound unbiased estimator (M.V.B.U.E.). Also called
minimum variance bound estimator (M.V.B.E.). In this case,
∂ ln L
= A(θ)[T − ψ(θ)])
∂θ
ψ 0 (θ)
and V (T ) = A(θ) .
M.V.U.E. of a parameter need not be M.V.B.E. of that pa-
rameter. Since it is not necessary that the minimum value of
the variance attained by unbiased estimators of a parameter is
the C.R.L.B.
Problems
1. S.T. the sample mean X̄ is the minimum variance bound
estimator of µ when a r.s. of size n is taken from N (µ, σ 2 ),
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
25
where σ is known.
2. S.T. for P (λ), sample mean X̄ is the M.V.B.E. of λ. Also
obtain the C.R.L.B.
Methods of Estimation
Appropriate statistics are to be identified for estimating the pa-
rameters. There are different methods for estimating param-
eters. The ones discussed here are the method of maximum
likelihood and methods of moments.
Method of Moments
This is the oldest method of estimation introduced by Karl Pear-
son. To estimate the k parameters of a population, we equate
the first k moments of the population to the first k moments of
the sample. Solving these k equations we get the k estimators.
That is, by equating µ0r = m0r , for r = 1, 2, . . . , k and solving,
Pn
where µ0r = E(X r ) and m0r = n1 i=1 Xir .
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
26
Properties
1. Moment estimators are asymptotically unbiased.
2. Moment estimators are consistent.
3. Under some general conditions, the distribution of mo-
ment estimators are asymptotically normal.
Problems
1. Estimate λ by the method of moments for a P (λ) distri-
bution.
2. Estimate µ and σ 2 by the method of moments for a
N (µ, σ 2 ) distribution.
3. Estimate p by the method of moments in sampling from
a B(N, p) population.
4. Estimate λ by the method of moments for a f (x; λ, α) =
αλ λ−1 −αx
Γ(λ) x e , x > 0.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
27
Method of Maximum Likelihood
Likelihood function gives the likelihood that the r.v. assumes
a particular value x1 , x2 , . . . , xn . The principle of maximum
likelihood estimation consists in finding the estimator of the
parameter which maximises L(θ). Thus, if there exists a statistic
T which maximises the L(θ) for variations in θ, then it is called
maximum likelihood estimator (M.L.E.). That is, the value of
θ, as a function of x1 , x2 , . . . , xn that maximises L(θ) will be the
M.L.E. This can be obtained by solving
∂L ∂2L
= 0 and if for that value of θ, < 0.
∂θ ∂θ2
The value of θ that maximises L(θ) will also maximise ln L(θ).
So M.L.E. can also be obtained by solving
∂ ln L ∂ 2 ln L
= 0 and if for that value of θ, < 0.
∂θ ∂θ2
Properties
1. Maximum likelihood estimators are consistent.
2. Maximum likelihood estimators are most efficient.
3. Maximum likelihood estimators are sufficient, if sufficient
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
28
statistic exists.
4. Maximum likelihood estimators are asymptotically nor-
mal.
5. Maximum likelihood estimators are not necessarily unbi-
ased.
6. Maximum likelihood estimators are invariant under 1-1
onto transformations.
Problems
1. Estimate λ by the method of maximum likelihood for a
P (λ) distribution.
2. Estimate M.L.E. for random sampling from N (µ, σ 2 ) pop-
ulation
(a) for µ when σ 2 is known.
(b) for σ 2 when µ is known.
(c) for µ and σ 2 simultaneously.
3. Find the M.L.E. of p in sampling from a B(N, p) popula-
tion.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
29
4. Find the M.L.E. of α if f (x; α) = αe−αx , x > 0.
5. Find the M.L.E. of θ for a U (0, θ) population.
Method of Minimum Variance
Minimum-variance unbiased estimator (MVUE) or uniformly
minimum-variance unbiased estimator (UMVUE) is an unbiased
estimator that has lower variance than any other unbiased esti-
mator for all possible values of the parameter. In this method
we choose those estimators which are unbiased and have mini-
mum variance. That is, if Tn is the estimator for θ, we should
have E(Tn ) = θ and variance of Tn is minimum. If a complete
sufficient statistic exists, the UMVUE will be its function.
Method of Least Squares
In method of least squares we are finding a function of sample
values X1 , X2 , . . . , Xn as an estimator of an unknown parameter
in such a way that the sum of squares of the differences between
the values of the variable and the corresponding expected values
become minimum. This method is mostly used in estimating
the parameters of a linear function and will lead to beast linear
unbiased estimators (BLUE).
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
30
Consider a r.s. X1 , X2 , . . . , Xn , of size n taken from
N (µ, σ 2 ). We know E(Xi ) = µ for all i. In this method we find
an estimator of µ such that (Xi − E(Xi ))2 = (Xi − µ)2
P P
is minimum. Using differential calculus we get X̄ as the least
square estimator of µ.
1.4 Interval Estimation
In point estimation, sample data is used to obtain a single num-
ber as an estimate of the unknown parameter. This value is
expectantly close to the unknown parameter. But, in general
the estimate thus obtained will not coincide with the true value
of the parameter. In contrast to point estimation, in interval es-
timation, sample data is used to obtain an interval within which
the parameter is expected to lie with a certain degree of con-
fidence. Hence, this interval is known as confidence interval or
fiducial interval.
Suppose we want to obtain the confidence interval for some
parameter θ of the population distribution. Here, the problem
is to determine two constants t1 and t2 such that P (t1 < θ <
t2 ) = 1 − α. The values of t1 and t2 depends on the sample
data. Hence, their values will be different for different samples.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
31
In other words, the interval [t1 , t2 ] will be random interval. Some
of these intervals will contain the true value of θ while the others
won’t. The confidence coefficient (1−α) assertion that, in a long
series of independent sample of the same size, the frequency of
cases in which this interval will contain the true value of the
parameter will be approximately 1 − α. Accordingly, confidence
interval is also known as 100(1 − α)% confidence interval.
To give an example, if α = 0.01, the 99% confidence interval
means that, if 100 samples of the same size are considered to
construct 100 confidence intervals, 99 of them will contain the
true value of the parameter.
Confidence Interval for Mean
Let X1 , X2 , . . . , Xn be a r.s. of size n from N (µ, σ 2 ).
Case I : σ is known
X̄−µ
When σ is known Z = √σ ∼ N (0, 1). Then, from the area
n
property of standard normal distribution, the 100(1-α)% C.I.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
32
for µ is given by
h i
P |Z| ≤ z α = 1 − α
2
h i
i.e., P −z α ≤ Z ≤ z α = 1 − α
2 2
" #
X̄ − µ
i.e., P −z α ≤ ≤ zα =1−α
2 √σ 2
n
σ σ
i.e., P −z α √ ≤ X̄ − µ ≤ z α √ =1−α
2 n 2 n
σ σ
i.e., P −X̄ − z α √ ≤ −µ ≤ −X̄ + z α √ =1−α
2 n 2 n
σ σ
i.e., P X̄ − z α √ ≤ µ ≤ X̄ + z α √ =1−α
2 n 2 n
Therefore, the 100(1-α)% C.I. for µ is
σ σ
X̄ − z α √ , X̄ + z α √ .
2 n 2 n
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
33
Case II : σ is unknown and n is large
When σ is unknown and n is large Z = X̄−µ
√S ∼ N (0, 1), where
n
2 1
Pn 2
S = n−1 i=1 (Xi − X̄) . Proceeding as above we get the
100(1-α)% C.I. for µ as
S S
X̄ − z α √ , X̄ + z α √ .
2 n 2 n
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
34
Case III : σ is unknown and n is small
X̄−µ
When σ is unknown and n is small S
√
∼ tn−1 . Now, by the
n
symmetry of t - distribution
h i
P |t| ≤ tn−1, α = 1 − α
2
h i
i.e., P −tn−1, α ≤ t ≤ tn−1, α = 1 − α
2 2
" #
X̄ − µ
i.e., P −tn−1, α ≤ ≤ tn−1, α =1−α
2 √S 2
n
S S
i.e., P −tn−1, α √ ≤ X̄ − µ ≤ tn−1, α √ =1−α
2 n 2 n
S S
i.e., P −X̄ − tn−1, α √ ≤ −µ ≤ −X̄ + tn−1, α √ =1−α
2 n 2 n
S S
i.e., P X̄ − tn−1, α √ ≤ µ ≤ X̄ + tn−1, α √ =1−α
2 n 2 n
Therefore, the 100(1-α)% C.I. for µ is
S S
X̄ − tn−1, α √ , X̄ + tn−1, α √ .
2 n 2 n
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
35
Note:
1. Let X̄ be the sample mean of a sample of size n from a
population which is not normal whose mean is µ and
variance σ 2 . Then by CLT,
σ2
X̄ → N µ, as n → ∞.
n
h i
Then the 100(1-α)% C.I. for µ is X̄ − z α √σ , X̄ + zα √σ .
2 n 2 n
2. Here z α is obtained from the normal table in such a way
2
that the area under the normal curve to its right is equal
to α2 .
3. If α = 0.05, z α = 1.96, therefore, when σ is known, 95%
h2 i
C.I. for µ is X̄ − 1.96 √σn , X̄ + 1.96 √σn .
4. If α = 0.02, z α = 2.326, therefore, when σ is known, 98%
h2 i
C.I. for µ is X̄ − 2.326 √σn , X̄ + 2.326 √σn .
5. If α = 0.01, z α = 2.58, therefore, when σ is known, 99%
h2 i
C.I. for µ is X̄ − 2.58 √σn , X̄ + 2.58 √σn .
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
36
Problems
1. Let X1 , X2 , . . . , Xn be a r.s. of size 9 from N (µ, σ 2 ). Ob-
tain the 97% C.I. for µ if X̄ = 3.2 and σ 2 = 2.
2. Let X̄ = 80 be the sample mean of a r.s. of size 100 from
N (µ, σ 2 ). Obtain the 96% C.I. for µ and σ = 2.
3. Let X1 , X2 , . . . , Xn be a r.s. of size 34 from N (µ, σ 2 ).
Obtain the 99% C.I. for µ if X̄ = 43 and S = 3.
4. Let X̄ = −2 be the sample mean of a r.s. of size 50 from
N (µ, σ 2 ). Obtain the 96% C.I. for µ and S 2 = 4.
Confidence Interval for the Difference of Means
Let X̄1 be the sample mean of a r.s. of size n1 from N (µ1 , σ12 )
and X̄2 be the sample mean of a r.s. of size n2 from N (µ2 , σ22 ).
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
37
Case I : σ1 and σ2 are known
σ2
When σ1 and σ2 are known, X̄1 ∼ N µ1 , 1
n1 and X̄2 ∼
σ22
N µ2 , n . Therefore,
2
σ12 σ2
X̄1 − X̄2 ∼ N µ1 − µ2 , + 2
n1 n2
and
(X̄1 − X̄2 ) − (µ1 − µ2 )
Z= r ∼ N (0, 1).
σ12 σ22
n + n
1 2
Then, from the area property of standard normal distribution,
the 100(1-α)% C.I. for µ1 − µ2 is given by
h i
P |Z| ≤ z α = 1 − α
2
h i
i.e., P −z α ≤ Z ≤ z α = 1 − α
2 2
−z α ≤ (X̄1 −r
X̄2 ) − (µ1 − µ2 )
i.e., P 2 ≤ zα
=1−α
σ 2 σ 2 2
n
1
+ n
2
1 2
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
38
" s
σ12 σ2
i.e., P −z α + 2 ≤ (X̄1 − X̄2 ) − (µ1 − µ2 )
2 n1 n2
s #
σ12 σ2
≤ zα + 2 =1−α
2 n1 n2
" s
σ12 σ2
i.e., P −(X̄1 − X̄2 ) − z α + 2 ≤ −(µ1 − µ2 )
2 n1 n2
s #
σ12 σ2
≤ −(X̄1 − X̄2 ) + z α + 2 =1−α
2 n1 n2
" s
σ12 σ2
i.e., P (X̄1 − X̄2 ) − z α + 2 ≤ (µ1 − µ2 )
2 n1 n2
s #
σ12 σ2
≤ (X̄1 − X̄2 ) + z α + 2 =1−α
2 n1 n2
Therefore, the 100(1-α)% C.I. for (µ1 − µ2 ) is
" s s #
σ12 σ2 σ12 σ22
(X̄1 − X̄2 ) − z α + 2 , (X̄1 − X̄2 ) + z α + .
2 n1 n2 2 n1 n2
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
39
Case II : σ1 and σ2 are unknown and n1 and n2
are large
When σ1 and σ2 are unknown and n1 and n2 are large
(X̄1 − X̄2 ) − (µ1 − µ2 )
Z= r ∼ N (0, 1).
S2 S2
n + n
1 2
1 2
Proceeding as above we get the 100(1-α)% C.I. for µ as
" s s #
S12 S2 S12 S22
(X̄1 − X̄2 ) − z α + 2 , (X̄1 − X̄2 ) + z α + .
2 n1 n2 2 n1 n2
Case III : σ1 = σ2 = σ is unknown and n1 and n2
are small
When σ1 = σ2 = σ is unknown and n1 and n2 are small
(X̄1 − X̄2 ) − (µ1 − µ2 )
r
(n1 −1)S12 +(n2 −1)S22
∼ t(n1 +n2 −2) .
1 1
n1 +n2 −2 n + 1
n2
Let
(n1 − 1)S12 + (n2 − 1)S22 ∗2
=σ .
n1 + n2 − 2
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
40
Then
(X̄1 − X̄2 ) − (µ1 − µ2 )
r ∼ t(n +n −2) .
∗2 1 2
σ n1 + n1
1 2
Proceeding as above the 100(1-α)% C.I. for µ1 − µ2 is given by
s
∗ 1 1
(X̄1 − X̄2 ) ± tn +n −2, α
σ + .
1 2 2 n1 n2
Confidence Interval for Variance
Let X1 , X2 , . . . , Xn be a r.s. of size n from N (µ, σ 2 ).
Case I : µ is unknown
When µ is unknown,
(n − 1)S 2
∼ χ2(n−1) .
σ2
By refereing χ2 - table we can find χ2(n−1),1− α and χ2(n−1), α such
2 2
that
h i
P χ2(n−1),1− α ≤ χ2 ≤ χ2(n−1), α = 1 − α
2 2
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
41
(n − 1)S 2
2 2
i.e., P χ(n−1),1− α ≤ ≤ χ(n−1), α = 1 − α
2 σ2 2
χ2(n−1),1− α χ2(n−1), α
" #
1
i.e., P ≤ 2 ≤
2 2
=1−α
(n − 1)S 2 σ (n − 1)S 2
" #
(n − 1)S 2 2 (n − 1)S 2
i.e., P ≥ σ ≥ =1−α
χ2(n−1),1− α χ2(n−1), α
2 2
" #
(n − 1)S 2 2 (n − 1)S 2
i.e., P ≤ σ ≤ =1−α
χ2(n−1), α χ2(n−1),1− α
2 2
Therefore, the 100(1-α)% C.I. for σ 2 is
" #
(n − 1)S 2 (n − 1)S 2
,
χ2(n−1), α χ2(n−1),1− α
2 2
Case II : µ is known
When µ is known,
Pn
(Xi − µ)2
i=1
∼ χ2(n) .
σ2
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
42
Proceeding as above we get the 100(1-α)% C.I. for σ 2 as
" Pn Pn #
i=1
(Xi − µ)2 i=1
(Xi − µ)2
, .
χ2n, α χ2n,1− α
2 2
Confidence Interval for Proportion of Success
Population proportion is the fraction of the population that has
certain characteristic. Suppose that a population consists of N
units of which X possesses a particular characteristic. Then
p= X N is the fraction or proportion of units having that char-
acteristic. Clearly, X ∼ B(N, p). In real life, we usually don’t
know facts about the entire population. So we use sample pro-
portion to estimate p. This sample proportion is denoted by
p0 = nx , where n is the sample size (or the number of trials) and
x is the number of units possessing the characteristic specified
within the sample.
Now, as n increases and when neither p nor q = 1 − p is too
small, p0 = nx → N p, pq
n . Therefore,
p0 − p
Z=q → N (0, 1)
p0 q 0
n
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
43
where p0 is the point estimate of p. By refereing to normal table
we can find z α such that
2
h i
P |Z| ≤ z α = 1 − α
2
h i
i.e., P −z α ≤ Z ≤ z α = 1 − α
2 2
−z α p0 − p
i.e., P ≤q ≤ zα = 1 − α
2 p0 q 0 2
n
" r r #
p0 q 0 p0 q 0
i.e., P −z α ≤ p0 − p ≤ z α =1−α
2 n 2 n
" r r #
0 p0 q 0 pq
i.e., P −p − z α ≤ −p ≤ p0 + z α =1−α
2 n 2 n
" r r #
0 p0 q 0 p0 q 0
i.e., P p − zα ≤ p ≤ p0 + z α =1−α
2 n 2 n
Therefore, the 100(1-α)% C.I. for p is
" r r #
0 p0 q 0 0 p0 q 0
p − zα , p + zα .
2 n 2 n
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
44
Note:
1. If α = 0.05, z α = 1.96, therefore 95% C.I. for p is
2
q q
0 p0 q 0 0 p0 q 0
p − 1.96 n , p + 1.96 n .
2. If α = 0.02, z α = 2.326, therefore 98% C.I. for p is
q 2 q
0 0 0 0
p − 2.326 pnq , p0 + 2.326 pnq .
0
3. If α = 0.01, z α = 2.58, therefore 99% C.I. for p is
2
q q
0 p0 q 0 0 p0 q 0
p − 2.58 n , p + 2.58 n .
Confidence Interval for Difference of Propor-
tions of Successes
x1
Let p1 = n1 be the proportion of success of a r.s. of size n1
x
from B(N1 , p1 ) and p2 = n2 be the proportion of success of
2
a r.s. of size n2 from B(N2 , p2 ). Now, as becomes n1 and n2
becomes large and when p1 and p2 areneither too small nor too
p q p q
big p01 − p02 ∼ N p1 − p2 , n1 1 + n2 2 . Hence,
1 2
(p01 − p02 ) − (p1 − p2 )
Z= r → N (0, 1),
p01 q10 p02 q20
n + n
1 2
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
45
where p01 and p02 are the point estimate of p1 and p1 respectively.
By refereing to normal table we can find z α such that
2
h i
P |Z| ≤ z α = 1 − α
2
h i
i.e., P −z α ≤ Z ≤ z α = 1 − α
2 2
0 0
− p2 ) − (p1 − p2 )
−z α ≤ (p1 r
i.e., P ≤ zα
=1−α
2 0 0
p1 q1 0 0
p2 q2 2
n + n
1 2
" s
p01 q10 p0 q 0
i.e., P −z α + 2 2 ≤ (p01 − p02 ) − (p1 − p2 )
2 n1 n2
s #
p01 q10 p0 q 0
≤ zα + 2 2 =1−α
2 n1 n2
" s
p01 q10 p0 q 0
i.e., P −(p01 − p02 ) − z α + 2 2 ≤ −(p1 − p2 )
2 n1 n2
s #
0 0 p01 q10 p0 q 0
≤ −(p1 − p2 ) + z α + 2 2
2 n1 n2
=1−α
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
46
" s
p01 q10 p0 q 0
i.e., P (p01 − p02 ) + z α + 2 2 ≥ (p1 − p2 )
2 n1 n2
s #
0 0 p01 q10 p0 q 0
≥ (p1 − p2 ) − z α + 2 2
2 n1 n2
=1−α
" s
p01 q10 p0 q 0
i.e., P (p01 − p02 ) − z α + 2 2 ≤ (p1 − p2 )
2 n1 n2
s #
0 0 p01 q10 p0 q 0
≤ (p1 − p2 ) + z α + 2 2
2 n1 n2
=1−α
Therefore, the 100(1-α)% C.I. for p1 − p2 is
" s s #
0 0 p01 q10 p0 q 0 p01 q10 p02 q20
(p1 − p2 ) − z α + 2 2 , (p01 − p02 ) + z α + .
2 n1 n2 2 n1 n2
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
47
Confidence Interval for Correlation Co-
efficient
Obtaining the confidence interval for Pearson’s r is a bit com-
plicated as r does not follow the normal distribution. It has a
negatively skewed distribution. Fisher’s Z-transform, by Prof.
R. A. Fisher convert r into a distribution that is approximately
normally distributed. That is, this transformation given by
1+r
Zr = 0.5 ln
1−r
is approximately normally distributed with mean µr =
1+ρ 1
0.5 ln 1−ρ and standard error σr = √n−3 . Therefore,
Zr − µr
Z= ∼ N (0, 1)
σr
Then, from the area property of standard normal distribution,
the 100(1-α)% C.I. for r is given by
h i
P |Z| ≤ z α = 1 − α
2
h i
i.e., P −z α ≤ Z ≤ z α = 1 − α
2 2
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
48
Zr − µr
i.e., P −z α ≤ ≤ zα = 1 − α
2 σr 2
h i
i.e., P −z α σr ≤ Zr − µr ≤ z α σr = 1 − α
2 2
h i
i.e., P −Zr − z α σr ≤ −µr ≤ −Zr + z α σr = 1 − α
2 2
h i
i.e., P Zr − z α σr ≤ µr ≤ Zr + z α σr = 1 − α
2 2
1+ρ
i.e., P Zr − z α σr ≤ 0.5 ln ≤ Zr + z α σr = 1 − α
2 1−ρ 2
1+ρ
i.e., P a ≤ 0.5 ln ≤b =1−α
1−ρ
where, a = Zr − z α and b = Zr + z α . Therefore, on simplifica-
2 2
tion we get,
e2a − 1 e2b − 1
P 2a ≤ ρ ≤ 2b =1−α
e −1 e −1
Hence, the 100(1-α)% C.I. for ρ is given by
h i
Zr − z α σr , Zr + z α σr
2 2
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
MODULE
TWO
Testing of Hypothesis
In practice it is not possible for a researcher to observe each
and every individual in a population. So, he/she may collect a
sample data and use it to answer questions about the popula-
tion. Hypothesis testing is one such procedure to make infer-
ences about the population.
The theory of testing statistical hypothesis deals with a spe-
cial class of decision problems. Many important decisions made
49
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
50
in the face of uncertainty involve a choice between two alterna-
tives. For example, an agronomist may have to decide whether a
new variety of wheat developed by him will significantly improve
the yield or not. An industrial engineer may have to determine
whether the new production process will significantly reduce the
cost or not. The sales manager of a soft drinks manufacturing
company may have to choose between newspapers and television
for launching his advertisement campaign.
The persons entrusted with taking such decisions face the
risk of making a wrong choice. Statistical techniques can be used
to evaluate or quantify such risks, and, if possible, to provide
criteria for minimizing the chances of making wrong decisions.
Such techniques fall into an area of statistics called ‘Testing of
Hypothesis’.
2.1 Basic Concepts
Hypothesis
Hypothesis means a testable proposition or supposition. Impor-
tant consideration after the formulation of the research problem
is the construction of hypothesis.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
51
Statistical Hypothesis
Hypothesis about the nature of a statistical population is called
a statistical hypothesis. It can be either parametric or non-
parametric.
Parametric Hypothesis
Parametric hypothesis is an assertion about the numerical value
of an unknown parameter, here, we assume that the functional
form of the distribution is known except the parameter or prob-
ability distribution of the population. Example: Let X ∼ P (θ).
Then H: θ = 3 or H: θ > 3 are parametric hypothesis.
Non-parametric Hypothesis
Non-parametric hypothesis is an assertion which involves no
population parameter. Unlike the parametric hypothesis, here
no assumptions about the underlying distribution is made. Non-
parametric does not mean that there isn’t any parameter. It
simply means that specifying the value of the parameter does
not specifies the distribution. Example: H : The attributes
obesity and food habits are independent or H : Median of a
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
52
population is 10.
Simple and Composite Hypothesis
If a statistical hypothesis completely specifies the probability
distribution it is referred to as a simple hypothesis, otherwise
composite. Examples:
Distribution Simple Composite
P (λ) H: λ = 2.7 H: λ 6= 2.7 or
H: λ < 2.7 or
H: λ > 2.7
1
B(10, p) H: p = 2 H: p 6= 12 or
H: p < 12 or
H: p > 12
B(n, p) H: p = 12 , n = 8 H: p = 12
(since n is unknown,
specifying the value of
p alone will not specify
the distribution) or
H: p > 12 , n = 8
Hence, a simple hypothesis specifies the numerical values of
all unknown parameters in the probability distribution of inter-
est.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
53
Null Hypothesis
The hypothesis which is to be tested is known as the null hypoth-
esis, denoted by H0 . Usually, it is a statement of no difference.
Alternative Hypothesis
The hypothesis which we accept when we reject the null hypoth-
esis or we reject when we accept the null hypothesis is called an
alternative hypothesis, denoted by H1 or HA .
Test of Hypothesis
A rule or procedure which enables us to decide whether to accept
or reject the null hypothesis or to determine whether observed
samples differ significantly from expected results are called tests
of hypothesis or tests of significance. It assesses the evidence
provided by data about some claim concerning a population.
Since the sample values x1 , x2 , . . . , xn can be taken as a point
in the n-dimensional space, we specify some region in the n-
dimensional space and see whether this point lies within this
region or not. Hence, tests of hypothesis partition the sample
space into two disjoint sets C and C 0 such that if the sample
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
54
belongs to C we reject H0 and if it belongs to C 0 we accept H0 .
Critical Region
The basis of testing of hypothesis is the partition of the sample
space into two exclusive regions, namely, the region of accep-
tance of H0 and a region of rejection of H0 . This subset C of
the sample space such that if the sample belongs to C, then H0
rejected is called critical region. In other words, set of all values
of the test statistic for which H0 rejected is called critical region.
It is denoted by ω.
Acceptance Region
The complement of the critical region in the sample space con-
stitutes the acceptance region. That is, the values of the test
statistic for which H0 is accepted is called the acceptance region.
Test Statistic
The statistic based on whose value the null hypothesis is rejected
or accepted is called the test statistic or test criterion, denoted
by ∆.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
55
In analogy to the above discussion we can say that, a test
partition the possible values of the test statistic into two parts.
One corresponding to critical region and the other corresponding
to acceptance region. The set of all values of the test statistic for
which H0 rejected is called critical region and the values of the
test statistic for which H0 is accepted is called the acceptance
region.
A test of statistical hypothesis is a two-action decision prob-
lem. The actions being “rejecting the null hypothesis” or “ac-
cepting the null hypothesis”. The decision is made based on the
sample data.
Type I and Type II Errors
Since the decision to accept or reject H0 will be made based on
data derived from some random process, it is possible that an
incorrect decision will be made. There are two types of errors
that can be made if we use a test. Rejecting H0 when it is true
and accepting H0 when it is false.
True State
Decision
H0 H1
Reject H0 Type I Error Correct Decision
Accept H0 Correct Decision Type II Error
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
56
Probability of Type I Error
Rejection of the null hypothesis when it is true is called type I
error. The probability of committing type I error is denoted by
α. That is,
α = P (Type I error) = P (Rej H0 |H0 is true).
Probability of Type II Error
Acceptance of the null hypothesis when it is false is called type
II error. The probability of committing type II error is denoted
by β. That is,
β = P (Type II error) = P (Acc H0 |H1 is true).
It would be ideal if we could find a test that simultaneously
minimises both α and β. But this is not possible unless we are
able to make the number of observations as large as we please
and in practice we are seldom able to do this. Since, α and
β are probabilities we know that α ≥ 0 and β ≥ 0. Suppose
no matter what H0 and H1 states and what observed values
occur in the sample, we use the test: accept H0 . With this test
we could never commit type I error. For this test, α = 0 and
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
57
β = 1. Similarly, for the converse of this test: reject H0 , α = 1
and β = 0. Neither of these tests is desirable, because they
maximise one of the two probabilities while minimising other.
Hence, the procedure used in practice is to limit probability of
type I error to a pre-assigned level α that is small (usually 0.01
or 0.05) and to minimise the probability of type II error.
Level of Significance
The fixed level of the probability of type I error is called the level
of significance. The significance level, denoted as α, is a measure
of the strength of the evidence that must be present in your
sample before you will reject the null hypothesis and conclude
that the effect is statistically significant. The significance level
or level of significance is the probability of rejecting the null
hypothesis when it is true. For example, a significance level of
0.05 indicates a 5% risk of concluding that a difference exists
when there is no actual difference. Lower significance levels
indicate that you require stronger evidence before you will reject
the null hypothesis. It is also called level of the test or size of
the test or size of the critical region or producer’s risk. The
researcher determines the significance level before conducting
the experiment.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
58
Best test
Among all the tests whose size is less than or equal to α, the
one for which P(Type II error) is minimum is called the best
test.
Power of a Test
The probability of rejecting the null hypothesis H0 when the
alternative hypothesis H1 is true is called power of the test or
power of the critical region. i.e.,
Power = P(rejecting H0 when it is not true)
= P(rejecting H0 when H1 is true)
= P(rejecting H0 |H1 )
= 1 − P(accepting H0 |H1 )
= 1 − P(type II error)
= 1−β
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
59
The larger the value of 1 − β for fixed α, the better is the test
in general. Sensitiveness of a test is its ability to ascertain the
correctness of the alternative hypothesis when it is true for fixed
α. Thus, power is a measure of sensitiveness of the test.
Most Powerful Test
Among all the tests whose size is less than or equal to α, the
one for which β is maximum is called the most powerful test.
Critical Value
The value of the test statistic which separates the critical re-
gion and acceptance region is called critical value. It is usually
referred as ∆α (Zα , tα etc.) depending on the sampling distri-
bution of the test statistic and the level of significance used. It
is also called the tolerance limit.
p-value
Traditionally the value of α is fixed as the maximum allowable
value and then proceed to test the null hypothesis in terms of the
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
60
test statistic. The pre specified value of α sets the correspond-
ing tolerance limit(s), i.e., critical value(s). If the calculated
value of the test statistic falls in the critical region correspond-
ing to the critical value(s), we reject H0 . But, the tolerance
limit varies with α. Hence the decision may also changes with
α. So, the question is which is the smallest value of α for which
H0 is rejected? The answer is P-value. P-value or probability
value is the smallest value of α for which H0 is rejected. That
is, for all alpha less than p, we accept the null hypothesis. It
is the observed level of significance corresponding to the ob-
served value of the test statistic. That is, it is the probability,
P (δ ∈ ω|H0 is true), where δ is the observed value of ∆ (test
statistic) and ω the critical region.
Steps Involved in Testing a Statistical Hypoth-
esis
1. State the null hypothesis H0 and alternative hypothesis
H1 .
2. Choose the level of significance α.
3. Determine the test statistic.
4. Determine the probability distribution of the test statistic.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
61
5. Determine the best critical region.
6. Calculate the value of the test statistic.
7. If the calculated value of the test statistic falls in the crit-
ical region reject the null hypothesis, otherwise accept it.
Neymann-Pearson Lemma
The best critical region or the most powerful critical region for
testing a simple H0 : θ = θ0 against the simple H1 : θ = θ1 ,
is given by the well known Neymann Pearson Lemma. That is,
the theorem specifies the critical region which has pre assigned
probability of type I error and a minimal probability of type II
error. This is the same as maximizing the power subject to the
condition that probability of type I error is a pre determined
constant (α).
Theorem 2.1.1. Let x1 , x2 , . . . , xn be n independent observa-
tions from the population f (x, θ). We shall denote the likelihood
of these observations when H0 is true by L(x, H0 ) and when
H1 is true by L(x, H1 ). Let H0 : θ = θ0 against H1 : θ = θ1
be the test to be performed. Then the BCR is that for which
kL(x, H1 ) ≥ L(x, H0 ) inside it, where k is positive and is to be
chosen such that the size of the critical region is α.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
62
Parametric Tests
2.2 Large Sample Tests
Standard test procedures are available to test the various hy-
potheses regarding the parameters of the populations. These
tests may be grouped into two.
i) large sample tests, and
ii) small sample tests.
For small sample tests, the exact sampling distribution of
the test statistic should be known.
In large sample tests, the normal distribution plays the key
role due to the central limit theorem. The theorem states that,
when the sample size is large, most of the statistics are normally
or approximately normally distributed.
Let Y be a statistic satisfying the conditions of centra limit
theorem. Then the statistic given by
Y − E(Y )
Z= p .
V (Y )
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
63
For large n, Z ∼ N (0, 1). If Z is chosen as the test statistic, the
critical region for a given significance level can be determined
from the standard normal table. The test based on normal
distribution is called ‘normal test’ or ‘Z test’.
As an illustration, consider a situation in which we want
to test the null hypothesis H0 : θ = θ0 against the two sided
alternative H1 : θ 6= θ0 . Let θ̂ be the estimate of θ. It appears
reasonable to accept the null hypothesis when θ̂ is close to θ0
and to reject it when θ̂ is much larger or much smaller the θ0 .
Hence it is logical to let the critical region consists of both tails
of the sampling distribution of the test statistic. Such a test is
referred to as a ‘two tailed test’ or a ‘two sided test’.
If we are testing the null hypothesis H0 : θ = θ0 against the
one sided alternative H1 : θ < θ0 , it would be reasonable to
reject H0 only if θ̂ is much smaller than θ0 . Since in this case
it would be logical to let the critical region consists of the left
hand tail of the sampling distribution of the test statistic.
Likewise, if we are testing the null hypothesis H0 : θ =
θ0 against the one sided alternative H1 : θ > θ0 , it would be
reasonable to reject H0 only if θ̂ is much larger than θ0 . Since
in this case it would be logical to let the critical region consists
of the right hand tail of the sampling distribution of the test
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
64
statistic.
Any test where the critical region consist only one tail of
the sampling distribution of the test statistic is called a ‘one
tailed test’, particularly they are called ‘left tailed test’ and
‘right tailed test’ respectively.
Best critical regions of normal test or Z - test
For the significance level α , to test H0 : θ = θ0 , the possi-
ble alternatives and corresponding critical regions are as in the
following figure
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
65
and is summerised in the following table.
Alternative hypothesis Best Critical Region
H1 ω
θ < θ0 Z < −zα
θ > θ0 Z > zα
θ 6= θ0 |Z| ≥ zα/2
For example, when α = 0.05, the best critical regions are
as follows.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
66
Test concerning mean of a population
By testing the mean of a population, we are actually testing the
significant difference between the population mean and the sam-
ple mean. In other words, we are deciding whether the sample
is drawn from the population having the mean proposed by the
null hypothesis H0 . Here, the null hypothesis to be tested is
H0 : µ = µ0
against one of the possible alternatives
1. H1 : µ < µ0 or
2. H1 : µ > µ0 or
3. H1 : µ 6= µ0
On the basis of a random sample of size n from a normal pop-
ulation with known variance σ 2 . The test statistic is given by
x̄ − µ0
Z= √
σ/ n
For the significance level α, the best critical regions are respec-
tively,
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
67
1. Z < −Zα
2. Z > Zα
3. |Z| ≥ Zα/2
Using the sample data, calculate the value of the test statistic
Z . If it lies in the critical region, reject H0 , otherwise accept
H0 .
Remark 2.2.1. If the population is given to be normal, the test
procedure is valid even for small samples, provided σ is known
Remark 2.2.2. When σ is unknown and n is large, in the
calculation of the statistic, we may replace σ by its estimate S,
the sample S.D. That is,
x̄ − µ0
Z= √ .
S/ n
Example 2.2.1. A sample of 25 items were taken from a popu-
lation with S.D. 10 and the sample mean is found to be 65. Can
it be regarded as a sample from a normal population with mean
=60, at 5% level of significance?
Solution: Given, n = 25, σ = 10, x̄ = 65, µ0 = 60 and
α = 0, 05.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
68
The null hypothesis to be tested is H0 : µ = 60 against H1 : µ 6=
60.
Even though the sample is small (n < 30), σ is known.
Therefore, the test statistic is
x̄ − µ0
Z = √
σ/ n
65 − 60
= √ = 2.5
10/ 25
Since the alternative is of the form H1 : µ 6= 60, the best
critical region is ω : |Z| ≥ Zα/2 . Now, zα/2 = z0.025 = 1.96.
Therefore,
|Z| = |2.5| = 2.5 > 1.96 = zα/2
That is |Z| > zα/2 . Hence, H0 is rejected.
i.e., the sample can not be regarded as drawn from a normal
population with µ = 60.
Example 2.2.2. A new product was introduced in the market
claiming that it has an average life of 200 hours with a standard
deviation of 21 hours. This clain came under severe criticism
from dissatisfied customers. A customer group tested 49 items
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
69
and found that they have an average life of 191 hours. Is the
claim of the manufacturer justified?
Example 2.2.3. The mean life of a sample of 100 light bulbs
produced by a company is found to be 1570 hours, with a S.D.
of 120 hours. Test the hypothesis that the mean life time of
all the bulbs produced by the company is 1600 hours against the
alternative that it is not, at a level of significance of 0.05.
Example 2.2.4. A sample of 400 observations were taken from
a population with S.D. 15. If the mean of the sample is 27, test
whether the hypothesis that the mean of the population is equal
to 24 against that mean is less than 24. (α = 0.05)
Example 2.2.5. A random sample of 25 light bulbs has a mean
life of 2000 hours with a S.D. of 200 hours. Test at 1% level
that the bulbs are taken from a lot having mean 2100 hours.
Test concerning difference of means of two pop-
ulations
By testing the equality of means of two normal populations,
we are actually testing the significant difference between two
sample means. Here, we are checking if the samples have come
from two populations having the same mean. Here, we want to
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
70
test the null hypothesis
H0 : µ1 = µ2
against one of the alternatives
1. H1 : µ1 < µ2
2. H1 : µ1 > µ2
3. H1 : µ1 6= µ2 .
Based on independent random samples of sizes n1 and n2 from
two populations having the means µ1 and µ2 and the known
variances σ1 2 and σ2 2 , the test statistic is
x¯1 − x¯2
Z=q .
σ1 2 σ2 2
n1 + n2
For the significance level α, the best critical regions (BCR)
are respectively
1. Z < −Zα
2. Z > Zα
3. |Z| ≥ Zα/2
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
71
Calculate the value of the statistic Z using the sample informa-
tion and if it lies in the critical region reject H0 , or otherwise
accept H0 .
Remark 2.2.3. When we deal with independent random sam-
ples from populations with unknown variances which may not be
normal, we can still use the above test with S1 substituted for
σ1 and S2 substituted for σ2 provided n1 and n2 are large.
x¯1 − x¯2
Z=q .
S1 2 S2 2
n1 + n2
Example 2.2.6. Suppose that 64 students from college A and
81 students from college B had mean heights 68.2” and 67.3”
respectively. If the standard deviation for heights of all students
is 2.43, is the difference between the two groups significant?
Example 2.2.7. A random sample of 1000 workers from fac-
tory A shows that the mean wages were Rs.47 with a standard
deviation of Rs.23. A random sample of 1500 workers from fac-
tory B gives a mean wage of Rs.30. Is there any significant
difference between their mean level of wages?
Example 2.2.8. Given in usual notations, n1 = 400, x¯1 = 250,
s1 = 40 n2 = 400, x¯2 = 220, s2 = 55 Test whether the two
samples have come from populations having the same mean?
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
72
Test concerning the proportion of suc-
cess of a population
By testing population proportion of success, we mean the test-
ing of the significant difference between population proportion
of success and the sample proportion of success. Now let us
familiarise the following notations
p : Population proportion of success
p0 : The assumed value of p (proposed by H0 )
q 0 = 1 − p0
x : The number of successes in the sample
n : Sample size
p0 : The proportion of success in the sample = nx
Here, the null hypothesis to be tested is
H 0 : p = p0
against one of the alternatives
1. H1 : p < p0
2. H1 : p > p0
3. H1 : p 6= p0 .
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
73
Based on a large sample of size n whose proportion of success
is p0 . The test statistic is
p0 − p
Z= q 0
p0 q0
n
For a given significance level , the BCR are respectively,
1. Z < −Zα
2. Z > Zα
3. |Z| ≥ Zα/2
Calculate the value of Z and if it lies in the critical region reject
H0 , or otherwise accept H0 .
Example 2.2.9. In a survey of 70 business firms, it was found
that 45 are planning to expand their capabilities next year. Does
the sample information contradict the hypothesis that 70% of the
firms are planning to expand next year. (Hint : Z= -1.04)
Example 2.2.10. In a die rolling experiment, getting 3 or 6
is identified as a success. Suppose that 9,000 times the die was
rolled resulting in 3240 successes. Do you have reasons to believe
that the die is an unbiased one?
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
74
Test concerning difference of two pop-
ulation proportions
By testing the difference of two population proportions, we are
testing the equality of two population proportions. In other
words, we are deciding whether the two samples have come from
populations having the same proportion of success. Let us con-
sider the following notations
p1 : Proportion of success of the first population
p2 : Proportion of success of the second population
x1 : Number of successes in the first sample
x2 : Number of successes in the second sample
n1 : Size of first sample
n2 : Size of second sample
x
p01 : Proportion of success in the first sample = n1
1
x2
p02 : Proportion of success in the second sample = n2
Suppose we want to test the null hypothesis
H 0 : p1 = p2
against one of the alternatives
1. H1 : p1 < p2
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
75
2. H1 : p1 > p2
3. H1 : p1 6= p2
Based on two independent random samples of sizes n1 and n2
with proportions of successes p1 0 and p2 0 respectively.
The test statistic is
p1 0 − p2 0
Z=r
p∗ q ∗ n1 + 1
n2
1
n1 p1 0 +n2 p2 0
where p∗ = n1 +n2 and q ∗ = 1 − p∗ .
For a given significance level α, the BCR are respectively
1. Z < −Zα
2. Z > Zα
3. |Z| ≥ Zα/2
Calculate the value of Z and if it lies in the critical region, reject
H0 or otherwise accept H0 .
Example 2.2.11. Before an increase in excise duty on tea, 800
persons out of a sample of 1000 were found to consume tea. Af-
ter an increase in the duty, 800 people out of 1200used to take
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
76
tea. Test whether there is significant decrease in the consump-
tion of tea after the increase in the duty. (Hint : Z=6.816)
Example 2.2.12. In a sample of 600 men from city A, 450were
found to be smokers. Out of 900 from city B, 450 were smokers.
Do the data indicate that the cities are significantly different with
respect to the smoking habit?
Example 2.2.13. Exercise 3 :A machine , in the long run,
produces 16 defective items out of every 500 produced. After the
machine is repaired, it produced 3 defective items in a batch of
100. Has the machine improved its performance?
2.3 Small Sample Tests
If the sample size n is less than 30, it is called small sample.
When the sample size is large, statistical validity of the test is
insured by central limit theorem. But, when the sample size is
small, as in most of the practical situations, the central limit
theorem does not apply. So stricter assumptions on the popula-
tion distributions is required to obtain statistical validity. One
common assumption is that the population under consideration
is having normal probability distribution. For small samples,
when the population is normal, the sampling distributions are
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
77
t, F and χ2 .
Note 2.3.1. Notations:
1 X 2
S2 = (xi − x̄)
(n − 1)
1X 2
s2 = (xi − x̄)
n
t - tests
The test of hypothesis based on the Student’s t distribution is
called t-test. The main applications of t-test are,
1. To test the significance of mean of a small sample from a
normal population
2. To test the significance of the difference between the means
of two independent samples taken from two normal popu-
lations.
3. To test the significance of the difference between the means
of two dependent samples taken from a normal population.
4. To test the significance of an observed correlation coeffi-
cient.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
78
5. To test the significance of an observed regression coeffi-
cient.
t- test for significance of population
mean
To test the mean of a population using Student’s t-test, the
following assumptions are made.
1. The parent population from which the sample is drawn is
normal.
2. The sample observations are independent and random.
3. The sample size should be small (i.e., n < 30).
4. The population standard deviation σ is unknown.
Note 2.3.2. When the population standard deviation is known,
X̄−µ
even if the sample size is small, σ/√n0 still has the standard
normal distribution, as discussed in the previous section.
By testing the mean of a normal population, we are actually
testing the significant difference between sample mean and the
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
79
hypothetical value of µ proposed by the null hypothesis. In
other words, we are testing whether the sample is drawn from
the population with the hypothetical mean.
Here we are testing
H0 : µ = µ0
against one of the alternatives
1. H1 : µ < µ0
2. H1 : µ > µ0
3. H1 : µ 6= µ0
The test statistic is
x̄ − µ0
t= ∼ tn−1 .
√S
n
Or
x̄ − µ0
t= ∼ tn−1 .
√s
n−1
For a given significance level, the BCR are respectively,
1. t < −tn−1,α
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
80
2. t > tn−1,α
3. |t| ≥ tn−1,α/2
Now calculate the value of t and compare it with the table
value of t for (n − 1) degrees of freedom and a given significance
level α.
Example 2.3.1. A sample of 10 observations gives a mean
equal to 38 and SD 4. Can we conclude that the population
mean is 40? (Hint : t = -1.5)
Example 2.3.2. A random sample of size 16 has 53 as mean
and the sum of squares of the deviations taken from the mean is
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
81
150. Can the sample be regarded as arisen from the population
with mean 56.
Example 2.3.3. A sample of size 8 from a normal population
is 6,8,11,5,9,11,10,12. Can such a sample be regarded as drawn
from a population with mean 7 at 2% level of significance?
Example 2.3.4. A personality test was conducted on a random
sample of 10 students from a University and the scores obtained
were 35, 60, 55, 50.5, 44, 47.5, 41.5, 49, 53.5, 50. Test whether
the average ”personality test score ”for the University students
is 50 at 5% level?
t- Test concerning difference of means
of two populations
By testing the equality of two population means, we are deciding
whether the two samples are drawn from populations having the
same mean. Assumptions :
1. The two populations from which the samples are drawn
follow normal distributions.
2. The sample observations are independent and random
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
82
3. The sample sizes are small (n1 < 30, n2 < 30)
4. The two population variances σ1 2 and σ2 2 are equal but
unknown.
Suppose we want to test the null hypothesis H0 : µ1 − µ2 = 0
against one of the alternatives
1. H1 : µ1 < µ2
2. H1 : µ1 > µ2
3. H1 : µ1 6= µ2 .
Based on independent random samples of sizes n1 and n2 from
two populations. The test statistic is
x¯1 − x¯2
t= r ∼ tn1 +n2 −2 .
(n1 −1)S12 +(n2 −1)S22
1 1
n1 +n2 −2 n1 + n2
Or
x¯1 − x¯2
t= r ∼ tn1 +n2 −2
n1 s21 +n2 s22
1 1
n1 +n2 −2 n1 + n2
For the significance level α, the best critical regions (BCR)
are respectively
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
83
1. t < −tn1 +n2 −2,α
2. t > tn1 +n2 −2,α
3. |t| ≥ tn +n −2,α/2
1 2
Here, tα is obtained by referring the t- table for n1 + n2 − 2
degrees of freedom. Calculate the value of the statistic t using
the sample information and if it lies in the critical region reject
H0 , or otherwise accept H0 .
Example 2.3.5. The mean life of a sample of 10 electric bulbs
was observed to be 1309 hours with a SD of 420 hours. A second
sample of 16 bulbs of a different batch showed a mean life of 1205
hours with a SD of 390 hours. Test whether there is significant
difference between the means at 5% level. (Hint: t=0.619)
Example 2.3.6. A random sample of 16 men from state A had
a mean height of 68 inches and sum of squares from the sample
mean 132. A random sample of 25 men from state B had the
corresponding values 66.5 and 165 respectively. Can the samples
be regarded as drawn from populations having the same mean?
Example 2.3.7. The following are samples from two indepen-
dent normal populations. Test the hypothesis that they have the
same mean assuming that the variances are equal at 5% level of
significance.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
84
Sample 1 : 14,18,12,9,16,24,20,21,19,17
Sample 2: 20,24,18,16,26,25,18.
t - test for dependent samples (Paired
t - test)
A situation similar to that discussed in the previous case occurs
when we have date obtained by observing the values of some
attribute of different experimental units of the population, ob-
served “before” and “after”. A typical case would be to measure
a certain reaction on human subjects before (Xi ) and after (Yi )
some treatment. The purpose is to decide whether the treatment
has some effect or not. Even though here we have the situation
of two samples x1 , x2 , . . . xn (values “before”) and y1 , y2 , . . . , yn
(values “after”), we can not apply the method discussed above.
Such a procedure would be incorrect, since in the present case,
the values (xi , yi ) are not independent as they are observations
on the same subject. However, under some assumptions, we can
use here one sample t-test. Imagine that the observations are
such that the differences di = xi − yi , i = 1, 2, . . . n have the
same normal distribution with means µ and variance σ 2 . The
values xi or yi separately need not be normally distributed. Now
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
85
we test the null hypothesis
H0 : µ = 0
against one of the alternatives
1. H1 : µ < 0
2. H1 : µ > 0
3. H1 : µ 6= 0
The test statistic is
d¯
t= S
,
√d
n
P q
di
where d¯ = 1 ¯ 2.
P
n and Sd = (n−1) (di − d)
Or
d¯
t= s ,
√ d
n−1
P q P
d
where d¯ = n i and sd = n1 ¯ 2 . For a given signifi-
(di − d)
cance level α, the best critical regions are respectively,
1. t < −tn−1,α
2. t > tn−1,α
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
86
3. |t| ≥ tn−1,α/2
Here, tα is obtained by referring the t- table for n − 1 degrees of
freedom. Now calculate the value of the statistic and if lies in
the critical region, reject the null hypothesis otherwise accept
null.
Example 2.3.8. To test the efficiency of sleeping tablets, a drug
company uses a sample of insomniacs. The time in minutes
until falling asleep is observed for each of them. Few days later,
the same persons are given a sleeping tablet and the time until
falling asleep is observed again. The measurements are given
below.
Persons A B C D E
No tablet 65 35 80 40 50
With tablet 45 15 61 31 20
Test whether the sleeping tablets are effective. (Hint: t=5.89)
Example 2.3.9. A certain stimulus administered on each of 10
patients resulted in the following changes in blood pressure. 5,
2, 8, -1, 3, -2, 1, 5, 4, 6. Can it be concluded that the stimulus
will be in general accompanied by an increase in blood pressure?
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
87
Testing the Correlation Coefficient
The correlation coefficient tells us about the strength and di-
rection of the linear relationship between x and y, where the
variables x and y are in interval or ratio scale. We perform a
hypothesis test of the “significance of the correlation coefficient”
to decide whether the linear relationship in the sample data is
strong enough to use to model the relationship in the popu-
lation. The hypothesis test lets us decide whether the value
of the population correlation coefficient ρ is “close to zero” or
“significantly different from zero”. We decide this based on
the sample correlation coefficient r and the sample size n. Let
(x1 , y1 ) , (x2 , y2 ) , . . . , (xn , yn ) be a random sample of size n from
a bivariate normal population. To test the null hypothesis
H0 : ρ = 0
against one of the alternatives
1. H1 : ρ < 0
2. H1 : ρ > 0
3. H1 : ρ 6= 0
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
88
The test statistic is
p
r (n − 2)
t= √ ∼ tn−2 ,
1 − r2
where r is the sample correlation coefficient. For the significance
level alpha the best critical regions are respectively
1. t < −tn−2,α
2. t > tn−2,α
3. |t| ≥ tα/2,n−2
When the sample is large,
p
r (n − 2)
Z= √ ∼ N (0, 1)
1 − r2
and the critical regions can be obtained from normal table.
Example 2.3.10. A scientist suspect that a person’s stress level
changes so does the amount of his or her impulse buying. To
test this hypothesis, he took a sample of size n = 72 and found
that r = 0.38. Test the hypothesis that this is suggestive of the
students suspicion?
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
89
Example 2.3.11. Test for the significance of correlation coef-
ficient at 5% level of significance if the sample of size n = 27
gives a sample correlation coefficient of 0.6.
Example 2.3.12. Consider a population consisting of all mar-
ried couples, and we want to investigate whether people tend to
marry partners of about same age. The following table gives
ages of 12 couples. Couple No. 1 2 3 4 5 6 7 8 9 10 11 12
Husband’s Age 30 29 36 72 37 36 51 48 37 50 51 36 Wife’s
Age 27 20 34 67 35 37 50 46 36 42 46 35 Test whether there is
a positive correlation between husband’s and wife’s age. (Hint:
r=0.969, t=12.4028)
Example 2.3.13. The following table gives data on day time
temperature (in F) and hot chocolate sale (in $) for a small
snack shop during 9 youth soccer matches in 2002 Match No. 1
2 3 4 5 6 7 8 9 Temperature 51 60 65 71 39 32 81 76 66 Sale
187 210 137 136 241 262 110 143 152 Test whether there is a
negative correlation between temperature and hot chocolate sale.
(Hint: r=-0.946, t=-1.895)
Chi-square Test for Population Variance
This test is conducted when we want to test if the given normal
population has a specified variance, say σ02 . Chi square test for
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
90
variance is generally a right tailed test. Here we are testing
H0 : σ 2 = σ 0 2
against
H1 : σ 2 > σ0 2
The test statistic is
ns2
χ2 =
σ02
or
(n − 1)S 2
χ2 =
σ02
For a given significance level α, the BCR is
ω : χ2 > χα 2 ,
where χα 2 is obtained by referring the chi square table for n − 1
d.f. and a for a given significance level α.
Calculate the value of the statistic and if it lies in the critical
region reject the null hypothesis, otherwise accept the null.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
91
Problems
1. A manufacturing process is expected to produce goods
with a specified weight with variance less than 5unts. A
random sample of 10 was found to have variance 6.2 units.
Is there reason to suspect that the process variance has
increased?
2. A farmer surveyed 4 plots of land and found the following
yields. 1269,1271,1263,1265. He believes that his produc-
tion have a SD of 2. Test at 5% level whether his data is
consistent with his supposition.
F Test for Equality of Two Population Vari-
ances
This test is used for testing the equality of variances (or stan-
dard deviations) of two normal populations. Draw a random
sample of size n1 from N µ1 , σ1 2 and a sample of size n2 from
S1 2 and S2 2 be their sample variances. Here
N µ2 , σ2 2 . Let
we want to test
H0 : σ 1 2 = σ 2 2
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
92
against
H1 : σ1 2 > σ2 2 .
The test statistic is
S12
F = ∼ F(n −1)(n −1) .
S22 1 2
For a given significance level α, the best critical region is given
by
ω : F > Fα ,
where Fα is obtained by referring F- table for (n1 − 1, n2 − 1)
degrees of freedom. Calculate the value of the test statistic
and if it lies in the critical region reject the null hypothesis ,
otherwise accept the null.
Notes :
1. Before starting the F test, compute S12 and S22 . Take the
larger value as the numerator in the F ratio and n1 corre-
sponds to the value in the numerator.
2. If the required F value is not plotted in the table, inter-
polate between the available values to determine the table
value.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
93
3. The test statistic can also be written in the form
n1 s 2
1
n1 −1
F = n2 s22
∼ F(n −1)(n −1)
.
1 2
n2 −1
Problems
1. Sample of sizes 10 and 18 taken from two normal popula-
tions gave s1 = 14 and s2 = 20. Test the hypothesis that
the samples have come from populations with the same
standard deviation.
2. The random samples of sizes 8 and 11 drawn from two
normal populations are characterized as follows.
Sample size Sum of obs. Sum of squares of obs.
8 9.6 61.52
11 16.5 73.26
Examine whether the two samples came from populations
having the same variance.
3. In an animal feeding experiment, the following results were
noted.
Diet Gain in weight
High protein X 13,14,10,11,12,16,10,8,11,12,9,12
Low protein Y 7,11,10,8,10,13,9
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
94
Test the equality of variances in weight gain due to the
two diets.
Chi-square test for goodness of fit
Consider a set possible events 1, 2, 3, ..., n arranged in ’n’ classes
or cells. Let these events occur with frequencies O1 , O2 , . . . , On
called ’observed frequencies. Let them be expected to occur with
frequencies E1 , E2 , . . . , En called the ’expected frequencies. The
expected frequencies are calculated on the assumption that the
data obeys a certain probability distribution such as binomial,
normal etc. We test the hypothesis that the assumed probability
distribution fits good for the given data against the alternative
that it is not. A measure of disagreement existing between the
observed and expected frequencies can be found by using the
chi square test statistic given by
n 2
2
X (Oi − Ei )
χ =
Ei
i=1
If χ2 = 0, the observed frequencies and the expected frequencies
will coincide, and this shows that there is perfect agreement
between theory and observation. When the value computed
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
95
from the data exceeds limit, we reject H0 . Thus the chi square
test for goodness of fit is basically a right tailed test. The best
critical region is χ2 > χα 2 , where χα 2 is obtained by referring
the chi square table for (n−1) degrees of freedom. Now calculate
the value of the test statistic and if lies in the critical region,
reject H0 or otherwise.
The following points to be considered while conduction this
test. If we have to estimate the parameters like ’p’ (for binomial
distribution) or λ (for Poisson distribution), to compute the
expected frequencies, then one or more degrees of freedom has
to be subtracted for each parameter estimated. If any of the
expected cell frequency is less than 5, then we combine (pool)
it with the preceding or succeeding cell frequency, so that the
resulting frequency is greater than 5. In this case also, one
degrees of freedom has to be adjusted.
Problems
1. Four coins are tossed 80 times. The distribution of number
of heads is given below.
No. of heads : 0 1 2 3 4
Frequency : 4 20 32 18 6
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
96
Apply chi square test at 1% level to test whether the coins
are unbiased. (Value of statistic=0.73)
2. The demand for refrigerators in a city are found to vary
day by day. In a study. the following data were obtained.
Test at 5% level of significance whether the demand de-
pends on the day of the week.
Days : Mon Tue Wed Thur Fri Sat
Demand : 115 126 120 110 125 124
3. Fit a Poisson distribution for the following data and test
for goodness of fit.
X : 0 1 2 3 4 5 6
F : 275 72 30 7 5 2 1
Chi - square test for independence of attributes
Let there be two attributes A and B, ‘A’ divided into ‘m’ classes
A1 , A2 , . . . Am and ‘B’ divided into ‘n’ classes B1 , B2 , . . . Bn .
The various cell frequencies can be expressed as in the following
table, having ‘m’ rows and ‘n’ columns.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
97
A\B B1 B2 ... Bj ... Bn Total
A1 f11 f12 ... f1j ... f1n f1.
A2 f21 f22 ... f2j ... f2n f2.
.. .. .. .. .. .. .. ..
. . . . . . . .
Ai fi1 fi2 ... fij ... fin fi.
.. .. .. .. .. .. .. ..
. . . . . . . .
Am fm1 fm2 ... fmj ... fmn fm.
Total f.1 f.2 ... f.j ... f.n N = f..
Here,
fij - frequency of the occurrence of the joint event (Ai , Bj )
fi. - frequency of occurrence of the event Ai
f.j - frequency of occurrence of the event Bj
Such a table is called m × n contingency table. We test the
hypothesis H0 that the attributes A and B are independent.
That is, there is no association between A and B. If H0 is true,
we have,
P Ai , Bj = P (Ai ) P Bj
fij f f
i.e., = i. x .j .
N N N
Hence,
fi. f.j
fij =
N
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
98
The test statistic is given by
m X
n 2
2
X Oij − Eij
χ =
Eij
i=1 j=1
The BCR is given by χ2 > χ2 α , where χ2 α is obtained by
referring the chi square table for (m − 1)(n − 1) degrees of free-
dom. Calculate the value of the statistic and if lies in the critical
region, reject the null hypothesis.
Problems
1. From the following table on some horticultural data, test
the hypothesis that the flower colour is independent of the
nature of leaves. Leaves Flowers
White flowers Red flowers
Flat leaves 99 20
Curled leaves 36 5
(Hint : χ2 = 0.249)
Note 2.3.3. For a 2 × 2 contingency table, where the cell
a b
frequencies are the calculated value of chi square
c d
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
99
is given by,
2
N (ad − bc)
χ2 = ,
(a + b) (c + d) (a + c) (b + d)
where N = a + b + c + d.
2. Consider the following 2 × 2 contingency table.
A1 A2
B1 7 1
B2 6 8
Apply chi square test at 5% level to test whether the two
attributes A and B are independent.
Yates’ Correction
Yates’ correction was proposed by F.Yates in 1934. We
have learnt that no expected cell frequency should be less
than 5 for applying chi square test. This is because, when
the expected frequencies are less than 5, the chi square
table values are not very reliable, especially for 1 degree
of freedom.
However, if chi square test is applied after considering
Yates’ correction, we will get a reduced chi square value.
Sometimes, the difference between the chi square value
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
100
with and without Yates’ correction is so much that it leads
to a totally different conclusion.
Consider the 2x2 contingency table
a b
c d
With Yates’ correction, the calculated chi square becomes
2
2 N |ad − bc| − N2
χ = ,
(a + b) (c + d) (a + c) (b + d)
3. Consider the previous example and test the independence
of attributes with Yates’ correction
4. In an experiment on immunization, of human beings, the
following results were obtained. Draw your inference on
the efficiency of the vaccine at 5% level.
Died Survived
Vaccinated 2 10
Not vaccinated 6 4
5. A driving school examined the result of 200 candidates
who were taking their test for the first time. They found
that out of 90 men, 52 passed and out of 110 women 59
passed. Do these data indicate at 1% level of significance
that a relationship between gender and the ability to pass
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
101
the test for the first time?
2.4 Analysis of Variance (ANOVA)
(Test for equality of several population means)
Test of significance of the difference between 2 sample means
is based on t-distribution. But when we have three or more sam-
ples to consider at a time this procedure is inadequate. There-
fore, for testing the hypothesis that all the samples are drawn
from the same population, i.e., they have the same mean the
technique of ANOVA is used.
The variation present in a set of observations under study
may be caused by known and unknown factors. Variations due
to unknown factors are known as random variations. In analysis
of variance, an attempt is made to separate the variation due to
known factors from the variation due to unknown factors. For
this, the total variance present in the whole set of observations
is partitioned into a number of component variances caused by
each set of independent factors. If the component variances do
not differ significantly, it is concluded that the effects of all the
factors are equal.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
102
Assumptions:
1. Observations yij ’s are independent and random.
2. Parent populations are normal with common variance σ 2
3. Different effects are additive in nature.
4. Error component follows N (0, σ 2 )
One-Way Classification
In one-way classification, observations are classified based on a
single criterion. For example, suppose we want to study the
effect of a variable, say fertilizer, in the yield of a crop. So, we
apply different fertilizers on different fields and try to find out
the difference in the effect of these fertilizers on yield. Suppose
we apply k fertilizers. Hence, we get k samples each of them
drawn from these k populations. (The different populations
here, are the fields where different fertilizers are applied). Here
the only one variable is the ‘fertilizer, which is the criterion for
classification.
Let us suppose that N observations yij , (i = 1, 2, . . . n; j =
1, 2, . . . ni ) of a random variable Y are grouped on some basis
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
103
Pk
into k classes of sizes n1 , n2 , . . . , nk , N = i=1
ni as exhibited
in the table below.
Class/ Sample Observations Total
Treatment
1 y11 y12 ... y1n1 T1
2 y21 y22 ... y2n T2
2
.. .. .. .. ..
. . . . .
i yi1 yi2 ... yin Ti
j
.. .. .. .. ..
. . . . .
k yk1 yk2 ... ykn Tk
k
Grand Total G
This scheme of classification according to 1 factor or criterion
is called one-way classification and its analysis is called one-way
ANOVA.
The total variation in the observations yij can be split into
two components.
1. Variation between the classes or variation due to
assignable causes (different bases of classification), com-
monly known as treatments. This can be detected and
controlled.
2. Variation within the classes or variation due to chance
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
104
causes. This cannot be controlled.
Mathematical model for one-way classification
Any observation belonging to the collected data has three com-
ponents, namely general mean effect, effect of the fertilizer and
the effect of unknown factors. i.e.,
yij = µ + αi + ij
where yij is any observation, µ is the general mean effect, αi
is the effect of ith treatment and ij is the effect of unknown
factors.
The null hypothesis is:
H0 : All the treatment means are homogeneous
That is,
H0 : µ1 = µ2 = . . . = µk
The main objective of ANOVA technique is to examine if
there is significant difference between the class means in view of
the inherent variability within the separate classes. The name
ANOVA is derived from partitioning of total variability into its
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
105
component parts.
The total sum of squares (T.S.S.),
ni
k X
X
T.S.S. = yij − ȳ..
i=1 j=1
is used as an overall measure of variability in data. Intuitively,
this is reasonable because, if we were to divide T.S.S. by the cor-
responding degrees of freedom N − 1, we would have the sample
variance. Sample variance is of course a standard measure of
variability. T.S.S. is partitioned as follows:
T.S.S. = S.S.Tr. +S.S.E.
Where, S.S.Tr is the S.S. due to treatments, i.e., between treat-
ments or classes And S.S.E. is the S.S. due to error, i.e., within
treatments or classes. The degrees of freedom (d.f.) due to
various sum of squares are as follows:
S.S. d.f.
T.S.S. N −1
S.S.Tr. k−1
S.S.E. N −k
Note 2.4.1. N − 1 = (k − 1) + (N − k)
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
106
Mean Sum of Squares (M.S.S.):
The S.S. divided by its corresponding degrees of freedom gives
the corresponding variance or the M.S.S.. Thus,
T.S.S.
M.S.T. =
N −1
S.S.Tr.
M.S.Tr. =
k−1
S.S.E.
M.S.E. =
N −k
Note 2.4.2. M.S.T.6= M.S.Tr. + M.S.E.
The M.S.E. is always an unbiased estimator of σ 2 . i.e.,
E(M.S.E.) = σ 2
Under H0 , M.S.Tr. is an unbiased estimator of σ 2 . i.e.,
E(M.S.Tr.) = E(M.S.E.) = σ 2 , under H0
Otherwise,
E(M.S.Tr.) > E(M.S.E.)
The test statistic is given by the variance ratio
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
107
M.S.Tr.
F = ∼ Fk−1,N −k .
M.S.E.
Under H0 , F = 1, otherwise F > 1. So, smaller values of
F support H0 and larger values of F support H1 . Hence, the
critical region is
F > Fα .
ANOVA Table
Sources of S.S. d.f. MS.S. F -ratio
Variation
Treatment S.S.Tr. k−1 M.S.Tr. = S.S.Tr.
k−1
F = M.S.Tr.
M.S.E.
Error S.S.E. N −k M.S.E. = S.S.E.
N −k ∼ Fk−1,N −k .
Total T.S.S. N −1
Calculation of Various Sum of Squares
G2
Correction Factor, C.F. =
N
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
108
T.S.S. = Sum of squares of all observations − C.F.
ni
k X
X
2
= yij − C.F.
i=1 j=1
(T1 )2 (T )2 (T )2
S.S.Tr. = + 2 + . . . + k − C.F.
n1 n2 nk
S.S.E. = T.S.S. − S.S.Tr.
Two-Way Classification
Suppose N observations are classified into k categories, say
A1 , A2 , . . . , Ak according to some criterion A; and into h cate-
gories, say B1 , B2 , . . . , Bh according to some criterion B, having
kh combinations (Ai , Bj ), i = 1, 2, . . . , k; j = 1, 2, . . . , h; often
called cells. The number of observations in each cell may be
equal or not, but we shall consider the case of one observation
per cell, so that N = kh (i.e., total number of cells is N = kh),
as shown in the table below.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
109
Criterion B 1 2 ... j ... h Row Total
Criterion A
1 y11 y12 ... y1j ... y1h R1
2 y21 y22 ... y2j ... y2h R2
.. .. .. .. .. ..
. . . . . .
i yi1 yi2 ... yij ... yih Ri
.. .. .. .. .. ..
. . . . . .
k yk1 yk2 ... ykj ... ykh Rk
Column Total C1 C2 ... Cj ... Ch G
This scheme of classification according to 2 factors or cri-
terion is called two-way classification and its analysis is called
two-way ANOVA. In two-way classification, the values of the
response variable are affected by two factors.
The total variation in the observations yij can be split into
three components.
1. Variation between the rows or variation due to criterion
A.
2. Variation between the columns or variation due to crite-
rion B.
3. Variation within the classes or variation due to chance
causes. This cannot be controlled.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
110
Mathematical model for two-way classification
An observation chosen from the collected data has four compo-
nents namely general mean effect, effect of the column variable,
effect of the row variable and the effect of unknown factors. i.e.,
yij = µ + αi + +βj + ij
where yij is the observation in ij th cell, µ is the general mean
effect, αi is the effect of Ai row, βj is the effect of factor Bj
column and ij is the effect of unknown factors.
The null hypothesis is:
H0A : Criterion A (Row) means are homogeneous
That is,
H0A : µA1 = µA2 = . . . = µAk
H0B : Criterion B (Column) means are homogeneous
That is,
H0B : µB1 = µB2 = . . . = µBh
T.S.S. is partitioned as follows:
T.S.S. = S.S.A. + S.S.B. + S.S.E.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
111
where, S.S.A. is the S.S. due to criterion A (i.e., the row S.S.),
S.S.B. is the S.S. due to criterion B (i.e., the column S.S.), and
S.S.E. is the S.S. due to error. The degrees of freedom (d.f.)
due to various sum of squares are as follows:
S.S. d.f.
T.S.S. kh − 1
S.S.A. k−1
S.S.B. h−1
S.S.E. (k − 1)(h − 1)
Note 2.4.3. kh − 1 = (k − 1) + (h − 1) + (k − 1)(h − 1))
Mean Sum of Squares (M.S.S.)
The S.S. divided by its corresponding degrees of freedom gives
the corresponding variance or the M.S.S.. Thus,
T.S.S.
M.S.T. =
kh − 1
S.S.A.
M.S.A. =
k−1
S.S.B.
M.S.B. =
h−1
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
112
S.S.E.
M.S.E. =
(k − 1)(h − 1)
Note 2.4.4. M.S.T.6= M.S.A. + M.S.B. + M.S.E.
The M.S.E. is always an unbiased estimator of σ 2 . i.e.,
E(M.S.E.) = σ 2
Under H0A , M.S.A. is an unbiased estimator of σ 2 . i.e.,
E(M.S.A.) = E(M.S.E.) = σ 2 , under H0A
Otherwise,
E(M.S.A.) > E(M.S.E.)
Under H0B , M.S.B. is an unbiased estimator of σ 2 . i.e.,
E(M.S.B.) = E(M.S.E.) = σ 2 , under H0B
Otherwise,
E(M.S.B.) > E(M.S.E.)
The test statistic is given by the variance ratio
M.S.A.
FA = ∼ Fk−1,(k−1)(h−1) .
M.S.E.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
113
M.S.B.
FB = ∼ Fh−1,(k−1)(h−1) .
M.S.E.
If FA > Fα , reject the null hypothesis H0A .
If FB > Fα , reject the null hypothesis H0B .
ANOVA Table
Sources of S.S. d.f. MS.S. F -ratio
Variation
Criterion A S.S.A. k−1 M.S.A. = S.S.A.
k−1 F = M.S.A.
M.S.E.
∼ Fk−1,(k−1)(h−1) .
Criterion B S.S.B. h−1 M.S.B. = S.S.B.
k−1 F = M.S.B.
M.S.E.
∼ Fh−1,(k−1)(h−1) .
Error S.S.E. (k − 1)(h − 1) M.S.E. = S.S.E.
N −k
Total T.S.S. N −1
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
114
Calculation of Various Sum of Squares
G2
Correction Factor, C.F. =
N
T.S.S. = Sum of squares of all observations − C.F.
ni
k X
X
2
= yij − C.F.
i=1 j=1
(R1 )2 (R )2 (R )2
S.S.A. = + 2 + . . . + k − C.F.
h h h
(C1 )2 (C )2 (C )2
S.S.B. = + 2 + . . . + h − C.F.
k k k
S.S.E. = T.S.S. − S.S.A. − S.S.B.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
MODULE
THREE
Non-parametric Tests
To use parametric tests, there should be some assumptions
about form of the underlying distribution or sample size should
be large. When such conditions are not satisfied we should opt
for non-parametric tests. Recall that, non-parametric does not
mean that there isn’t any parameter. It means that, knowing
the value of the parameter does not completely specifies the
distribution.
115
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
116
3.1 Advantages and Disadvantages
Advantages
1. Easily understandable.
2. Short calculations.
3. Assumption of distribution is not required.
4. Applicable to all types of data.
Disadvantages
1. Less efficient as compared to parametric tests.
2. The results may or may not provide an accurate answer
because they are distribution free.
3.2 Applications of Non-Parametric
Test
The conditions when non-parametric tests are used are listed
below:
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
117
1. When the population distribution is not known
2. When the sample size is not large to enough for normal
approximation
3. When the data is qualitative.
3.3 Test for Randomness
Wald-Wolfowitz Run Test
In order to arrive at a conclusion about a distribution on the
basis of a sample we often need to have a random sample, i.e.,
sample is taken without any bias or pre-designed rule of selec-
tion. Hence, often we test the randomness of a sample. Wald-
Wolfowitz run test, developed by Abraham Wald and Jacob
Wolfowitz, is a statistical analysis that helps determine the ran-
domness of data. Runs test are applicable to either quantitative
or qualitative data. It is used to test whether a sequence of two
possible outcomes is random or to test a sample is random or
for determining whether the two populations follow the same
distribution.
Given an ordered sequence of two or more types of symbols,
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
118
a run is a succession of two or more identical symbols which are
preceded and followed by a different symbol or no symbol. If
the number of runs is too large or too small, it is unlikely that
the points are random. The length of a run is the number of
like symbols in a run. For applying the run test it is important
to maintain the order in which the samples are chosen.
Test Procedure
In order to test if a sequence of dichotomous data is random the
following procedure is used. Let A and B be the two possible
dichotomous outcomes. Let n1 be the number of observations
of type A, n2 be that of B and R be the number of runs. The
test statistic R can take any value between 2 and n = n1 + n2 .
A very large and a very small value of R supports H1 . If R lies
below the lower critical value or above the upper critical value,
we reject the null and conclude that the sequence is not random.
In case of numeric data the dichotomy is usually obtained
by comparing each number with a focal point like a median or
mean. First arrange the observations in increasing order of their
magnitude. The observations above the median are labeled A
and that below are labeled B. Then the above procedure is used
to conduct the test.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
119
For determining whether the two populations follow the
same distribution, the following procedure is used. A sample
of size n1 and n2 are taken from populations X and Y respec-
tively. Combining the two populations, n1 + n2 = n samples
are arranged in increasing order of their magnitudes keeping
track of the population. A clustering of samples from the same
population preceded and followed by that from the other forms
a run. The total number of runs, R, in the combined sample of
X’s and Y ’s when arranged in increasing order can be used as
a test statistic of H0 . When the two samples are drawn from
two populations having the same distribution, the sample must
be very well mixed. That is, under H0 the X and Y symbols
are expected to be well mixed. If R lies between lower and
upper critical values we accept H0 , otherwise reject H0 .
A test based on R is appropriate only for two-sided (general)
alternatives. Tables of critical values are available.
Normal Approximation
For n1 or n2 ≥ 20, we can use normal approximation. The test
statistic is
R − µR
Z= ,
σR
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
120
2n n 2n1 n2 (2n1 n2 −n1 −n2 )
where µR = n +n 1 2
+ 1 and σR = (n1 +n2 )2 (n1 +n2 −1) . H0 is
1 2
rejected if |Z| ≥ z α .
2
3.4 Problem of Location - Sign Test
When the population is not normal and sample size is small or
the data is not quantitative, to test the problem of location we
cannot use parametric tests and we go for non-parametric tests.
Here our aim is to check if the sample provide strong evidence
that the central location of the distribution (M , the median) is
at the point M0 . Two important tests we study in this regard
are:
1. One Sample Sign Test
2. One Sample Wilcoxon Signed Rank Test.
These are non parametric alternative for one sample t-test.
One Sample Sign Test
This test is one of the simplest of all non-parametric tests. The
sample is assumed to be taken from a population which is con-
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
121
tinuous, symmetrical and non-normally distributed. The null
hypothesis is
H 0 : M = M0
against one of the possible alternatives
1. H1 : M > M0
2. H1 : M < M0
3. H1 : M 6= M0
Test Procedure
Let X1 , X2 , . . . , Xn be a r.s. of size n from a continuous popu-
lation. A positive sign is given to an observed values Xi if it is
greater than the hypothesised median M0 and a negative sign
is given to an observed values Xi if it is less than the hypothe-
sised median M0 . If Xi = M0 , ignore the zeros and reduce the
number of observations accordingly. Let N + be the number of
positive signs and N − be the number of negative signs. Then,
N = N + + N −.
If the sample truly comes from the distribution with median
M0 , then it is expected that nearly half of the sample observa-
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
122
tions will be > M0 and nearly half of the observations will be
< M0 .
Since the median divides the data in such a way that the
number of observations above it is equal to the number of ob-
servations below it, the probability of a positive sign is same as
the probability of the negative sign.
Once the observations are converted as above, under H0 ,
they constitute a set of N independent random variables from
Bernoulli population with probability of success, p = 12 , where,
p is the probability of getting either a positive sign or negative
sign. Then under H0 , N + ∼ B(N, 12 ) and N − ∼ B(N, 12 ).
Hence, the null hypothesis can also be written as
1
H0 : p =
2
against one of the possible alternatives
1
1. H1 : p > 2
1
2. H1 : p < 2
1
3. H1 : p 6= 2
We reject H0 , in favor of
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
123
1
:: either N − is too small or alterna-
H1 : M > M 0 p > 2
tively N + is too big
1
H1 : M < M0 p < 2 :: if either N + is too small or alter-
natively N − is too big
1
:: if any one of N + or N − is too big
H1 : M 6= M0 p 6= 2
or too small.
For small samples, the test statistic is S = min(N + , N − ).
Let k be the observed value of S. The p - value is P (S ≤ k). If
p < α, we reject H0 . For two tailed test, H0 is rejected if p < α2 .
When sample size is large,
N
S− 2
Z= q → N (0, 1).
N
4
Applying continuity corrections, the test statistic is
N
S + 0.5 − 2
Z= q → N (0, 1).
N
4
The p - value is the cumulative probability corresponding to z,
the calculated value of Z. That is, p = P (Z ≤ z). If p < α, we
reject H0 . For two tailed test, H0 is rejected if p < α2 .
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
124
Paired Sample Tests
Two Sample Sign Test
Two sample sign test can be used as the non-parametric version
of paired t test, like to study the effectiveness of any programme
or treatment implemented on the same group of subjects. It can
also be used where two samples are taken from two populations
which have continuous symmetrical distributions and known to
be non-normal such that the probability of having, “the first
sample value less than the corresponding second sample value”
as well as the probability of having ”the first sample value more
than the corresponding second sample value”, is 12 .
The null hypothesis is
H 0 : M1 = M2
against one of the possible alternatives
1. H1 : M1 > M2
2. H1 : M1 < M2
3. H1 : M1 6= M2
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
125
Test Procedure
Give a positive sign if Xi − Yi > 0 and negative sign if Xi −
Yi < 0, i = 1, 2, . . . n where, Xi and Yi respectively denote
the corresponding observations in the first and second samples.
Obviously, the sample sizes has to be equal. Now, P (Xi − Yi >
0) = P (Xi − Yi < 0) = 12 . The two sample sign test is carried
out as in the case of one sample with this set of signs.
The null hypothesis can also be written as
1
H0 : p =
2
against one of the possible alternatives
1
1. H1 : p > 2
1
2. H1 : p < 2
1
3. H1 : p 6= 2
We reject H0 , in favor of
H1 : M1 > M2 p > 12 :: either N − is too small or alter-
natively N + is too big
p < 12 :: if either N + is too small or
H 1 : M1 < M 2
alternatively N − is too big
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
126
1
:: if any one of N + or N − is too
H1 : M1 6= M2 p 6= 2
big or too small.
For small samples, the test statistic is S = min(N + , N − ).
Let k be the observed value of S. The p - value is P (S ≤ k). If
p < α, we reject H0 . For two tailed test, H0 is rejected if p < α2 .
When sample size is large,
N
S− 2
Z= q → N (0, 1).
N
4
Applying continuity corrections, the test statistic is
N
S + 0.5 − 2
Z= q → N (0, 1).
N
4
The p - value is the cumulative probability corresponding to z,
the calculated value of Z. That is, p = P (Z ≤ z). If p < α, we
reject H0 . For two tailed test, H0 is rejected if p < α2 .
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
127
3.5 Problem of Equality of two Popu-
lations
Median Test
Median test is used for comparing the distribution of two pop-
ulations. The null hypothesis
H0 : M1 = M2
is tested against the two sided alternative
H1 : M1 6= M2 .
For each sample, the proportion of samples below C is
checked and then check their differences. If the difference is
significantly large, the null hypothesis is rejected. Otherwise
accept the null. Even though the method is fast, it has the
disadvantage that the decision may vary with C. Typically, C
can be any quantile. We consider C to be the median.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
128
Test Procedure
Let X1 , X2 , . . . , Xn and Y1 , Y2 , . . . , Yn be two independent
1 2
random samples. For each sample, check the proportion of
samples below C. Let U be the number of X observations less
than C and V be the number of Y observations less than C.
Then,
U ∼ B(n1 , p1 )
and
V ∼ B(n2 , p2 ),
where p1 is P (X < C) and p2 is P (Y < C). Then the null
hypothesis can be stated as
H 0 : p1 = p2
against the alternative
H1 : p1 6= p2 or p1 − p2 6= 0.
The appropriate test statistic is
U V
−
n1 n2
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
129
U V
and the null is rejected if n1 − n2 > Cα .
Alternatively, median of the combined sample is obtained
and the data is represented in a contingency table as below.
Sample I Sample II Total
Above the Median a b a+b
Below the Median c d c+d
Total n1 = a + c n2 = b + d N
where N = a + b + c + d.
Now, calculate
n1
n2
a
p= n
b .
a+b
If p < α reject the null hypothesis.
Large Sample Approximation
In case of large samples (n1 + n2 ≥ 20), we have the Moods
median test which is based on chi-square statistic. Here, no cell
frequency should be less than 5. First median of the combined
sample is obtained and the data is represented in a contingency
table as above.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
130
Under the null hypothesis, the test statistic is
N (|ad − bc| − N2 )2
χ2 = ∼ χ2 (1).
(a + b)(c + d)(a + c)(b + d)
Wilcoxon Rank Sum Test
Wilcoxon rank sum test is used to test whether the two popu-
lation have the same distribution. That is,
H0 : FX (x) = FY (x).
If it is known that the populations are of the same form, then
the null will be medians are equal. That is,
H 0 : M1 = M2
and the possible alternatives are
1. H1 : M1 > M2
2. H1 : M1 < M2
3. H1 : M1 6= M2
It is a non parametric analogue of two sample t test. It is as-
sumed that the two populations are continuous.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
131
Test Procedure
Suppose that a sample of size n1 is taken from a continuous
papulation X and n2 from a continuous papulation Y , and the
samples are independent. Combine the two independent sam-
ples and obtain the ranks of the combined data. Let R1 be the
sum of the ranks of the first sample and R2 that of the second.
The test statistic W , is the rank sum associated with the smaller
sample.
The critical regions are:
H1 C.R.
M1 > M 2 W > TU
M1 < M 2 W < TL
M1 6= M2 W < TL or W > TU
Normal Approximation
For large values of n1 and n2 , say (≥ 10), the test statistic is
n1 (N +1)
W− 2
Z= n1 n2 (N +1)
∼ N (0, 1),
12
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
132
where N = n1 + n2 . Reject H0 : M1 = M2 against the alterna-
tive
1. H1 : M1 > M2 if Z > zα
2. H1 : M1 < M2 if Z < −zα
3. H1 : M1 6= M2 if |Z| ≥ zα/2
The critical values can be obtained from standard normal table.
Mann-Whitnney U Test
Mann-Whitnney U test is also used to test the equality of cen-
trality of two continuous distributions. Although it is closely
related to Wilcoxon Rank sum test, it is slightly different. The
null hypothesis is
H0 : FX (x) = FY (x) ∀x
against
H1 : FX (x) = FY (x − θ) ∀ x and θ 6= 0.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
133
If the populations are assumed to be of the same form, then
H0 : M1 = M2
and the possible alternatives are
1. H1 : M1 > M2
2. H1 : M1 < M2
3. H1 : M1 6= M2
It is also a non parametric analogue of two sample t test. It is
assumed that the two populations are continuous.
Test Procedure
Suppose that a sample of size n1 is taken from a continuous
papulation X and n2 from continuous papulation Y . Combine
the two independent samples and arrange them in increasing
order of magnitude. The statistic U1 is the sum of the number
of y observations that are less than each of the x observations in
the combined ordered arrangement. Similarly, the statistic U2
is the sum of the number of x observations that are less than
each of the y observations in the combined ordered arrangement.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
134
The possibility of xi = yj for some i, j can be omitted. Let
(
1 if yj < xi
T (xi , yj ) =
o if yj > xi
where i = 1, 2, . . . , n1 and j = 1, 2, . . . , n2 . Let
n1 n2
X X
U1 = T (xi , yj )
i=1 j=1
and
n 1 n2
X X
U2 = (1 − T (xi , yj )).
i=1 j=1
U1 = 0 (or U2 = n1 n2 ) if all xi ’s are < yj ’s. Hence, mostly
FX > FY (M1 < M2 ).
U1 = n1 n2 (or U2 = 0) if all xi ’s are > yj ’s. Hence, mostly
FX < FY (M1 > M2 ).
The test statistic U = min(U1 , U2 ). If H0 is true U should
be close to mn
2 . If U = U1 is too small, reject H0 in favour of
H1 : FX > FY and if U = U2 is too small, reject H0 in favour of
H1 : FX < FY .
Therefore, the rejection region is U ≤ cα for one tailed test
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
135
and U ≤ c α for two tailed test.
2
Normal Approximation
For large values of n1 and n2 , say (≥ 10), the test statistic is
n1 n2
U− 2
Z=q ∼ N (0, 1).
n1 n2 (n1 +n2 +1)
12
Reject H0 : M1 = M2 against the alternative
1. H1 : M1 > M2 if Z < −zα
2. H1 : M1 < M2 if Z > zα
3. H1 : M1 6= M2 if |Z| ≥ zα/2
The critical values can be obtained from standard normal table.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
136
3.6 Problem of Equality of Several
Population Medians
Kruskal-Wallis Test
Kruskal-Wallis test is a non-parametric analogue of one-way
ANOVA and is a generalisation of the Wilcoxon rank sum test
to more than two populations. It is also called one-way ANOVA
on ranks. It tests equality of two or more population medians.
The test statistic used is called the H statistic. The hypothesis
for the test are
H0 : The population medians are equal.
in other words
H 0 : M1 = M2 = . . . = Mk
against
H1 : The population medians are not equal.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
137
Test Procedure
Combine all the samples and rank them. Let Ri denote the sum
of the ranks in ith group. The test statistic is
k
X R2
12
H= i
− 3(n + 1),
n(n + 1) ni
i=1
where, n = n1 + n2 + . . . + nk and ni is the sample size of the
tth sample. The test statistic follows chi-square distribution if
the number of samples in each group is more than 5. Reject the
null hypothesis if H ≥ χ2k−1 otherwise accept the null.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
138
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
MODULE
FOUR
Quality Control
Quality control is a procedure or a set of procedures performed
to ensure the quality criterion of a product or service. Statis-
tical quality control (S.Q.C.) is the use of statistical methods
in the monitoring and maintaining the quality of products and
services, which determines its fitness to use. S.Q.C. is one of
the most important application of statistics in industry, based
on the theory of probability and sampling.
139
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
140
Here, quality does not mean the highest standard, but the
quality which conforms to the standards specified for measure-
ment. In many cases, it will be lower than the highest possible
standard. In fact, consistency in quality standard is expected
rather than the absolute standard. Quality control, therefore,
covers all the factors and processes of production which may be
broadly classified as follows:
1. Quality of materials: Material of good quality will re-
sult in smooth processing thereby reducing the waste and
increasing the output. It will also give better finish to the
end products.
2. Quality of manpower: Trained and qualified personnel
will give increased efficiency due to the better quality pro-
duction through the application of skill and also reduce
production cost and waste.
3. Quality of machines: Better quality equipment will re-
sult in efficient work due to lack or scarcity of breakdowns
and thus reduce the cost of defectives.
4. Quality of management: A good management is im-
perative for increase in efficiency, harmony in relations,
and growth of business and markets.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
141
4.1 Basis of S.Q.C.
The basis of statistical quality control is the degree of ‘vari-
ability’ in the size or the magnitude of a given characteristic of
the product. Variation in the quality of manufactured prod-
uct in the repetitive process in industry is inherent and in-
evitable. These variations are broadly classified as being due
to two causes, viz., i) chance causes, and ii) assignable causes.
1. Chance Causes: Some “stable pattern of variation” or
“a constant cause system” is inherent in any particular
scheme of production and inspection. The variation due to
these causes is beyond the control of human hand and can-
not be prevented or eliminated under any circumstances.
One has got to allow for variation within this stable pat-
tern, usually termed as allowable variation. The range of
such variation is known as ’natural tolerance of the pro-
cess’.
2. Assignable Causes: The second type of variation at-
tributed to any production process is due to non-random
or the so-called assignable causes and is termed as pre-
ventable variation. The assignable causes may creep in
at any stage of the process, right from the arrival of the
raw materials to the final delivery of goods. Some of the
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
142
important factors of assignable causes of variation are sub-
standard or defective raw materials, new techniques or op-
erations, negligence of the operators, wrong or improper
handling of machines, faulty equipment, unskilled or in-
experienced technical staff, and so on. These causes can
be identified and eliminated and are to be discovered in
a production process before it goes wrong, i.e., before the
production becomes defective.
The main purpose of Statistical Quality Control (S.Q.C.) is
to devise statistical techniques which would help us in separating
the assignable causes from the chance causes, thus enabling us
to take immediate remedial action whenever assignable causes
are present. Hence, a production process is said to be in a state
of statistical control, if it is governed by chance causes alone, in
the absence of assignable causes of variation.
Benefits of Statistical Quality Control
1. It provides a means of detecting error at inspection.
2. It leads to more uniform quality of production.
3. It improves the relationship with the customer.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
143
4. It reduces inspection costs.
5. It reduces the number of rejects and saves the cost of ma-
terial.
6. It provides a basis for attainable specifications.
7. It points out the bottlenecks and trouble spots.
8. It provides a means of determining the capability of the
manufacturing process.
9. It promotes the understanding and appreciation of quality
control.
Process Control And Product Control
Process Control: The main objective in any production pro-
cess is to control and maintain a satisfactory quality level of the
manufactured product so that it conforms to specified quality
standards. In other words, we want to ensure that the pro-
portion of defective items in the manufactured product is not
too large. This is termed as ‘process control’ and is achieved
through the technique of “Control Charts” pioneered by W.A.
Shewhart in 1924.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
144
Product Control: By product control we mean controlling
the quality of the product by critical examination at strategic
points and this is achieved through ‘Sampling Inspection Plans’
pioneered by H.F. Dodge and H.C. Romig. Product control aims
at guaranteeing a certain quality level to the consumer regard-
less of what quality level is being maintained by the producer.
In other words, it attempts to ensure that the product marketed
by a department does not contain a large number of defective
(unsatisfactory) items. Thus, product control is concerned with
classification of raw materials, semi-finished goods or finished
goods into acceptable or rejectable items.
Control Limits, Specification Limits,
Tolerance Limits
1. Control Limits: These are limits of sampling variation
of a statistical measure (e.g. mean, range, or fraction-
defective) such that if the production process is under con-
trol, the values of the measure calculated from different ra-
tional sub-groups will lie within these limits. Points falling
outside control limits indicate that the process is not oper-
ating under system of chance causes, i.e., assignable causes
of variation are present, which must be eliminated. Con-
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
145
trol limits are used in ‘Control Charts’. Control limits,
also known as natural process limits, are horizontal lines
drawn on a statistical process control chart, usually at a
distance of ±3 standard deviations of the plotted statistic
from the statistic’s mean.
2. Specification Limits: When an article is proposed to be
manufactured, the manufacturers have to decide upon the
maximum and the minimum allowable dimensions of some
quality characteristics so that the product can be gainfully
utilised for which it is intended. If the dimensions are
beyond these limits, the product is treated as defective and
cannot be used. These maximum and minimum limits of
variation of individual items, as mentioned in the product
design, are known as ‘specification limits’.
3. Tolerance Limits: Variation due to chance causes is
inherent in the outputs under any scheme of production.
These are limits of variation of a quality measure of the
product between which at least a specified proportion of
the product is expected to lie (with a given probability),
provided the process is in a state of statistical quality con-
trol. Almost all the variations due to chance causes lie
within µ ± 3σ limits and those limits are called natural
tolerance limits. The width 6σ is called the natural toler-
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
146
ance.
Control limits describe what a process is capable of produc-
ing (sometimes referred to as the “voice of the process”), while
tolerances and specifications describe how the product should
perform to meet the customer’s expectations (referred to as the
“voice of the customer”).
Acceptance Sampling
A buyer draws a sample of a few units from a lot that is to be
accepted by him if the number of defectives is less otherwise the
lot is to be rejected. This technique of sampling inspection and
decision making is called acceptance sampling.
Variables and Attributes
The characteristics of a product which are measurable and can
be expressed in their respective units of measurement are called
variables. For example, length, temperature, time etc.
The characteristics of a product which are not measurable
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
147
but can be identified by their presence or absence are called
attributes. This applies to things that can be judged by visual
examination. For example, spots in enamel painting, crack on
a glass tumbler etc.
Control Charts
Control charts are used to identify whether a manufacturing pro-
cess is under statistical control, that is, it governed by chances
causes alone. If there is an influence of assignable cause, neces-
sary actions should be taken to bring the process under control.
This technique was introduced by Dr. Walter A. Shewart. It is
based on the theory of probability and sampling.
Control charts consists of 3 horizontal lines besides the usual
x and y axes, which are the central line and two control limits
on either side of central line.
1. Central Line: Central line (C.L.) denotes the average
value of the characteristics, which is drawn at the mean
of the observations.
2. Upper Control Limit: Upper control limit (U.C.L.) is
at 3σ distance above the central line.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
148
3. Lower Control Limit: Lower control limit (L.C.L.) is
at 3σ distance below the central line. If the 3σ distance
below the central line is below the x - axis, then the Lower
control limit is the x - axis.
A chart is drawn for one statistical measure of observations.
If this measure follows normal law, 99.73% of the observations
are expected to lie with in the control limits. Otherwise, 8/9 of
the observations are expected to lie with in the control limits.
Hence, control charts protect against both types of errors - not
noticing the trouble when there is and looking for trouble when
there is not.
At a time, a few units, say n, are chosen. Those units form
one subgroup. On the basis of all the subgroups, the chart is
drawn with a central line and the control limits. Subgroup or
sample numbers are represented in x - axis and the measures of
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
149
subgroups of observations are represented in y - axis. For each
subgroup or sample one point is plotted. The chart contains
the points of all the subgroups. If any point lies in the region
outside control limits, called out of control region, it indicates
that there is some assignable cause of variation and the process
is not under statistical control.
Standards Known: Sometimes the nature of the process
may be known from the past. On that basis the control limits
can be found out and the present sample need be considered.
When the past standards are not known the control limits are
found on the basis of the sample under consideration. We will
be considering this kind of charts.
Revised Control Limits: From a sample under consider-
ation those items which fall in the ‘out of control’ region are
deleted and then the control limits are found again. Those con-
trol limits are the revised control limits.
Rational Subgroup: A rational subgroup is a group of
units produced under the same set of conditions. Rational sub-
groups are meant to represent a “snapshot of the process. They
also reflect how your data are collected, and represent the in-
herent (common cause) variation in your process at any given
time.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
150
For many processes, you can form rational subgroups by
sampling multiple observations that are close together in time,
but still independent of each other. For example, a die cut ma-
chine produces 100 plastic parts per hour. The quality engineer
measures five randomly selected parts at the beginning of every
hour. Each sample of five parts is a subgroup.
Types of Control Charts
There are two types of control charts:
1. Control charts for variables
2. Control charts for attributes.
Control chart for variables: As we have defined earlier,
variables are the quality characteristics of manufactured prod-
ucts which are measurable in their respective units of measure-
ments. These types of variables are continuous in nature and
they follow the Normal distribution. For this purpose we use
two Charts:
1. the chart for mean (X̄) and range (R)
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
151
2. the chart for mean (X̄) and standard deviation (σ)
Control chart for Attributes To control the quality of
products which are governed by the attributes we use, the p -
chart for proportion of defectives, the np - chart for number of
defectives and the c - chart for number of defects per item or
the unit.
3σ Control limits
3σ limits were proposed by Dr. Shewhart for his control charts
from various considerations, the main being probabilistic con-
siderations. Consider the statistic T = {X1 , X2 , . . . , Xn }. Let
E(T ) = A and V (T ) = σ 2 . If the statistic T is normally dis-
tributed, then from the fundamental area property of the normal
distribution, we have
P [µ − 3σ < T < µ + 3σ] = 0.9973
or
P [|T − µ| > 3σ] = 0.0027.
Here µ − 3σ is called L.C.L. and µ + 3σ is called U.C.L. If for
the ith sample, the observed T lies between L.C.L. and U.C.L.,
there is nothing to worry, otherwise a danger signal is indicated.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
152
4.2 Control Charts for Variables
These charts may be applied to any quality characteristic that
is measurable. In order to control a measurable characteristic
we have to exercise control on the measure of location as well
as the measure of dispersion. Usually X̄ and R charts are em-
ployed to control the mean and dispersion respectively of the
characteristic.
X̄ - Chart
No production process is perfect enough to produce all the items
exactly alike. Some amount of variation, in the produced items,
is inherent in any production scheme. This variation is the to-
tality of numerous characteristics of the production process viz.,
raw material, machine setting and handling, operators, etc. As
pointed out earlier, this variation is the result of (i) chance
causes, and (ii) assignable causes. The control limits in the
X̄ chart are so placed that they reveal the presence absence of
assignable causes of variation in the average - mostly related to
machine setting.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
153
Procedure for the Construction X̄ - Chart
1. Compute the mean of each sample of the same size n, say
x̄1 , x̄2 , . . . , x̄k , where k denote the number of samples.
x̄1 +x̄2 +...+x̄k
¯=
2. Compute the C.L., C.L.X̄ = x̄ k
3. Compute the range or standard deviation (as required) of
each sample of the same size n, say R1 , R2 , . . . , Rk and
s1 , s2 , . . . , sk respectively.
R1 +R2 +...+Rk s1 +s2 +...+sk
4. Compute R̄ = k or s̄ = k as re-
quired.
5. Obtain the U.C.L. and L.C.L. using the appropriate for-
mula.
√
(a) When µ = µ0 and σ = σ 0 are known, and A = 3/ n,
U.C.L.x̄ = µ0 + Aσ 0
L.C.L.x̄ = µ0 − Aσ 0
(b) When standards are not given
i. If both µ and σ are unknown, using their esti-
√
¯ and σ̂ = dR̄ and A2 = 3/(d2 n), then
mates x̄
2
¯ + A2 R̄
U.C.L.x̄ = x̄
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
154
¯ − A2 R̄
L.C.L.x̄ = x̄
ii. If the limits are to be obtained in terms of s̄ sam-
√
ple rather than R̄, σ̂ = Cs̄ and A1 = 3/(C2 n),
2
then
¯ + A1 s̄
U.C.L.x̄ = x̄
¯ − A1 s̄.
L.C.L.x̄ = x̄
R - Chart
R - chart shows variability within the process. The control limits
in the R - chart are so placed that they reveal the presence
absence of assignable causes of variation in the range - mostly
related to negligence on the part of the operator.
Procedure for the Construction R - Chart
1. Compute the range of each sample of the same size n, say
R1 , R2 , . . . , Rk , where k denote the number of samples.
R1 +R2 +...+Rk
2. Compute the C.L., C.L.R = R̄ = k
3. The 3 - σ control limits for R - chart are E(R) ± 3σR ,
where E(R) is estimated by R̄
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
155
(a) When σ is known,
U.C.L.R = D2 σ
L.C.L.R = D1 σ
(b) When σ is unknown, σR is estimated from the rela-
tion σR = d3 σ̂ = d3 dR̄ ,
2
U.C.L.R = D4 R̄
L.C.L.R = D3 R̄.
s or σ - Chart
Since standard deviation is an ideal measure of dispersion, a
combination of control chart for mean (X̄) and standard devi-
ation (s), known as X̄ and s - charts (or X̄ and σ - charts) is
theoretically more appropriate than a combination of X̄ and R
- charts for controlling process average and process variability.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
156
Procedure for the Construction σ - Chart
1. Compute the standard deviation of each sample of the
same size n, say s1 , s2 , . . . , sk , where k denote the number
of samples.
s1 +s2 +...+sk
2. Compute the C.L., C.L.s = s̄ = k
3. The 3 - σ control limits for s - chart are E(s) ± 3σs ,
where E(s) is estimated by s̄
(a) When σ is known,
U.C.L.R = B2 σ
L.C.L.R = B1 σ
(b) When σ is unknown, σs is estimated from the relation
σs = C3 σ̂ = C3 Cs̄ ,
2
U.C.L.R = B4 s̄
L.C.L.R = B3 s̄.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
157
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
158
4.3 Control Charts for Attributes
In spite of wide applications of X̄ and R -(or σ) charts as a
powerful tool of diagnosis of sources of trouble in a production
process, their use is restricted because of the following imita-
tions.
1. They are charts for variables only, i.e., for quality charac-
teristics which can be measured and expressed in numbers.
2. In certain situations they are impracticable and un-
economical.
As an alternative to X̄ and R -charts, we have the control
chart for attributes which can be used for quality characteristics:
1. which can be observed only as attributes by classifying
an item as defective or non defective i.e., conforming to
specifications or not
2. which are actually observed as attributes even though they
could be measured as variables, eg., go and no-go gauge
test results.
There are two control charts for attributes
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
159
1. Control chart for fraction defective (p - chart) or the num-
ber of defectives (np - chart or d - chart).
2. Control chart for the number of defects per unit (c - chart).
p - chart
p - chart is the most widely used control chart for attributes.
This chart is for the fraction defectives (fraction rejected as not
attaining the standards specified). The objective of p - chart
is to ascertain and to evaluate the quality of the items and to
observe the changes in quality over a period of time. This is an
immediate guide for correcting the causes of bad quality.
While dealing with attributes a process will be adjudged in
statistical control if all the samples or sub-groups are ascertained
to have the same population proportion p.
If ‘d’ is the number of defectives in a sample of size n, then
d is a binomial variate with parameters n and p. Therefore,
E(d) = np and V (d) = npq, q = 1 − p.
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
160
d
Let the sample proportion of defective is p = n. Then,
1 1
E(p) = E(d) = np = p
n n
and
1 1 pq
V (p) = V (d) = 2 npq = , q = 1 − p.
n2 n n
Procedure for the Construction p - Chart
1. Compute the fraction defectives of each sample of size n,
say p1 , p2 , ldots, pk , where k denotes the number of sam-
ples.
p1 +p2 +...+pk
2. Compute the C.L., C.L.p = p̄ = k
3. The 3 - σ control limits for p - chart are E(p) ± 3σp ,
where E(p) is estimated by p̄. Therefore,
r
p̄(1 − p̄)
U.C.L.p = p̄ + 3
n
r
p̄(1 − p̄)
L.C.L.p = p̄ − 3
n
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
161
np - chart or d - chart
If instead of p, the sample proportion defective, we use d, the
number of defectives in the sample, then we have the d - chart.
The number of defectives d = np will always be an integer. If
the number of items inspected is the same on each occasion,
then plotting the number of defectives (d) is more convenient
than plotting the fraction defectives (p). in navethep chart, if
n = 100 then we have the percentage defective chart or the 100
p - chart.
Procedure for the Construction d - Chart
1. Compute the number of defectives of each sample of size n,
say d1 , d2 , ldots, dk , where k denotes the number of sam-
ples.
d1 +d2 +...+dk
2. Compute the C.L., C.L.d = d¯ = np̄ = k
3. The 3 - σ control limits for p - chart are E(d) ± 3σd ,
where E(d) is estimated by d¯ = np̄. Therefore,
p
U.C.L.d = np̄ + 3 np̄(1 − p̄)
p
L.C.L.d = np̄ − 3 np̄(1 − p̄)
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
162
Remark 4.3.1. p and d - Charts for Fixed Sample Size
If the sample sizes remains constant for each sample (as we
¯
considered) p̄ = nd .
c - Chart
In many manufacturing or inspection situations, the sample size
n i.e., the area of opportunity is very large (since the opportu-
nities for defects to occur are numerous) and the probability p
of the occurrence of a defect in any one spot is very small such
that np is finite. In such situations from statistical theory we
know that the pattern of variations in data can be represented
by Poisson distribution, and consequently 3 − σ control limits
based on Poisson distribution are used. Since for a Poisson dis-
tribution, mean and variance are equal, if we assume that c, the
number of defects, is Poisson variate with parameter, λ, we get
Procedure for the Construction c - Chart
1. Compute the number of defects of each unit, say
c1 , c2 , . . . , ck , where k denotes the number of units
(samples).
c1 +c2 +...+ck
2. Compute the C.L., C.L.c = c̄ = k
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
163
3. The 3 - σ control limits for c - chart are E(c) ± 3σc ,
√
where E(c) is estimated by c̄ and σc by c̄. Therefore,
√
U.C.L.c = c̄ + 3 c̄
√
L.C.L.c = c̄ − 3 c̄
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
164
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
BIBLIOGRAPHY
[1] Rohatgi V. K. and Saleh, A.K. Md. E. (2009): An Intro-
duction to Probability and Statistics. 2nd Edn. (Reprint),
John Wiley & Sons.
[2] Gupta, S.P. Statistical Methods, Sultan Chand and Sons:
New Delhi.
[3] S.C.Gupta and V. K. Kapoor, Fundamentals of Mathemat-
ical Statistics, Sultan Chand & Sons.
[4] Mood, A.M. Graybill, F.A. and Boes, D.C. (2007): Intro-
duction to the Theory of Statistics, 3rd Edn., (Reprint),
Tata McGraw-Hill Pub. Co. Ltd.
165
School of Distance Education, University of Calicut
STA4C04 : Statistical Inference and Quality Control
166
[5] John E. Freund, Mathematical Statistics, Pearson Edn,
New Delhi.
[6] Grant E.L., Statistical quality control, McGraw Hill.
[7] Montogomery, D.C. (2009): Introduction to Statistical
Quality Control, 6th Edition, Wiley India Pvt. Ltd.
[8] K.X. Joseph: Subsidiary Statistics.
School of Distance Education, University of Calicut
You might also like
- System Dynamics and Respinse Kelly Solutions100% (6)System Dynamics and Respinse Kelly Solutions447 pages
- 7QC ProcessImprovementVariationProblem Solving and ToolsNo ratings yet7QC ProcessImprovementVariationProblem Solving and Tools106 pages
- Lesson - 5.1 - Design of Experiments - Improve - PhaseNo ratings yetLesson - 5.1 - Design of Experiments - Improve - Phase39 pages
- 18MEO113T - DOE - Unit 1 - AY2023-24 ODDNo ratings yet18MEO113T - DOE - Unit 1 - AY2023-24 ODD120 pages
- 20 Reliability Testing and VerificationNo ratings yet20 Reliability Testing and Verification13 pages
- What Is Data? Data Is A Set of Values of Subjects With Respect To Qualitative or Quantitative VariablesNo ratings yetWhat Is Data? Data Is A Set of Values of Subjects With Respect To Qualitative or Quantitative Variables10 pages
- Chapter5 Statistical Process Control Questions100% (1)Chapter5 Statistical Process Control Questions3 pages
- Seven QC Tools Tool #5: Part 1-Run ChartNo ratings yetSeven QC Tools Tool #5: Part 1-Run Chart6 pages
- X-Bar and R Charts: NCSS Statistical SoftwareNo ratings yetX-Bar and R Charts: NCSS Statistical Software26 pages
- 8 - Analyze - Hypothesis Testing Non Normal Data - P2No ratings yet8 - Analyze - Hypothesis Testing Non Normal Data - P237 pages
- 5) DOE Design and Analysis Using MinitabNo ratings yet5) DOE Design and Analysis Using Minitab48 pages
- Probability Distribution: Shreya Kanwar (16eemme023)No ratings yetProbability Distribution: Shreya Kanwar (16eemme023)51 pages
- Graded Quiz - Using Probability Distributions - CourseraNo ratings yetGraded Quiz - Using Probability Distributions - Coursera10 pages
- Henry-1473226926-Lean Six Sigma Black BeltNo ratings yetHenry-1473226926-Lean Six Sigma Black Belt8 pages
- Process Capability Analysis and Process Analytical TechnologyNo ratings yetProcess Capability Analysis and Process Analytical Technology43 pages
- Statistical Analysis in Finance Session 4: Hypothesis TestingNo ratings yetStatistical Analysis in Finance Session 4: Hypothesis Testing32 pages
- A Statistical Process Control Case Study: Thomas K. Ross, PHDNo ratings yetA Statistical Process Control Case Study: Thomas K. Ross, PHD16 pages
- Interpret The Key Results For Attribute Agreement Analysis100% (1)Interpret The Key Results For Attribute Agreement Analysis28 pages
- SPC Charts - Statistical Process Control Charts PDFNo ratings yetSPC Charts - Statistical Process Control Charts PDF6 pages
- Making Sustainability Work: Best Practices in Managing and Measuring Corporate Social, Environmental, and Economic ImpactsFrom EverandMaking Sustainability Work: Best Practices in Managing and Measuring Corporate Social, Environmental, and Economic ImpactsNo ratings yet
- Introduction To Inferential Statistics: Jomel R. AlanzalonNo ratings yetIntroduction To Inferential Statistics: Jomel R. Alanzalon36 pages
- 1.1 Parametric and Nonparametric Statistical InferenceNo ratings yet1.1 Parametric and Nonparametric Statistical Inference8 pages
- Basics of Inference: Department of Statistics Ram Lal Anand CollegeNo ratings yetBasics of Inference: Department of Statistics Ram Lal Anand College34 pages
- Autoregressive Integrated Moving AverageNo ratings yetAutoregressive Integrated Moving Average10 pages
- Modern Optimized Transportation Solutions For Public Transport Using Agile MethodologyNo ratings yetModern Optimized Transportation Solutions For Public Transport Using Agile Methodology5 pages
- Glossary of Mathematical Terms (CSTM 0120)No ratings yetGlossary of Mathematical Terms (CSTM 0120)6 pages
- Alexandra Maceda - TASK SHEET 3.3-2 Interpret and Analyze Program EvaluationNo ratings yetAlexandra Maceda - TASK SHEET 3.3-2 Interpret and Analyze Program Evaluation4 pages
- Jakobsson, - Rmann - Jakobsson, Sverrir - The Routledge Research Companion To The Medieval Icelandic Sagas-TaylorNo ratings yetJakobsson, - Rmann - Jakobsson, Sverrir - The Routledge Research Companion To The Medieval Icelandic Sagas-Taylor377 pages
- 7.2.4.3 Lab - Using Wireshark To Examine FTP and TFTP Captures0% (1)7.2.4.3 Lab - Using Wireshark To Examine FTP and TFTP Captures14 pages
- Compilers: Tools For Scientists and EngineersNo ratings yetCompilers: Tools For Scientists and Engineers42 pages
- Pwa Buildings Cad Standards Manual Ver 2.0100% (1)Pwa Buildings Cad Standards Manual Ver 2.077 pages
- Linear Vibration Analysis of Cantilever Plates Partially Submerged in FluidNo ratings yetLinear Vibration Analysis of Cantilever Plates Partially Submerged in Fluid13 pages
- 1426357293.3273chapter 2 In-Class ExercisesNo ratings yet1426357293.3273chapter 2 In-Class Exercises12 pages
- The Importance of Waste Management Knowledge To Encourage Householdwastesorting Behaviour in Indonesia 2252 5211 1000309No ratings yetThe Importance of Waste Management Knowledge To Encourage Householdwastesorting Behaviour in Indonesia 2252 5211 10003095 pages
- Sta 711: Homework 1 Solutions: Fields and σ-fieldsNo ratings yetSta 711: Homework 1 Solutions: Fields and σ-fields3 pages
- Adventist University of The Philippines College of Nursing Intervention PlanNo ratings yetAdventist University of The Philippines College of Nursing Intervention Plan3 pages
