Wilcoxon Signed Rank test (one-sample)

Introduction: The sign test is based on the signs ignoring their magnitude. But
Wilcoxon Signed Rank test takes sign as well as its magnitude into
consideration.This test is more powerful and sensitive than ordinary sign test.
Procedure:
Let x1,x2,…..xn be a random sample drawn from a population of size n and
let M be the median of the given distribution (or) for the given
observations

H0: M = M0 (The median divides the distribution equally)

H1: M  M0 (or) M > M0 (or) M< M0

[Any one of these 3 conditions depends on the given problem]
Level of significance:
Appropriate level of significance is % (given/chosen)
Test Statistic:
Step1: Calculate di=xi-M0, i=1,2,…n
Step2: Arrange calculated di’s in ascending order irrespective of the sign.
If any value of di is zero, leave or omit that observation and then decide n.
Step3: Calculate the absolute value of dii.e |di|
Step4: Rank the values of |di| after discarding zero differences
Step5: Compute T+=Sum of the ranks of the +ve differences
Compute T-=Sum of the ranks of the -ve differences
Step6: Find, T=minimum(T+,T-)
Inference:

1) If it is a two tailed test (): Accept H0, if T>Tα/2,Otherwise reject H0.

2) If it is a right tailed test (>): Accept H0, if T>Tα,Otherwisereject H0.
3) If it is a left tailed test (<):Accept H0, if T<-Tα,Otherwisereject H0.

1
 Problems on Wilcoxon Signed Rank test
1) If a random sample taken at public play ground, the time required to play a
tennis match by players are: 36, 29, 44, 28, 40, 50, 39, 47, 33mins.
Use Wilcoxon signed rank test and test at 5% level of significance whether the
average time required to play a match is 35 mins at this public play ground.

H0: µ = 35
H1: µ35
Level of significance:
Appropriate level of significance is α=5% (given)
Test Statistic:
xi di=xi-35 Ascending order |di| Ranks T+ T-
of di irrespective given to
of sign |di|
36 1 1 1 1 1
29 -6 -2 2 2 2
44 9 4 4 3 3
28 -7 5 5 4 4
40 5 -6 6 5 5
50 15 -7 7 6 6
39 4 9 9 7 7
47 12 12 12 8 8
33 -2 15 15 9 9
32 13

Therefore n=9, T=minimum(T+,T-)= minimum(32,13) = 13

Inference:

Being a two tailed test () ,the table value of T from Wilcoxon signed rank
test at 5% level of significance for sample number n=9 is Tα/2= 5.
Here, T > Tα/2, i.e 13>5, We accept H0.

Hence we conclude that the average time required to play a match is 35mins.

2
 2) The following data gives a random sample of 15 measurements of Octane ratings
of certain kind of Gasoline
97.5, 95.2, 97.3, 96, 96.8, 100.3, 97.4, 95.3, 93.2, 99.1, 96.1, 97.6, 98.2, 98.5,
94.9.
Test whether the median is 98.5 or not using Wilcoxon signed rank test.

H0: M0 = 98.5
H1: M098.5
Level of significance:
Appropriate level of significance is α=5% (chosen)
Test Statistic:
xi di=xi-98.5 Ascending order |di| Ranks T+ T-
of di irrespective given to
of sign |di|
97.5 -1 -0.3 0.3 1 1
95.2 -3.3 0.6 0.6 2 2
97.3 -1.2 -0.9 0.9 3 3
96 -2.5 -1 1 4 4
96.8 -1.7 -1.1 1.1 5 5
100. 1.8 -1.2 1.2 6 6
3
97.4 -1.1 -1.7 1.7 7 7
95.3 -3.2 1.8 1.8 8 8
93.2 -5.3 -2.4 2.4 9 9
99.1 0.6 -2.5 2.5 10 10
96.1 -2.4 -3.2 3.2 11 11
97.6 -0.9 -3.3 3.3 12 12
98.2 -0.3 -3.6 3.6 13 13
98.5 0 -5.3 5.3 14 14
94.9 -3.6
10 95

Therefore n=14 (after omitting one observation because one di=0),

T=minimum(T+,T-)= minimum(10,95) = 10

Inference:

Being a two tailed test () ,the table value of T from Wilcoxon signed rank
test at 5% level of significance for sample number n=14 is Tα/2 = 21.
Here, T < Tα/2, i.e 10<21, We reject H0.

Hence we conclude that the median is not equal to 98.5.

3
 3) A random sample of size 15 infants of one month shows the following pulse
rates (beats/min):119, 120, 125, 122, 118, 117, 126, 114, 115, 123, 121, 120, 124,
127, 126
Is there an evidence to suggest that the median pulse rate of one month infants is
different from 120 beats/min.

H0: M0 = 120
H1: M0120
Level of significance:
Appropriate level of significance is α=5% (chosen)
Test Statistic:
xi di=xi-120 Ascending order |di| Ranks T+ T-
of di irrespective given to
of sign |di|
119 -1 -1 1 1.5 1.5
120 0 1 1 1.5 1.5
125 5 -2 2 3.5 3.5
122 2 2 2 3.5 3.5
118 -2 -3 3 5.5 5.5
117 -3 3 3 5.5 5.5
126 6 4 4 7 7
114 -6 -5 5 8.5 8.5
115 -5 5 5 8.5 8.5
123 3 -6 6 11 11
121 1 6 6 11 11
120 0 6 6 11 11
124 4 7 7 13 13
127 7
126 6
61 30

Therefore n=13 (after omitting two observations because two di’s are 0),
T=minimum(T+,T-)= minimum(61,30) = 30

Inference:

Being a two tailed test () ,the table value of T from Wilcoxon signed rank
test at 5% level of significance for sample number n=13 is Tα/2 = 17.
Here, T >Tα/2, i.e30>17, We acceptH0.

Hence we conclude that the median pulse rate of one month infants is different

4
 from 120 beats/min.

Problems on Wilcoxon Signed Rank test (two samples/paired sample test)

4) In order to determine if smoking results increased heart activity, a random sample of
size 20 smokers was taken. Their pulse rate of size 20 was taken. Their pulse
rate(beats/min) before smoking and after smoking a certain brand of cigarettes are
being measured and the result obtained are :
Before 70 69 72 74 66 68 69 70 71 69 73 72 68 72 67 70 68 69 70 71
After 69 72 71 74 68 67 72 72 70 75 75 73 71 72 69 71 72 70 71 71
Test whether thereis a change in heart activity among smokers
H0: µx =µy
H1: µx≠ µy
Level of significance:
Appropriate level of significance is α=5% (chosen)
Test Statistic:
xi yi di=xi-yi Ascending order |di| Ranks given T+ T-
of di irrespective to
of sign |di|
70 69 1 -1 1 4.5 4.5
69 72 -3 -1 1 4.5 4.5
72 71 1 -1 1 4.5 4.5
74 74 0 -1 1 4.5 4.5
66 68 -2 1 1 4.5 4.5
68 67 1 1 1 4.5 4.5
69 72 -3 1 1 4.5 4.5
70 72 -2 1 1 4.5 4.5
71 70 1 -2 2 10.5 10.5
69 75 -6 -2 2 10.5 10.5
73 75 -2 -2 2 10.5 10.5
72 73 -1 -2 2 10.5 10.5
68 71 -3 -3 3 14 14
72 72 0 -3 3 14 14
67 69 -2 -3 3 14 14
70 71 -1 -4 4 16 16
68 72 -4 -6 6 17 17
69 70 -1
70 71 -1
71 71 0
18 135
Therefore n=17 (after omitting three observations because three di’s are 0),
T=minimum(T+,T-) = minimum(18,135) = 18
Inference:
Being a two tailed test (≠) , the table value of T from Wilcoxon signed rank test at 5%

5
 level of significance for sample number n=17 is T =35
Here, T <Tα/2, i.e 18<35, We reject H0.
Hence we conclude that there will be a change in the heart activity among smokers.

5) In a random sample of 10 days in two cities of Andhra Pradesh, the following

temperatures are recorded:
City A 108 104 109 116 111 113 114 112 102 112
City B 105 109 110 109 106 112 108 107 104 108
Test whether the average of temperatures of two cities are same.

H0: µx =µy
H1: µx≠ µy
Level of significance:
Appropriate level of significance is α=5% (chosen)
Test Statistic:
xi yi di=xi- yi Ascending order |di| Ranks T+ T-
of di irrespective given to
of sign |di|
108 10 -3 -1 1 1.5 1.5
5
104 10 -5 1 1 1.5 1.5
9
109 11 -1 -2 2 3 3
0
116 10 7 -3 3 4 4
9
111 10 5 4 4 5 5
6
113 11 1 -5 5 7 7
2
114 10 6 5 5 7 7
8
112 10 5 5 5 7 7
7
102 10 -2 6 6 9 9
4
112 10 4 7 7 10 10
8
39.5 15.5
+ -
Therefore n=10, T=minimum(T ,T ) = minimum(39.5,15.5) = 15.5
Inference:
Being a two tailed test (≠) , the table value of T from Wilcoxon signed rank test at 5%
level of significance for sample number n=10 is T =8
Here, T >Tα/2, i.e 15.5> 8, We accept H0.
Hence we conclude that there is no significant different between the average
temperatures in both the cities.

6
 Mann-Whitney Test (or) U-test

Procedure:
H0: µx = µy

H1: µxµy (or) µx> µy (or) µx<µy

[Any one of these 3 conditions depends on the given problem]
Level of significance:
Appropriate level of significance is % (given/chosen)
Test Statistic:
Step1: Combine the samples and arrange all the values in ascending order
Step2: Assign ranks to all the arranged values.
Step3: i) Find the sum of the ranks for each of the original values in the samples
ii) Let us denote the sum of the ranks of the first sample as R1
iii) Let us denote the sum of the ranks of the second sample as R2
vi) Let us denote the number of observations in the first sample as n1
v) Let us denote the number of observations in the second sample as n2
( n¿¿ 1+ 1)
Step4: Calculate U 1=n1 n 2+ n1 −R1 ¿
2
( n¿¿ 2+1)
Calculate U 2=n1 n 2+ n2 −R2 ¿
2
U=minimum{U1,U2}

Inference:

7
 1) If it is a two tailed test ( ): Accept H0, if U>Uα/2,Otherwise reject
H 0.
2) If it is a right tailed test (>): Accept H0, if U>Uα,Otherwisereject H0.
3) If it is a left tailed test (<):Accept H0, if U<-Uα, Otherwisereject H0.

Problems on Mann-Whitney test

1) Nicotine content of two brands of cigarettes was found as follows:
BrandA 2.1 4.0 6.3 5.4 4.8 3.7 6.1 3.3
BrandB 4.1 0.6 3.1 2.5 4.0 6.2 1.6 2.2 1.9 5.4
Test whether the distribution of nicotine in two brands is equal using U-test at
5% level of significance.
H0: µx = µy
H1: µx µy
Level of significance:
Appropriate level of significance is α=5% (given)
Test Statistic:
xi yi Combined Ascending order of Rank Ranks of Ranks of
sample the values in the s the first the second
combined sample sample sample
2.1 4. 2.1 0.6 1 4 12
1
4.0 0. 4.0 1.6 2 10.5 1
6
6.3 3. 6.3 1..9 3 18 7
1
5.4 2. 5.4 2.1 4 14.5 6
5
4.8 4. 4.8 2.2 5 13 10.5
0
3.7 6. 3.7 2.5 6 9 17

8
 2
6.1 1. 6.1 3.1 7 16 2
6
3.3 2. 3.3 3.3 8 8 5
2
1. 4.1 3.7 9 3
9
5. 0.6 4.0 10.5 14.5
4
3.1 4.0 10.5
2.5 4.1 12
4.0 4.8 13
6.2 5.4 14.5
1.6 5.4 14.5
2.2 6.1 16
1.9 6.2 17
5.4 6.3 18
93 78
Here, n1=8, n2=10, R1=93, R2=78
( n¿¿ 1+ 1) 8 ( 8+1 )
Calculate U 1=n1 n 2+ n1 −R1=8× 10+ −93=23 ¿
2 2
( n¿¿ 2+1) 10 ( 10+1 )
Calculate U 2=n1 n 2+ n2 −R2 =8 ×10+ −78=57 ¿
2 2
U=minimum{U1,U2}=minimum{23,57}=23
Inference:
Being a two tailed test ( ) ,the table value of U from Mann-Whitney test
tables for n1=8, n2=10 at 5% level of significance is 17 i.eUα/2 = 17.
Here, U>Uα/2, i.e23>17, We accept H0.
Hence we conclude that the average nicotine in both the brands are same.
2) The following data represents the weight in Kgs of a personal luggage carried in
an aircraft by the members of two baseball clubs
Club A 34 39 41 28 33
Club B 36 40 35 31 39 36
Use α=0.05 to test the hypothesis that the two clubs carry same amount of the
luggage.
H0: µx = µy
H1: µx µy
Level of significance:
Appropriate level of significance is α=0.05 (given)
Test Statistic:
xi yi Combined Ascending order of Rank Ranks of Ranks of
sample the values in the s the first the second
combined sample sample sample
34 36 34 28 1 4 6.5

9
 39 40 39 31 2 8.5 10
41 35 41 33 3 11 5
28 31 28 34 4 1 2
33 39 33 35 5 3 8.5
36 36 36 6.5 6.5
40 36 6.5
35 39 8.5
31 39 8.5
39 40 10
36 41 11
27.5 38.5
Here, n1=5, n2=6, R1=27.5, R2=38.5
( n¿¿ 1+ 1) 5 (5+ 1 )
Calculate U 1=n1 n 2+ n1 −R1=5 × 6+ −27.5=17.5 ¿
2 2
( n¿¿ 2+1) 6 ( 6+1 )
Calculate U 2=n1 n 2+ n2 −R2 =5× 6+ −38.5=12.5 ¿
2 2
U=minimum{U1,U2}=minimum{17.5,12.5}=12.5
Inference:
Being a two tailed test ( ) ,the table value of U from Mann-Whitney test
tables for n1=5, n2=6 at 5% level of significance is 3i.e Uα/2 = 3.
Here, U>Uα/2, i.e12.5>3, We accept H0.
Hence we conclude that the average luggage carried by the members of the
two baseball clubs are same.

3) Suppose that two drugs A and B are being compared. Suppose 3 people took
drug A and 4 people took drug B, the number of minutes that took to relieve
from the pain are given below:
Drug A 9 11 15
Drug B 6 8 10 13
Are the mean number of minutes that took to get relieved from the pain are same
for both the drugs? Test at 1% level of significance.
H0: µx = µy
H1: µx µy
Level of significance:
Appropriate level of significance is α=1% (given)
Test Statistic:

10
 xi yi Combined Ascending order of Rank Ranks of Ranks of
sample the values in the s the first the second
combined sample sample sample
9 6 9 6 1 3 1
11 8 11 8 2 5 2
15 10 15 9 3 7 4
13 6 10 4 6
8 11 5
10 13 6
13 15 7
15 13
Here, n1=3, n2=4, R1=15, R2=13
( n¿¿ 1+ 1) 3 ( 3+1 )
Calculate U 1=n1 n 2+ n1 −R1=3 × 4+ −15=3 ¿
2 2
( n¿¿ 2+1) 4 ( 4+1 )
Calculate U 2=n1 n 2+ n2 −R2 =3× 4 + −13=9 ¿
2 2
U=minimum{U1,U2}=minimum{3,9}=3
Inference:
Being a two tailed test ( ) ,the table value of U from Mann-Whitney test
tables for n1=3, n2=4 at 1% level of significance is 1i.e Uα/2 = 1.
Here, U>Uα/2, i.e3>1, We accept H0.
Hence we conclude that the mean number of minutes that took to get relieved
from the pain are same for both the drugs.

4) Students in X-class were divided into two groups according to the methods of
instruction. The following data gives the scores obtained by the two groups of students
in the final examination. Test whether the new method is more advantageous at 5%
level of significance
New method 431 225 372 681 576 628 605 689 476 359 269 654
Old method 655 603 466 860 382 580 750 855 776 640 532 630 670
Solution :H0: µx = µy H1: µx> µyLevel of significance: α=5% (given)
xi yi Combined Ascending order of Ranks Ranks of Ranks of
sample the values in the the first the second
combined sample sample sample
431 655 431 225 1 6 18
225 603 225 269 2 1 12
372 466 372 359 3 4 7
681 860 681 372 4 20 25

11
 576 382 576 382 5 10 5
628 580 628 431 6 14 11
605 750 605 466 7 13 22
689 855 689 476 8 21 24
476 776 476 532 9 8 23
359 640 359 576 10 3 16
269 532 269 580 11 2 9
654 630 654 603 12 17 15
670 655 605 13 19
603 628 14
466 630 15
860 640 16
382 654 17
580 655 18
750 670 19
855 681 20
776 689 21
640 750 22
532 776 23
630 855 24
670 860 25
119 206
Here, n1=12, n2=13, R1=119, R2=206
( n¿¿ 1+ 1) 12 ( 12+ 1 )
Calculate U 1=n1 n 2+ n1 −R1=12 ×13+ −119=115 ¿
2 2
( n¿¿ 2+1) 13 ( 13+1 )
Calculate U 2=n1 n 2+ n2 −R2 =12× 13+ −206=41¿
2 2
U=minimum{U1,U2}=minimum{115,41}=41

Inference:
Being a two tailed test ( ) , the table value of U from Mann-Whitney test tables for
n1=12, n2=13 at 1% level of significance is 1 i.e Uα =47.
Here, U < Uα, i.e41<47, We reject H0.Hence we conclude that the new method is
more advantageous than the old method of instruction.

12
 Kolmogorov-Smirnov test (one-sample)

Procedure:
H0: The data follows given distribution

H1: The data do not follow given distribution

Level of significance:
Appropriate level of significance is % (given/chosen)
Test Statistic:
Step1: Count the number of given observations n (say)
Step2: Arrange the given observations in ascending order without any repetitions say x(i)
Step3: Generate the following table

i x(i) F(x(i)) i
n
i−1
n ( ni )−F( x )
i ( i−1n )
F ( xi )−

13
Step4:
D
+¿=Max
{( )
i
n }
−F ( xi ) ¿

D
{
−¿=Max F ( x i )− (i−1n )}¿
Now D= Max{D+,D-}

Inference:
If calculated D < The table value of D at α% level of significance,
we accept H0. Otherwise reject H0.

1. The sequence of numbers 0.63, 0.49, 0.24, 0.57, 0.76, 0.89 have been
generated. Use the Kolmogorov-Smirnov test to determine if the
hypothesis that the numbers are uniformly distributed in the interval [0,1].

H0: The numbers are uniformly distributed

H1: The numbers are not uniformly distributed

Level of significance:
Appropriate level of significance is 5% (chosen)
Test Statistic:
Step1: Count the number of given observations n=6
Step2: Arrange the given observations in ascending order without any repetitions say x(i)
0.24, 0.49, 0.57, 0.63, 0.76, 0.89
Step3: Generate the following table

14
 i x(i) F(x(i))
n
i i−1
n ( ni )−F( x )
i ( i−1n )
F ( xi )−

1 0.24 0.24 1/6=0.1666 0/6=0.0000 -0.0734 0.2400

2 0.49 0.49 2/6=0.3333 1/6=0.1666 -0.1567 0.3234
3 0.57 0.57 3/6=0.5000 2/6=0.3333 -0.0700 0.2367
4 0.63 0.63 4/6=0.6666 3/6=0.5000 0.0366 0.1300
5 0.76 0.76 5/6=0.8333 4/6=0.6666 0.0733 0.0934
6 0.89 0.89 6/6=1.0000 5/6=0.8333 0.1100 0.0567

Step4:
D
+¿=Max
{( ) ( )} =0.1100
i
n
−F xi ¿

D
{
−¿=Max F ( x i )− ( )} =0.3234
i−1
n
¿

Now D= Max{D+,D-}=0.3234

Inference:
The table value of D for n=6 at 5% level of significance isD0.05=0.521.
Here, D<Dαwe accept H0

2. Ten points are taken in an interval of length one metre. The distance of
each point from the start of the interval is (in meters) as follows:
0.414, 0.523, 0.229, 0.942, 0.097, 0.394, 0.572, 0.486, 0.273, 0.358
The points may be supposed to be chosen at random and independently of
each other if and only if the 10 observations form a random sample from
the uniform distribution over the interval [0,1]. Examine whether this is
borne out by the data by using Kolmogorov-Smirnov test.

H0: The numbers are uniformly distributed

H1: The numbers are not uniformly distributed

Level of significance:
Appropriate level of significance is 5% (chosen)
Test Statistic:

15
Step1: Count the number of given observations n=10
Step2: Arrange the given observations in ascending order without any repetitions say x(i)
0.097, 0.229, 0.273, 0.358, 0.394, 0.414, 0.486, 0.523, 0.572, 0.942
Step3: Generate the following table
i x(i) F(x(i)) i
n
i−1
n ( ni )−F( x )i ( i−1n )
F ( xi )−

1 0.09 0.097 0.1 0 0.003 0.097

7
2 0.22 0.229 0.2 0.1 -0.029 0.129
9
3 0.27 0.273 0.3 0.2 0.027 0.073
3
4 0.35 0.358 0.4 0.3 0.042 0.058
8
5 0.39 0.394 0.5 0.4 0.106 -0.006
4
6 0.41 0.414 0.6 0.5 0.186 -0.086
4
7 0.48 0.486 0.7 0.6 0.214 -0.114
6
8 0.52 0.523 0.8 0.7 0.277 -0.177
3
9 0.57 0.572 0.9 0.8 0.328 -0.228
2
10 0.94 0.942 1.0 0.9 0.058 0.042
2
Step4:
D
+¿=Max
{( ) ( )} =0.328
i
n
−F xi ¿

D
{
−¿=Max F ( x i )− ( )} =0.129
i−1
n
¿

Now D= Max{D+,D-}=0.328
Inference:
The table value of D for n=10 at 5% level of significance isD0.05 =0.410.
Here, D<Dα we accept H0.

3. It is desired to check whether pinholes in electrolytic tin plate are uniformly

distributed across a plated coil on the basis of the following distances in
inches of 10 pin holes form one edge of a long strip of tin plate 30 inches
wide: 4.8, 14.8, 28.2, 23.1, 4.4, 28.7, 19.5, 2.4, 25.0, 6.2
16
 Test the null hypothesis at the 0.05 level of significance

H0: The pinholes are uniformly distributed

H1: The pinholes are not uniformly distributed
Level of significance:
Appropriate level of significance is 0.05 (given)
Test Statistic:
Step1: Count the number of given observations n=10
Step2: Arrange the given observations in ascending order without any repetitions say x(i)
2.4, 4.4, 4.8, 6.2, 14.8, 19.5, 23.1, 25.0, 28.2, 28.7
Step3: Generate the following table
i x(i) F(x(i))=
x(i)
i
n
i−1
n ( ni )−F( x )
i ( i−1n )
F ( xi )−

30
1 2.4 0.08 0.1 0 0.02 0.08
2 4.4 0.1466 0.2 0.1 0.0534 0.0466
3 4.8 0.16 0.3 0.2 0.14 -0.04
4 6.2 0.2066 0.4 0.3 0.1934 -0.0934
5 14.8 0.4933 0.5 0.4 0.0067 0.0933
6 19.5 0.65 0.6 0.5 -0.05 0.15
7 23.1 0.77 0.7 0.6 -0.07 0.17
8 25.0 0.8333 0.8 0.7 -0.0333 0.1333
9 28.2 0.94 0.9 0.8 -0.04 0.14
1 28.7 0.9566 1.0 0.9 0.0434 0.0566
0
Step4:
D
+¿=Max
{( ) ( )} =0.1934
i
n
−F xi ¿

D
{
−¿=Max F ( x i )− ( )} =0.17
i−1
n
¿

Now D= Max{D+,D-}=0.1934
Inference:
The table value of D for n=10 at 5% level of significance isD0.05 =0.410.
Here, D<Dα we accept H0 .

4. The following service times ( in mins) of a random sample of 15 customers

were recorded at a bank. 7, 3, 2, 7, 5, 4, 9, 15, 6, 6, 1, 6, 8, 10, 12.
Test the hypothesis that the service times are exponentially distributed
with mean time of 5 minutes.

17
 H0: The data follows exponential distribution with mean 5 minutes i.e θ=5
H1: The data do not follow exponential distribution with mean 5 minutes
θ≠5
Level of significance:
Appropriate level of significance is 5% (chosen)
Test Statistic:
Step1: Count the number of given observations n=15
Step2: Arrange the given observations in ascending order without any repetitions say x(i)
1,2,3,4,5,6,7,8,9,10,12,15
Step3: Generate the following table
i x(i) F(x(i))=1−e−θx i
n
i−1
n ( ni )−F( x )
i ( i−1n )
F ( xi )−

1 1 0.9936 0.0666 0 -0.9266 0.9936

2 2 0.9995 0.1333 0.066 -0.8662 0.9333
6
3 3 0.9999 0.2000 0.133 -0.7999 0.8666
3
4 4 0.9999 0.2666 0.200 -0.7333 0.7999
0
5 5 1 0.3333 0.266 -0.6667 0.7333
6
6 6 1 0.4000 0.333 -0.6000 0.6666
3
7 7 1 0.4666 0.400 -0.5334 0.6000
0
8 8 1 0.5333 0.466 -0.4667 0.5333
6
9 9 1 0.6000 0.533 -0.4000 0.4666
3
10 10 1 0.6666 0.600 -0.3334 0.4000
0
11 12 1 0.7333 0.666 -0.2666 0.3334
6
12 15 1 0.8000 0.733 -0.2000 0.2666
3
Step4:
D
+¿=Max
{( ) ( )} =-0.2000
i
n
−F xi ¿

D
{
−¿=Max F ( x i )− ( )} =0.9936
i−1
n
¿

Now D= Max{D+,D-}=0.9936
Inference:
18
The table value of D for n=15 at 5% level of significance isD0.05 =0.338
Here, D>Dα we reject H0 .

5. The following are 15 measurements of the boiling point of a silicon

compound ( in degrees celsius) :
166, 141, 136, 153, 170, 162, 155, 146, 183, 157, 148, 132, 160, 175, 150.
Use the Kolmogorov-Smirnov test at the 0.01 level of significance to test
the null hypothesis that the boiling points come from a normal population
with µ=160 degrees celsius and σ=10 degrees celsius.

H0: The data follows normal distribution with µ=160 and σ=10
H1: The data do not follow normal distribution with µ=160 and σ=10
Level of significance:
Appropriate level of significance is 0.01 (given)
Test Statistic:
Step1: Count the number of given observations n=15
Step2: Arrange the given observations in ascending order without any repetitions say x(i)
132, 136, 141, 146, 148, 150, 153, 155, 157, 160, 162, 166, 170, 175, 183
Step3: Generate the following table
i x(i)
F(z=¿
x−μ
σ
)
i
n
i−1
n ( ni )−F( x )
i ( i−1n )
F ( xi )−

1 132 F(132-160/10)=F(-2.8)=0.0026 0.0666 0.0000 0.0640 0.0026

2 136 F(136-160/10)=F(-2.4)=0.0082 0.1333 0.0666 0.1251 -0.0584
3 141 F(141-160/10)=F(-1.9)=0.0287 0.2000 0.1333 0.1713 -0.1046
4 146 F(146-160/10)=F(-1.4)=0.0808 0.2666 0.2000 0.1858 -0.1192
5 148 F(148-160/10)=F(-1.2)=0.1151 0.3333 0.2666 0.2182 -0.1515
6 150 F(150-160/10)=F(-1.0)=0.1587 0.4000 0.3333 0.2413 -0.1746
7 153 F(153-160/10)=F(-0.7)=0.2420 0.4666 0.4000 0.2246 -0.1580
8 155 F(155-160/10)=F(-0.5)=0.3085 0.5333 0.4666 0.2248 -0.1581
9 157 F(157-160/10)=F(-0.3)=0.3821 0.6000 0.5333 0.2179 -0.1512
10 160 F(160-160/10)=F(0)=0.5000 0.6666 0.6000 0.1666 -0.1000
11 162 F(162-160/10)=F(0.2)=0.5793 0.7333 0.6666 0.1540 -0.0873
12 166 F(166-160/10)=F(0.6)=0.7257 0.8000 0.7333 0.0743 -0.0076
13 170 F(170-160/10)=F(1.0)=0.8413 0.8666 0.8000 0.0253 0.0413
14 175 F(175-160/10)=F(1.5)=0.9332 0.9333 0.8666 0.0001 0.0666
15 183 F(183-160/10)=F(2.3)=0.9893 1.0000 0.9333 0.0107 0.0560
Step4:
D
+¿=Max
{( )
i
n } =-0.2413
−F ( xi ) ¿

19
D
{
−¿=Max F ( x i )− (i−1n )}¿=0.0666
Now D= Max{D+,D-}=0.2413
Inference:
The table value of D for n=15 at 5% level of significance is D 0.01 =0.404
Here, D<Dα we accept H0 .

Kolmogorov-Smirnov test (two-samples)

Procedure:
H0: f1(x)=f2(y) i.e the distributions are same

H1: f1(x)≠f2(y)
Level of significance:
Appropriate level of significance is % (given/chosen)
Test Statistic:
Step1: Combine the samples
Step2: Arrange all the values in ascending order without repetitions x(i) then count
the observations say n
Number of observations∈the first sample ≤ x (i)
Step3: Calculate Sm ( x )=
n

Number of observations∈the second sample ≤ x (i)

Step4: Calculate T n ( x )=
n

Step5: Generate the following table

Sample 1 Sample 2 Combined Ascending order of

samples the combined
sample without
repetitions
(ordered Statistics) Sm(x) Tn(x) Sm(x)- Tn(x) Tn(x)-Sm(x)
x(i)

{S ( x ) −T n (x)} ¿
Step6: D+¿=Max
m, n
m

−¿=Max {T n ( x ) −Sm (x) } ¿

Dm , n
−¿ }¿

D=max ⁡{D+¿ ,D
m ,n
m ,n ¿

20
Inference:
If calculated D < The table value of D at α% level of significance,
we accept H0. Otherwise reject H0.

1. The following data represent the life times of batteries of two different
brands in hours.
Brand A 40 30 40 45 55 30
Brand B 50 55 45 55 60 40
Test whether the two brands of the batteries are same by Kolmogorov -
Smirnov test at 5% level of significance.

H0: µx=µy
H1: µx≠ µy
Level of significance:
Appropriate level of significance is 5% (given)
Test Statistic:
Step1: Combine the samples
Step2: Arrange all the values in ascending order without repetitions x(i) then count
the observations, n=6
Number of observations∈the first sample ≤ x (i)
Step3: Calculate Sm ( x )=
n
Number of observations∈the second sample ≤ x (i)
Step4: Calculate T n ( x )=
n
Step5: Generate the following table
Brand A Brand B Combined Ascending order of
samples the combined
sample without
repetitions
(ordered Statistics) Sm(x) Tn(x) Sm(x)- Tn(x) Tn(x)-Sm(x)
x(i)
40 50 40 30 2/6 0/6 2/6 -2/6
30 55 30 40 4/6 1/6 3/6 -3/6
40 45 40 45 5/6 2/6 3/6 -3/6
45 55 45 50 5/6 3/6 2/6 -2/6
55 60 55 55 6/6 5/6 1/6 -1/6
30 40 30 60 6/6 6/6 0 0
50
55

21
 45
55
60
40
3
+¿=Max {S m ( x ) −T n (x)} = =0.5 ¿
Step6: D 6
m, n
−¿=Max {T n ( x ) −Sm (x) }=0 ¿
Dm , n
D=max ¿ ¿
Inference:
The table value of D for n=6 at 5% level of significance isD 0.05 =0.521.
Here, D<Dα we accept H0
2. As a part of a training program, some trainees are instructed by two different methods
and the scores obtained in the tests are :
Method A 71 75 65 69 73 66 68 71 74 68
Method B 72 77 84 78 69 70 77 73 65 75
Test whether the two different training methods are significant or not by Kolmogorov -
Smirnov test at1% level of significance.

H0: µx=µyH1: µx≠ µy

Level of significance: Appropriate level of significance is 1% (given)
Test Statistic:
Step1: Combine the samples
Step2: Arrange all the values in ascending order without repetitions x(i) then count
the observations, n=13
Number of observations∈the first sample ≤ x (i)
Step3: Calculate Sm ( x )=
n
Number of observations∈the second sample ≤ x (i)
Step4: Calculate T n ( x )=
n
Step5: Generate the following table
Brand A Brand B Combine Ascending order of
d samples the combined
sample without
repetitions
(ordered Statistics) Sm(x) Tn(x) Sm(x)- Tn(x) Tn(x)-Sm(x)
x(i)
71 72 71 65 1/13 1/13 0 0
75 77 75 66 2/13 1/13 1/13 -1/13
65 84 65 68 4/13 1/13 3/13 -3/13
69 78 69 69 5/13 2/13 3/13 -3/13
73 69 73 70 5/13 3/13 2/13 -2/13
66 70 66 71 7/13 3/13 4/13 -4/13
68 77 68 72 7/13 4/13 3/13 -3/13
71 73 71 73 8/13 5/13 3/13 -3/13
74 65 74 74 9/13 5/13 4/13 -4/13
68 75 68 75 10/13 6/13 4/13 -4/13

22
 72 77 10/13 8/13 2/13 -2/13
77 78 10/13 9/13 1/13 -1/13
84 84 10/13 10/13 0 0
78
69
70
77
73
65
75
4
+¿=Max {S m ( x ) −T n (x)} = =0.3077 ¿
Step6: D 13
m, n
−¿=Max {T n ( x ) −Sm (x) }=0 ¿
Dm , n
D=max ¿ ¿Inference: The table value of D for n=13 at 1% level of significance is D0.01 =0.433.
Here, D<Dα we accept H0
3. The following are the no. of sales which a sample of 9 sales people of industrial
chemicals in California and a sample of 6 sales people of industrial chemicals in Oregon
made over a certain fixed period of time
California 59 68 44 71 63 46 69 54 48
Oregon 50 36 62 52 70 41
Test whether the sales in the two states are distributed equally by using Kolmogorov -
Smirnov test at 1% level of significance.

H0: µx=µy
H1: µx≠ µy
Level of significance:Appropriate level of significance is 1% (given)
Test Statistic:
Step1: Combine the samples
Step2: Arrange all the values in ascending order without repetitions x(i) then count
the observations, n=15
Number of observations∈the first sample ≤ x (i)
Step3: Calculate Sm ( x )=
n
Number of observations∈the second sample ≤ x (i)
Step4: Calculate T n ( x )=
n
Step5: Generate the following table
California Orego Combined Ascending order of
n samples the combined
sample without
repetitions
(ordered Statistics) Sm(x) Tn(x) Sm(x)- Tn(x) Tn(x)-Sm(x)
x(i)
59 50 59 36 0/15 1/15 -1/15 1/15
68 36 68 41 0/15 2/15 -2/15 2/15
44 62 44 44 1/15 2/15 -1/15 1/15
71 52 71 46 2/15 2/15 0 0
63 70 63 48 3/15 2/15 1/15 -1/15
46 41 46 50 3/15 3/15 0 0
69 69 52 3/15 4/15 -1/15 1/15

23
 54 54 54 4/15 4/15 0 0
48 48 59 5/15 4/15 1/15 -1/15
50 62 5/15 5/15 0 0
36 63 6/15 5/15 1/15 -1/15
62 68 7/15 5/15 2/15 -2/15
52 69 8/15 5/15 3/15 -3/15
70 70 8/15 6/15 2/15 -2/15
41 71 9/15 6/15 3/15 -3/15
3
+¿=Max {S m ( x ) −T n (x)} = =0.2 ¿
Step6: D 15
m, n
2
−¿=Max {T n ( x ) −Sm (x) }= =0.1333 ¿
15
Dm , n
D=max ¿ ¿
Inference: The table value of D for n=15 at 1% level of significance is D0.01 =0.404.
Here, D<Dα we accept H0

4. To compare two kinds of bumper guards 6 of each kind were mounted on a certain kind
of compact car . Then each car was run into a concrete wall at 5 miles per hour and the
following are the cost of repairs in dollars.
Bumper guard 1 107 148 123 165 102 119
Bumper guard 2 134 115 112 151 133 129
Test whether the two bumper gaurds are equally efficient at 5% level of significance by
using Kolmogorov -Smirnov test at 5% level of significance.

H0: µx=µy
H1: µx≠ µy
Level of significance:Appropriate level of significance is 1% (given)
Test Statistic:
Step1: Combine the samples
Step2: Arrange all the values in ascending order without repetitions x(i) then count
the observations, n=125
Number of observations∈the first sample ≤ x (i)
Step3: Calculate Sm ( x )=
n
Number of observations∈the secon d sample ≤ x (i)
Step4: Calculate T n ( x )=
n
Step5: Generate the following table
Bumper Bumper Combined Ascending order of
Gaurd1 Gaurd2 samples the combined
sample without
repetitions
(ordered Statistics) Sm(x) Tn(x) Sm(x)- Tn(x) Tn(x)-Sm(x)
x(i)
107 134 107 102 1/12 0/12 1/12 -1/12
148 115 148 107 2/12 0/12 2/12 -2/12
123 112 123 112 2/12 1/12 1/12 -1/12
165 151 165 115 2/12 2/12 0 0
102 133 102 119 3/12 2/12 1/12 -1/12

24
 119 129 119 123 4/12 2/12 2/12 -2/12
134 129 4/12 3/12 1/12 -1/12
115 133 4/12 4/12 0 0
112 134 4/12 5/12 -1/12 1/12
151 148 5/12 5/12 0 0
133 151 5/12 6/12 -1/12 1/12
129 165 6/12 6/12 0 0
2
+¿=Max {S m ( x ) −T n (x)} = =0.1666¿
Step6: D 12
m, n
1
−¿=Max {T n ( x ) −Sm (x) }= =0.0833 ¿
12
Dm , n
D=max ¿ ¿Inference: The table value of D for n=12 at 5% level of significance is D0.05 =0.375.
Here, D<Dα we accept H0

25