Sampling distributions

TOPICS:

 Sampling distribution of ¿

 Sampling distribution of ( p )

 Sampling distribution of (x− y )

 Sampling distribution of ( p1− p2 )

1 Sampling distribution of sample mean, x :

Let x 1 , x 2 ,… … . x n be a random sample (iid) of size n is drawn from a

population with mean μ and variance σ 2.

Then, the sampling distribution of sample mean is

2
σ
x N (μ , ).
n

CSE -200
2
POPULATION μ,σ

Sample (SRSWR/SRSWOR)
X 1 , X 2 , X 3 ,… … … … , X n(ls).

Sample mean , X
 T, Estimator

So that, we compute, E ( T )∧V (T )

Since x 1 , x 2 ,… … . x n be random sample (iid) of size n,

therefore
E ( X 1) =E ( X 2 ) =… … … … .. E ( X n ) =μ

2
V ( X 1 ) =V ( X 2 )=… … … … ..V ( X n )=σ

If the population mean, μ is unknown, it is estimated by sample mean,

x in LS and SS

∴ We need to find E ( T )=E( X)

E ( X )=E { X 1 + X 2+ … … … . X n
n }
1
¿
n
[ E ( X 1 ) + E ( X 2 ) + … … … … .. E ( X n ) ]
1
¿ [ μ+ μ+… … … … . μ ]
n

1
¿ (nμ)
n

¿μ

E ( X )=μ , sample mean x is an unbiased estimator of POP. Mean μ.

And V ( X )=V { X 1 + X 2 +… … … . X n
n }
()
2
1
¿ . V ( X 1+ X 2+ … … … . X n )
n
2
∴ V ( α + βX )=β V ( X)
 1
2[
¿ V ( X 1 ) +V ( X 2 ) +… … … … .. V ( X n ) ]
n

∴( X ' s is indep.)

1 2 2
¿ 2
[ σ + σ +… … … … . σ 2 ]
n
2
1 ( 2) σ
¿ 2
nσ =
n n
2
σ
⇒ V ( X )=
n

If the sample size n is large (as n → ∞ ¿ according to central limit

theorem, all most all distributions converges to normal dist.
Therefore as → ∞ ,
2
σ
X N (μ , ).
n

Note:
2
σ
We have, The sampling distribution of sample mean is x N (μ , )
n

σ
1 Standard error of x , SE( x )=
√n
2 If n → ∞ , SE ( x ) → 0

3 If σ 2 is unknown, it is estimated by s2 in LS, then sampling

s
2
s
distribution of sample mean is x N (μ , ) and SE( x )=
n √n

4 If σ 2 is unknown, it is estimated by S2 in SS, then sampling

distribution of sample mean is for from the normality and it follows
S
students t- distribution and SE( x )=
√ n−1
5 The Standardized normal variable (Test Statistic),
 T −E (T )
Z= N (0 ,1)
SE (T )

x −μ
Z= N (0 , 1)
σ
√n
x −μ
Z= N (0 , 1)
s , If σ 2 is unknownis LS
√n
x−μ
t= t n−1 ,
S 2
If σ is unknownis SS
√n

6 The General form of confidence limits or Interval estimation for μ,

T ∓ Z α SE ( T ) , for LS

T ∓t α SE ( T ) , for SS

Where Z α ∧t α ( Statistical table value )

Z α =1.96∧2.58 respectively for 5 %∧1 % lelel of significance

95% Control limits: [ ¿ , UL ] = [ T −Z α SE (T ) , T + Z α SE ( T ) ] , for LS

[ T −1.96 SE (T ) , T +1.96 SE ( T ) ] , for LS

[ x−1.96 SE ( T ) , x+1.96 SE ( T ) ] , for LS

[ x−1.9 6
s
√n
, x +1.96
s
√n
,
] for LS

99% Control limits: [ ¿ , UL ] = [ T −Z α SE (T ) , T + Z α SE ( T ) ] , for LS

[ T −2.58 SE (T ) , T +2.58 SE ( T ) ] , for LS

[ x−2.58 SE ( T ) , x+2.58 SE ( T ) ] , for LS
 [ x−2.58
s
√n
, x +2.58
s
√n
,
] for LS

7 The Maximum error of the Estimator T is

E=Z α SE ( T ), for LS
E=t α SE ( T ) , for SS

3 Sampling distribution of ( x− y ) :

CSE -B
CSE -A
2 POPULATION
POPULATION μ1 , σ 1 2
μ2 , σ 2

Sample (SRSWR/SRSWOR) Sample

(SRSWR/SRSWOR)
X 1 , X 2 , X 3 ,… … … … , X n .
1
Y 1 ,Y 2 , Y 3 , … … … … ,Y n .
2

Sample mean , x Sample mean , y

2 2
Sample Variance , s 1 Sample Variance , s 2
2 2
σ1 σ2
x N (μ 1 , ) y N (μ2 , )
n1 n2
Let x , y ; s21 , s 22 be sample means, sample variances of a r . Samples
x 1 , x 2 ,… … . x n and y 1 , y 2 , … … . y n of size n1∧n2 (LS) are drown from two indep
1 2

population with weans μ1∧μ2 and variance σ 21∧σ 22 respectively, then

For n1∧n2 , and CLT, we have

E(X ± Y ¿=E ( X ) ± E(Y )

and
V(X ± Y ¿=V ( X ) +V (Y )

Now the sampling distribution of difference of sample means

2 2
σ σ
( x− y ) N ¿ , 1 + 2 ¿
n 1 n2
2 2
s1 s 2
( x− y ) N ¿ , + ¿, if Pop Variances are unknown
n1 n2

The Standardized normal variable (Test Statistic),

T −E (T )
Z= N (0 ,1)
SE (T )

( x − y )−( μ1−μ 2)
Z= N (0 , 1)

√
2 2
s1 s 2
+
n1 n2

Testing of Hypothesis

Mean Variance proportion

Population

2
μσ P

unknown
 WR/WOR
¿ 30 , SS / ¿ 30 , LS

Sample X 1 , X 2 , X 3 ,… … … … X n

Estimator x : sample mean

T
(Statistic) 2
s :sample variance

p :sample proportion

2
S :mean square

Reliability of the estimator, T =?

So that , The reliability of the estimator for estimating the unknown

population parameter, at a given confidence of ( 1−α ) % leads the
concept of testing of hypothesis

Basic definitions

Null hypothesis

Alternative hypothesis (One – Tailed alternative & Two – Tailed

alternative)

Level of significance
Type-I & Type –II errors

Critical / rejection region

Power of the test

1 Problem :

A population consists of five numbers 2 , 11, 6, 8, 3. Consider all

possible samples of size 2 that can be drawn from this population.

Determine the following by using the sampling techniques of simple

random sample with replacement (SRSWR) and simple random
sample without replacement (SRSWOR).

 1 The mean of the population

 2 The variance of the population
 3 The S.D of the population.
 4 The mean of the sample
 5 The variance of the sample
 6 The S.D of the sample
 7 The mean square of the sample
 8 Sampling distribution of sample mean.
 9 The S.E of sample mean of sampling distribution.

SOLUTION:
POPULATION

Y 1=¿ 2, Y 2=¿ 3, Y 3=¿ 6 ,

1 Population Mean , μ,
Y 4 =¿8, Y 5=¿11
Given that
population size N = 5,

population units (Obvers ions) : 2 ,3, 6, 8,11

Y 1 +Y 2 +−−−−−Y N
now, population mean μ=
N

2+ 3+6+8+ 11 30
¿ = =6
5 5

μ=6

2 Population Variance, σ 2
N
1
σ = ∑ ( Y i−μ )
2 2

N i=1
5
1
σ = ∑ ( Y i−μ )
2 2

5 i=1

2 1
⇒ σ = 5 {( 2−6 ) + ( 3−6 ) + ¿ ( 6−6 ) + ( 8−6 ) + ( 11−6 ) }
2 2 2 2 2

2 1
σ = { 16+9+ 0+4 +25 }
5

1
¿ ( 54 )=10.8
5
2
σ =10.8

3 Population SD, σ

σ =√ PopulationVariance , σ 2

¿ √ 10.8

σ =3.29

POP {x , y , z}
{ } { }
( x , x ) , ( x , y ) , (x , z ) , ( x , y ) ,( x , z )
¿( y ,x),( y, y),( y, z) ¿ , ,( y , z )
¿ ( z , x ) , ( z , y ) , (z , z ) ¿,,

WR WOR , (32)=( Nn )=3

N=3

n= 2 , 9=32=Nn

CASE - 1 simple random sample with replacement (SRSWR)

POP. Size N = 5
Y 1=¿ 2, Y 2=¿ 3, Y 3=¿ 6 ,
Sample size n = 2
Y 4 =¿8, Y 5=¿11

SRSWR
∴ The no of samples N n=5 2=25(LS)

Samples:
 Mean of each sample X 1 =2 X 2 =2.5 X 3 =4 X 4 =5 X 5 =6.5

sample (2,2) (2,3) (2,6) (2,8) (2,11)

Mean of each sample X 6 =2.5 3 4.5 55 7

sample (3,2) (3, 3) (3,6) (3,8) (3,11)

Mean of each sample 4 4.5 6 7 8.5

sample (6,2) (6,3) (6,6) (6,8) (6,11)

Mean of each sample 5 5.5 7 8 9.5

sample (8,2) (8,3) (8,6) (8,8) (8,11)

Mean of each sample 6.5 7 X 23 =8.5 X 24 =9.5 X 25 =11

sample (11,2) (11,3) (11,6) (11,8) (11,11)

Now, we can compute Sample Characteristics (Estimators/

statistics)

4 sample mean (mean of means each sample) is

x=
∑ of the obs
no of obs

X 1 + X 2 +… … … . X n −wr
¿
n wr

2+ 2.5+4 … … … .+9.5+11
¿
25

150
¿
25

¿6
 ⇒ x=6 (¿ μ=Population mean)

⇒ sample mean , x , isUBE for Population mean , μ .

5 Sample variance, s2:

n
1
s2= ∑ ( x −x )2
n i=1 i
25
1
s = ∑ ( x i−6 )
2 2

25 i=1

{ }
( 2−6 )2+ (2.5−6 )2+ … … … . ( 6.5−6 )2
1 ( 2.5−6 )2+ … … … . (7−6 )2
¿ .
25
.
2 2
( 6.5−6 ) +… … … . ( 11−6 ) .

1
¿ [ 135 ]
25
2
s =5.4

6 Sample sd, s :

s= √ sample variance

s= √ s 2
¿ √ 5.4

s=2.32

7 Sample Mean Square, S2:

n−1
1
2
S= ∑
n−1 i=1
( x i−x )
2
 { }
( 2−6 )2 + ( 2.5−6 )2 +… … … . ( 6.5−6 )2
1 ( 2.5−6 )2 +… … … . ( 7−6 )2
¿ .
24
.
2 2
( 6.5−6 ) + … … … . (11−6 ) .

1
¿ [ 135 ] =¿
24
2
S =5.625

8 Sampling distribution of sample mean.

2
σ
We have, x N (μ , )
n

10.8
x N (6 , )
2

x N (6 , 5.4)

T, E(T), V(T)

9 Standard error of sample mean

S.E (( T )=+ √ V (T )

S.E ( ( x )=+ √ V ( x )
¿ √ 5.4

¿ 2.32

Note: If σ is unknown,
2
σ is estmated by s in LS
2 2
 2
s
x N (μ , )
n

5.4
x N (6 , )
2

x N (6 , 2.32)

CASE - 2 simple random sample without replacement (SRSWOR)

Y 1=¿ 2, Y 2=¿ 3, Y 3=¿ 6 ,

POP. Size N = 5
Y 4 =¿8, Y 5=¿11
Sample size n = 2

SRSWOR
N 5
( ) ()
The no. of samples n = 2 =10.

Samples
 Mean of each sample X 1 =2.5 X 2 =4 X 3 =5 X 4 =6.5

sample (2,3) (2,6) (2,8) (2,11)

Mean of each sample 4.5 55 7

sample (3,6) (3,8) (3,11)

Mean of each sample 7 8.5

sample (6,8) (6,11)

Mean of each sample X 10 =9.5

sample (8,11)

Mean of each sample

sample

Now, inSRSWOR,

1 sample mean(mean of means each sample) is

x=
∑ of the obs
no of obs

X 1 + X 2 +… … … . X n
¿
nwor

2.5+4 … … … .+9.5
¿
10

60
¿
10

¿6

⇒ x=6=μ
 ⇒ x is UBE for μ

2 Sample variance :
10
1
s2= ∑ ( x i−x ) 2
n wor i=1

{ }
( 2.5−6 )2+ … … … . ( 6.5−6 )2
2 2
1 ( 2.5−6 ) + … … … . (7−6 )
¿ .
10
.
2 2
( 6.5−6 ) + … … … . ( 9.5−6 ) .

1
¿ [ 45 ] =4.5
10

3 Sample sd : s= √ sample variance

s= √ s 2
¿ √ 4.5

s=2.12

4 Sample Mean Square, S2:

n−1
1
2
S= ∑
(n−1)wor i=1
( x i−x )
2

{ }
( 2.5−6 )2 +… … … . ( 6.5−6 )2
2 2
1 ( 2.5−6 ) +… … … . ( 7−6 )
¿ .
9
.
2 2
( 6.5−6 ) +… … … . ( 9.5−6 ) .

1
¿ [ 45 ] =¿
9
2
S =5
4 Sampling distribution of sample mean.

We have

( )
2
σ
x N μ,
n

(
x N 6,
10.8
2 )
x N (6 , 5.4)

5 Standard error of sample mean

S.E (( T )=√ V (T )

S.E ( ( x )=√ V ( x )
¿ √ 5.4

¿ 2.32

Note: If σ is unknown
2
σ is estmated by S in SS
2 2

( )
2
S
x N μ,
n

x N 6, ( 52 )
x N (6 , 2.5)
Practice problems

1 Let the population is 3, 6, 9, 15 and 27. Consider all samples of size two
which can be drawn wr/without replacement from this population.
Then. find
(a)The population mean
(b) The population standard deviation
(c)The mean of the sampling distribution of means
(d) The standard deviation of the sampling distribution of means.

2 A population consists of six numbers 4, 8, 12, 16, 20, 24. Consider all
samples of size two which can be drawn wr/without replacement from
this population. Then, find
(a)The population mean
(b) The population standard deviation
(c)The mean of the sampling distribution of means
(d) The standard deviation of the sampling distribution of means.

CASE STUDY

A company claims that, the mean thermal efficiency of diesel

engines produced by them is 32.3 %. To test this claim be
accepted or not for a random sample of 40 engines were
examined which showed that the average thermal efficiency of
31.4 % and s.d of 1.6% at 5% level of significance.

Sol : Diesel company

MTE(parameter)

H 0 : μ=32.3
 H 1 : μ ≠ 32.3 % (T −T )

H 1 : μ >32.3 %(R−T )

H 1 : μ <32.3 %( L−T )

Los, α =5 %∨1 %

Sample(wr/wor)

⏟
X , X , X , X , X ,… … … … , X =n ( ¿ 30 LS )
1 2 3 4 5 40

Sample data

Sample size, n=40 ( ¿ 30 , LS )

Sample mean, x = 31.4

Sample SD, s = 1.6

T −E (T )
Under H 0, the test statistic, Z= SE (T ) N (0 ,1)
 x−μ0
Z= N (0 ,1)
s∨√ n

31.4−32.3
⇒ Z cal=
1.6∨ √ 40

−0.9
Z cal=
0.253

Z cal=−3.55

|Z cal|=3.55

Conclusion:

Since |Z 3.55|>3 , we reject H 0.

|Z 3.55|<3 , we do n' t reject H 0

¿ cosider level of significance , α at 5 %∨1 % .

LARGE/ SMALL SAMPLE TEST PROCEDURE TESTING OF

HYPOTHESIS

Let X 1 , X 2 , … … .. X n be a random sample of size n(SS / LS) is drown from a

population with parameters μ∧σ 2.

In practice the parameters are unknown, these are estimated by

sample estimators(T).

Now , we test the hypothesis for testing parameters with following

procedure.

STEP-1 To set up a Null Hypothesis (H ¿¿ 0)¿:

A definite statement about population parameters is called null

hypothesis, H 0.

Example:
 H 0 : μ=20 (Normal distribution)

2
H 0 :σ =5 (Normal distribution)
1
H 0 : P=
3
(Binomial distribution)

1
H 0 :θ=
3
(Exponential distribution)

H 0 : λ=20 (Poisson distribution)

STEP -2 To set up an Alternative Hypothesis ( H 1):

An alternative hypothesis is an hypothesis, which is

complementary of H 0, denoted by H 1.

Every null hypothesis H 0, possess any of the following

alternatives, namely Two – tailed(T-T), right – tailed(R-T) & left –
tailed (L-T)

i.e., H 0 ; μ=μ0

Against The alternative hypothesis

H 1 ; μ ≠ μ0 (T −T )

H 1 ; μ> μ 0 (R−T )

H 1 ; μ< μ 0 (L−T )

STEP – 3

To fix the level of significance α in advance at

α =5 % level∨1 % level

Where the level of significance α is the size of type –I error.

Type –I Error: Pr { Reject H 0 ∨H 0 is true } =α , 0< α <1

Type –II Error: Pr { Accept H 0∨H 0 is false }=β , 0< β <1

Power of the test: the number (1- β )is called power of the test

Note: In industrial Engineering we termed as

Type –I Error is Producer’s risk

Type –II Error is consumer’s risk

STEP – 4

From the sample data, we can collect/compute the following

information

Sample Size, n

Sample mean, x
1
Sample variance( s2 ¿
2
s=
n
∑ ( x−x )2
Sample sd, s
1
Sample mean square ( S2)
2
S=
n−1
∑ ( x−x )2

Sampling distribution

Standard error

STEP – 5 compute the Test statistic:

According to central limit theorem, the test statistic is standard

normal variable for LS is
 T −E (T )
Z= N (0 ,1)
SE (T )

Ex:

1 Test for the single population mean ( H 0: μ=μ 0)

x−μ0
Z= N (0 ,1)
s∨√ n

2 Test for significance difference of two POP mean (or) Test for

equality two POP means H 0 : μ1=μ2

x− y
Z= N (0 ,1)

√ s 21 s 22
+
n1 n2

X +Y n1 p 1+ n2 p2
Where, ^P= n +n = n + n ^
and Q=1− ^
P
1 2 1 2

3 Test for single population mean H 0 : μ=μ 0 ( SS )

x−μ0
t= t n−1
s
√ n−1
4 Test for equality two POP. Means (or) H 0 : μ1=μ2

Test for significance difference of two POP. Means

H 0 : ( μ1−μ2 ) =0(SS)

( x− y )
t= t n +n −2

√ 1 1
1 2
2
S( + )
t
n1 n2

n s +n s
Where S = 1 1 2 2 2
2 2
2
or S =
∑ ( x−x )2 + ∑ ( y− y )2
t t
n1 +n2 −1 n + n −2
1 2

5 Test for equality of two population variances: H 0 :σ 21=σ 22

 2
S1
F= 2
F( n −1 ,n −1) where S21 > S22
S2 1 2

2 2
n s n s
Where S = 1 1 and S22= 2 1
2
1
n1−1 n2−1

STEP – 6 Decision/ Conclusion

From sample data, we can calculate test statistic Z cal and compared
with Z(tab)
α for either rejecting or accepting
H 0.

If |Z cal|>3 , we reject H 0.

Otherwise, ie |Z cal|<3, there exist the following three cases

Z ( tab ) t alues
Alternativ
(Level of Significance) Decision
e Hypo. H 1
α =5 % α =1 %

H 1: μ ≠ μ 0 If |Z cal|≤ Z ( tab ) ,
Z ( tab )=1.96 Z ( tab )=2.58 Don’t reject H 0, at
(T −T ) α los

H 1: μ> μ0 If Z cal ≤ Z ( tab ) ,

Null Hypothesis

Z ( tab )=1.645 Z ( tab )=2.52 Don’t reject H 0, at

(R−T ) α los

H 1: μ< μ0
If Z cal ≤ Z ( tab ) ,
Z ( tab )=−1.645 Z ( tab )=−2.52
(L−T ) Reject H 0, at α los
:

Problem-1

To determine whether the mean breaking strength of synthetic fiber

produced by a certain company is 8 kg or not. A random sample of 50
fibers was tested yielding a mean breaking strength of 7.8 kg and S.D 0.5
kg.
 a) Test the hypothesis at 5% level of significance.
b) Determine the 95% & 99% confidence limits for the population
mean
c) Maximum error of T.

Synthetic fiber

Sol

Points : H0 :
MBS μ=8 ( μ 0 )

Sample size n=50(¿ 30 , LS) H1:

μ ≠ 8(T −T )

Single pop.

Quantitative measure (Variable) sample

x−μ0
Test statistic, Z= ⏟
X , X ,………… X
1 2 50=n()
s∨√ n

Sample mean, x = 7.8

Sample SD, s = 0.5

Now the test procedure is

Step-1 H 0 : The mean breaking strength of synthetic fibre is 8 kg

H 0: μ=8(μ0 )

Step-2 H 1 : The mean breaking strength of synthetic fibre is not equal

to8 kg
H 1 : μ ≠ 8 kg

Step 3: Level of significance, α

(tab) (tab)
α =5 % , Z 0.05 =Z α =1.96

Step-4 Sample data :

 Sample size , n=50(¿ 30 , LS)

Sample mean, x=7.8

Sample SD, s=0.5

Step- 5 Under H 0, the test statistic,

x−μ0
Z=
s∨√ n

7.8−8
¿ =−2.82
0.5
√50
|Zcal|=2.83

Step 6 : Decision :

Since |Z (2.82|>Z (1.96

cal ) tab )
, we reject H 0 at 5 % LOS .

Thus, the mean breaking strength of synthetic fibres is not 8 kg.

b) Control limits or Interval estimation

The General form of confidence limits or Interval estimation for μ,

T ± Z α SE (T ) , for LS

T ± t α SE ( T ) , for SS

Where Z α ∧t α ( Statistical table value )

Z α =1.96∧2.58 respectively for 5 %∧1 % lelel of significance

Now,

If x 1 , x 2 ,… … . x n be a random sample (iid) of size n is drawn from a

population with mean μ and variance σ 2.

Then, the sampling distribution of sample mean is

 2
σ
x N (μ , ).
n

The General form of confidence limits or Interval estimation for μ,

T ± Z α SE (T ) , for LS

T ± t α SE ( T ) , for SS

Where Z α ∧t α ( Statistical table value )

Z α =1.96∧2.58 respectively for 5 %∧1 % lelel of significance

95% Control limits:

[ ¿ , UL ] = [ T −Z α SE (T ) , T + Z α SE ( T ) ] , for LS

=[ T −1.96 SE (T ) , T +1.96 SE ( T ) ] , for LS

=[ x−1.96 SE ( T ) , x+1.96 SE ( T ) ] , for LS

[ s
= x−1.96 , x +1.96
√n
s
√n
, for LS
]
[
= 7.8−(1.96)
0.5
√ 50
, 7.8+(1.96)
0.5
√50
,
] for LS

=[7.66, 7.93 ]

99% Control limits:

[ ¿ , UL ] = [ T −Z α SE (T ) , T + Z α SE ( T ) ] , for LS

=[ T −2.58 SE (T ) , T +2.58 SE ( T ) ] , for LS

=[ x−2.58 SE ( T ) , x+2.58 SE ( T ) ] , for LS

[ s
= x−2.58 , x +2.58
√n
s
√n
, for LS
]
[
= 7.8−(2.58)
0.5
√ 50
, 7.8+(2.58)
0.5
√50
,
] for LS
 =[7.61, 7.98]

c) Maximum Error

The Maximum error of the Estimator T is

E=Z α SE ( T ), for LS
E=t α SE ( T ) , for SS

Now,
E=Z α SE ( T ), for LS

¿ 1.96 SE (T ) at α =5 %
s 0.5
¿ ( 1.96 ) =(1.96) = 0.13
√n √ 50
And Similarly,

Maximum Error
E=Z α SE ( T ), for LS

¿ 2.58 SE (T ) at α =1 %
s 0.5
¿(2.58) = (2.58) =0.18
√n √ 50

Problem-2 :

Ten individuals are chosen at random from the population and their
heights are found to be in inches 63 , 64 , 63 ,65 ,66 , 70, 69, 70, 71 & 69.
Is it reasonable to believe that the average height 68 inches ?

a) Test the hypothesis at 5% level.

b) Determine the 95% & 99% confidence limits for the population
mean
c) Maximum error
 POP

Sol: H0
μ=68(μ¿¿ 0)¿
H1 μ ≠ 68

⏟
X , X , … … … … X =n (ss )
1 2 10
63 ,64 ,… … …… ….. 69

Sample mean, x
Sample SD, s
x−μ0
t= t n−1
s
√ n−1
Step 1: H 0 : The mean height of individual is 68

i.e., H 0 : μ=68 ( μ0)

Step 2: H 1 : The mean height is Not equal to 68

H 1 : μ ≠ 68(T −T )

Step 3 LOS, α =5 %, and n = 10(ss)

(tab) (tab)
t (n−1 ,α /2) =t (10−1 , 0.05/ 2) = 2.262

Step 4 From sample data

Sample size , n = 10 (¿ 30 , SS ¿.

And calculate,

Sample mean, x=
∑x
n

1 1
Sample variance, s = n ∑ ( x−x ) = n ∑ x −x
2 2 2 2

Sample SD, s= √ s2
 S.No Sample, X X
2

1 63 3969

2 64 4096

3 63 3969

4 65 4225

5 66 4356

6 70 4900

7 69 4761

8 70 4900

9 71 5041

10 69 4761

Tota 10 10

∑ x =¿670 ∑ x 2=¿ 44978

l i=1 i=1

Sample mean, x=
∑ x = 670 =67
n 10

1
Sample variance, s = n ∑ x −x
2 2 2

1
¿ ( 44978 )−67 2=8.8
10

Sample SD, s= √ s2 =√ 8.8=2.96

Step 5 Under H 0 , the test statistics for SS

x−μ0
t= t n−1
S /√n

(Or)
 x−μ0
t= t n−1
s
√ n−1
67−68
⇒ t cal=
2.96∨√ 10−1

= 1.01
⇒|t cal|=1.01

Step 6 Decision

Since |t cal|1.02 ≤t (tab )2.262, Don’t reject H 0 at 5% los

Hence, The average height of an individual is 68 inches

Note:
1 1
2
s=
n
∑ ( x−x )2 and
2
S=
n−1
∑ ( x−x )2
⇒ n s2=∑ ( x −x )2 ⇒(n−1)S 2=∑ ( x−x )2

⇒ n s2=(n−1)S 2
2 2
s S
⇒ =
n−1 n

s S
=
√ n−1 √ n

b) Control limits or Interval estimation

The General form of confidence limits or Interval estimation for μ,

T ± Z α SE (T ) , for LS

T ± t α SE ( T ) , for SS
 Where Z α ∧t α ( Statistical table value )

Now,

If x 1 , x 2 ,… … . x n be a random sample (iid) of size n is drawn from a

population with mean μ and variance σ 2. Then, the sampling distribution
of sample mean is students t- distribution

The General form of confidence limits or Interval estimation for μ,

T ± t α SE ( T ) , for SS

Where Z α ∧t α ( Statistical table value )

95% Control limits:

[ ¿ , UL ] = [ T −t α SE ( T ) , T +t α SE ( T ) ] , for SS

=[ T −2.262 SE ( T ) ,T +2.262 SE ( T ) ] , for SS

[ x−2.262 SE ( T ) , x +2.262 SE (T ) ] , for SS

[ x−2.262
s
√ n−1
, x+2.262
s
√ n−1
, for SS
]
[ 67− ( 2.262 )
2.96
√9
,67+(2.262)
2.96
√9
,
] for LS

=[64.76, 69.23]

99% Control limits:

[ ¿ , UL ] = [ T −t α SE ( T ) , T +t α SE ( T ) ] , for SS
 =[ T −3.250 SE ( T ) ,T +3.250 SE (T ) ] , for SS

¿ [ x−3.25 SE (T ) , x+3.25 SE ( T ) ] , for SS

[
= x−3.25
s
√ n−1
, x+ 3.25
s
√ n−1 ]
, for SS

[
¿ 67−(3.25)
2.96
√9
, 67+(3.25)
2.96
√9 ]
, for SS

=[63.79, 70.20]

Maximum Error

The Maximum error of the Estimator T is

E=Z α SE ( T ), for LS
E=t α SE ( T ) , for SS

Now,
E=t α SE ( T ), for SS
s
¿ 2.262 at α =5 %
√ n−1
2.96
¿ ( 2.262 ) =¿ 2.23
√9
And Similarly, Maximum Error
E=T α SE ( T ), for SS
s
¿ 3.25 at α =1 %
√n−1
2.96
¿ ( 3.25 ) =¿3.2
√9
Problem 3
Can it be conclude that, the average lifespan of Indian is more than 70
years if a replace of 100 Indians has an average lifespan of 71.8 years
with S.D 8.9 years, at 5% LOS

Sol:

The average lifespan of an Indian is 70 years i.e., H 0: μ=70

Against H 1 : μ>70 (The average lifespan of an Indian is more than 70 years

Under H 0,
x−μ
Z= N ( 0 ,1)
∂∨√ 10

71.8−70
¿ =2.02
8.9
√ 400
Z cal=4

Were Z cal> 1.645 , we reject H 0 at 5 % LOS .

Hence, The average lifespan of an Indian is more than 70 years.

Problem

A sample of 400 items is taken from a population whose standard

deviation is 10. The mean of the sample is 40. Test whether the sample
has come from a population with mean 38.

The problem is under test 1 because n ¿25 large sample and single
population variables.

Step 1 : H 0 : there is no significant difference between means μ=μ 0

Step 2 : H 1 : μ ≠ μ 0

Step 3 : The Level of significance,

 LOS=5 % ,⇒ Z tab =1.96

Step 4 : Sample data :

Sample size , n=400

Sample mean, x=40

μ=38 ,∂=10

Step 5 : Under H 0, the list statistic

x−μ
Z= (0 , 1)
∂∨√ 10

40−38
¿
10
√ 400
Z cal=4

Step 6 : Conclusion

Were Z cal> Z tab , then we reject H 0 at 5 % LOS .

Problem

An industrial safety program was recently instituted in the computer

chip industry. The average weekly loss (averaged over 1 month) in man
– hours due to accidents in 10 similar plants both before and after the
program are as follows:

Plant: 1 2 3 4 5 6 7 8 9 10

Before: 30 18 24 32 16 15 23 25 28 18

After: 23 21 22 28 14 15 24 21 23 16

Determine at 5% level of significance, whether the safety program has

been proven to be effective.
problem

The life time of electric bulbs for a random sample of 10 from a large
consignment gave the following data.

Item: 1 2 3 4 5 6 7 8 9 10
Life in ‘000 3. 3.
1.2 4.6 4.1 5.2 3.8 4.3 4.4 5.6
hrs: 9 9
Can we accept the hypothesis that the average life time of bulbs is
4000 hrs.

POP

Sol: H0 μ=4000

H1 μ ≠ 4000

⏟
X , X , … … … … X =n
1 2 10
1.2 , 4.6 , …… …… ….. 5.6

Sample mean, x

Sample SD, s

Step 1 : H 0 :The average life time of bulb is 4000hrs

H 0 : μ=4(μ 0)

Step 2 : H 1 : the average life of bulbs is not 4000hrs

H 0 : μ≠ 4(T −T )
Step 3 : The line of significance is 95% α =5 %
( tab ) ( tab ) ( tab )
t =t =t (9 , 0.025)=¿
( n−1 ,
α
2 ) ( 10−1,
0.05
2 )
Step 4 : From Sample data :

Item Life in ‘000hrs (X) X

2

1 1.2 1.44
2 4.6 21.1
3 3.9 15.21
4 4.1 16.81
5 5.2 27.04
6 3.8 14.44
7 3.9 15.21
8 4.3 18.49
9 4.4 19.36
10 5.6 31.36
∑ X=41 ∑ X 2=180.46

Sample X =
∑ x = 41 =4.1
n 10

√
Sample S.D S = √ s 2= 1 ∑ x 2−x 2=√ 18.046−4.12=1.11
n

Step 5 :
X −μ0 4.1−4
t= = =0.27
( )
S 1.1
√ n−1 √ 10−1

Step 6 :

|t cal|<|ttab| , we don ' t reject H 0

The average life time of bulb is 4000hrs.
Problem:

A company claims that, the mean thermal efficiency of diesel

engines produced by them is 32.3 %. To list this claim, a random sample
of 40 engines were examined which showed that the average thermal
efficiency of 31.4 % and sd of 1.6% at 5% level of significance.

Diesel comp.

Sol MTE

Points : H0 : μ=32.3 %

Sample size n=40(¿ 30 , LS)

Single pop. H 0 : μ≠ 32.3 % (T −T )

Quantitative measure ⏟
X , X ,… … … … X
1 2 40

Sample mean, x = 31.4%

Sample SD s = 1.6 %

Now , test procedure

Step 1 : H 0 : The mean thermal efficiency of diesel engines is 32.3%

i.e., H 0 : μ=32.3 (μ 0)

Step 2 : H 1 : The mean thermal efficiency of diesel engines is ≠ 32.3% (T –

T)

Step 3 : α : Level of significance,

(tab) (tab)
α =5 % , Z 0.05 =Z α =1.96

Step 4 : Sample data :

Sample size , n=40(¿ 30 , LS)

 Sample mean, x=31.4

Sample SD, s=1.6

Step 5 : Under H 0, the list statistic

x−μ0
Z= N (0 ,1)
s∨√ n

31.4−32.3
⇒ Z cal=
1.6∨ √ 40

0.9
¿−
0.253

¿−3.55

⇒|Z cal|=3.55

Step 6 : Decision :

Since |Z (3.55|>Z (1.96

cal ) tab )
, we reject H 0 at 5 % LOS .

Hence, the company doesn’t produced diesel engines with thermal

efficiency of 32.3 %

Problem :

Two file servers are compared according to their response time for
retrieving a small file. The mean response time of 50 such requests
submitted to server I was measured to be 682 ms with a SD of 25ms. A
similar measurement on server II resulted a mean of 675 was with an SD
of 28ms. Test the hypothesis that server I provide better response than
server II ?

SOL:

Server -I = Server-II
H0 =
μ1 μ2
X 1 , X 2 , … … … … X 50=n 1
Y 1 ,Y 2 , … … … … Y 50=n 2

x=682 , s21=25 y=675 , s22=28

Step 1:
H 0 : there is no significance difference between mean response times of
servers I and II

i.e., H 0 : μ1=μ2

Step 2 : H 1 : The server I provide better response time than server II

H 1 : μ 1< μ 2 (L−T )

Step 3 : The LOS

(tab)
α =5 % (Consider), Z 0.05 =−1.645

Step 4 : From Sample data :

Sample size , n1=50(LS) n2 =50(LS)

Sample mean, x=682 y=675

Sample SD’s s1=25 s2=28

Step 5 : Under H 0, the list statistic,

x− y
Z= N (0 ,1)

√ ∂21 ∂22
+
n1 n2
 682−675
⇒ Z cal= =1.32

√ 252 28 2
+
50 50

Step 6 :

Since |Z 1.32|> Z −1.645 , Don ; t reject H 0 at 5 % LOS .

( tab )

Hence, we conclude that there is no significance difference

between two servers with respect mean response time.

Problem - 4

The mean yield of wheat from a district A was 210 pounds with S.D 10
pounds per acre from a sample of 100 plots. In another district B the
mean yield was 220 pounds with S.D 12 pounds from a sample of 150
plots. Test whether there is any significant difference between the mean
yields of crops.

district A district B
H0 = = μ2
μ1
X 1 , X 2 , … … … … X 100 =n1 Y 1 ,Y 2 , … … … … Y 150 =n2

x=210 and s1=10 y=220 and s2=12

Step 1: H 0 :The mean yield of crop in two districts is same (OR)

There is no significant difference between the mean yields of crops

in the two districts.
H 0 : μ1=μ2 (OR) H 0 : μ1−μ2=0
 Step 2: H 1 : The mean yield of crops in two districts is different H 1 : μ 1 ≠ μ2

Step 3 : Let us consider, Level of significance is α =¿ 5%

(tab) (tab)
At α =5 % , Z α =Z 0.05 =1.96

Step 4: From Sample data :

Sample sizes, n1=100 n2 =150

Sample means x=210 y=220

Sample sd’s s1=10 s2=12

Step 5: Under H 0, the test statistic

x− y
Z= N (0 ,1)

√ s 21 s 22
+
n1 n2

210−220
Z cal= =−7.14

√
2 2
10 12
+
100 150

|Z cal|=7.14

Step 6 : Conclusion

Since, |Z (7.14
cal )
|> Z (1.96
tab )
, we reject H 0 at 5 % LOS .

Hence, there is some significance difference between mean yield two

districts.
5 problem

A reading test is given to an elementary school that consists of Indo –

American children and Anglo – American children. The samples are
given below.

Indo - American 9 11 13 11 15 9 12 14
Anglo - 10 12 10 14 9 8 10
American
Test the significance difference between two groups in averages of
reading test at 5% LOS.

Sol :

I-A = A-A
H0 =
μ1 μ2

H 1 μ1 ≠ μ2
X 1 , X 2 , … … … … X 8=n1 Y 1 ,Y 2 , … … … … Y 7 =n2

9 , 11 , ……. 14 10 , 12 , 12 , ……..10

Compete x=? y=?

2 2
s1=? s2=?

Step-1 H 0 :There is no significance difference between I-A and A-A

with respect means of reading test

H 0 : μ1=μ2
Step-2 H 1 There is some significance difference between I-A and A-A

with respect means

H 1 : μ 1 ≠ μ2 (T −T )

(tab) (tab)
Step-3 LOS , α =5 % , t n +n −2 ,α /2=,t 8+ 7−2, 0.025=2.160.
1 2

Step-4 from the sample data,

we need to calculate the following sample means and Sample variances (

sample estimators)

Sample sizes n1=8 n2 =7

Indo – Anglo – X
2
Y
2

American(x) American(y)
9 10 81 100
11 12 121 144
13 10 169 100
11 14 121 196
15 9 225 81
9 8 81 64
12 10 144 100
14 196
∑ x =94 ∑y =73 ∑ x 2=1138 ∑ x 2=¿ 785

Mean of first sample, x=

∑ x =11.75
n1

Mean of second sample, y=

∑ y =10.43
n2
 1
Sample variance of 1 st sample s1= n
2
∑ x 2−x 2
1

1 2
¿ (1138)−11.75 =4.187
8

and
2 1 2
Sample variance of 2 nd sample s2= 7 ( 785 )−10.43 =3.36

2 2
2 n1 s 1 +n2 s 2 8 (4.187)+7(3.36)
St = = =4.38
n1 +n2 −2 8+7−2

Step – 5 Under H 0 , the test statistic

( x− y )
t= t n +n −2

√ 1 1
1 2
2
S( + )
t
n1 n2

11.75−10.43
⇒ t cal=

√ 4.38 ( 18 + 17 )
⇒ t cal=1.22

|t cal|=1.22
Step-6 Decision,

Since |t (1.22|<t (2.16

cal ) tab )
, Don t ' t reject H 0 at 5 % LOS .

Hence , there is no significance difference between in two groups of

children of I-A & A-A with respect reading test