Scribd
Upload
815 views10 pages

Sign Test

The sign test is a nonparametric test used to test whether the median of a distribution equals a fixed value. It can be used as an alternative to one-sample or paired t-tests. The sign test works by ranking observations as greater than or less than the hypothesized median and calculating the probability of obtaining the observed number of rankings or more extreme under the null hypothesis that the probability of being above or below the median is 0.5. The Kruskal-Wallis test is a nonparametric test used to determine if three or more independent samples were selected from populations with the same distribution. It works by ranking all observations together and calculating a test statistic based on the sum of ranks for each sample group.

Uploaded by

ShubhamMaheshwari
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
815 views10 pages

Sign Test

The sign test is a nonparametric test used to test whether the median of a distribution equals a fixed value. It can be used as an alternative to one-sample or paired t-tests. The sign test works by ranking observations as greater than or less than the hypothesized median and calculating the probability of obtaining the observed number of rankings or more extreme under the null hypothesis that the probability of being above or below the median is 0.5. The Kruskal-Wallis test is a nonparametric test used to determine if three or more independent samples were selected from populations with the same distribution. It works by ranking all observations together and calculating a test statistic based on the sum of ranks for each sample group.

Uploaded by

ShubhamMaheshwari
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
  • Sign Test
  • The Kruskal-Wallis Test

Sign test

This non parametric test called thesign testis used to test the
null hypothesis that is
assumed and says that the median of the distribution is equal
to a particular fixed value.
The sign test can be used in the place of parametric tests, like
one sample t-test, or in
the place of the paired sample t test.
The sign test can be used to test that kind of data which is of
ordered categorical type.
Thesign test is usually used in the kind of data where it is
possible to rank the observations.
The researcher should keep in mind that the Wilcoxon Signed
Rank Sum test is used in such
situations, and is also more powerful than thesign test.

Procedure
a sample consisting of n number of observations is
drawn.
it is assumed that some s+ of the observations are
greater than the population median
it is also assumed that some s- observations are
smaller than the population median.
the observations in the sample, which are absolutely
equal to the population median, are ignored by the
researcher.
The value of the sum of s+ and s- are therefore less
than the sample size n in the sign test and therefore
considered as n/.

Contd
The first step of conducting asign testinvolves
choosing the value of s, which is nothing but the
maximum value among the s+ and s-
Then, the usage of the tables of the binomial
distribution is done in thesign test. This table of
binomial distribution is done in the sign test in
order to find the probability of the value of s or
something higher than s under the assumption
that the probability is 0.5 and the n= n/ .
If the sign test is two sided, then the probability
obtained from the tables of binomial distribution is
doubled.

The Kruskal-Wallis Test

This test is used to test


whether several
populations have the
same median.

12
The test statistic is H
n(n 1)

R 2j

- 3(n 1) ,

where nj is the number of observations in the jth sample,


n is the total number of observations, and
Rj is the sum of ranks for the jth sample.

If each n j 5 and the null hypothesis is true,


then the distribution of K is 2 with dof c - 1,
where c is the number of sample groups.
In the case of ties, a corrected statistic should be computed:

Hc

H
(t 3j t j )

1-

n n

where tj is the number of ties in the


jth sample.

EXAMPLE

1. Three machines are used in the packaging of 10kg of wheat


flour. Each machine is designed so as to pack on average of
10kg of flour per bag. Samples of six bags were selected from
each machine and the amount of wheat packaged in each
bag is shown below:
Machin
e1

15.
8

15.
9

16. 15.
2
7

16. 15.
3
8

Machin
e2

16.
5

16

15. 15.
4
9

16. 16.
2
1

Machin
e3

15.
7

16.
4

16. 15.
2
9

15. 16.
7
3

Use a 5 per cent level of significance to test the hypothesis


that the amount of wheat packaged by the three machines is
the same.
Ho: amount of wheat packaged by the three machines is same.
H1: amount of wheat packaged by at least two machines is
different.

Solution:

Pool the elements of the different samples and rank them.


These ranking are shown below.

Machine1

R1

Machine2

R2

Machine3

R3

15.8

5.5

16.5

18

15.7

15.9

16

10

16.4

17

16.2

13

15.4

16.2

13

15.7

15.9

15.9

16.3

15.
5

16.2

13

15.7

15.8

5.5

16.1

11

16.3

15.
5

Total =

50.
5

Total =

61

Total =

59.
5

R1(Total of ranks from machine 1) = 50.5


R2(Total of ranks from machine 2) = 61
R3(Total of ranks from machine3) = 59.5
Therefore,
12
H
n(n 1)

R2
j

n j

- 3(n 1)

= 12/18(19)[50.5/6+61/6+59.5/6] -3(18+1)
=12/342[425.04+621.17+590.04] -57
= 12/342[1636.25] -57
= 19635/342 -57
= 57.41 57

= o.41

We know that H follows a


distribution with 2 degree of
freedom. The sample value of of
0.41 is to be compared with the
critical value of , which in the
present case is 5.99. As sample is
less than the critical , the null
hypothesis is accepted. Therefore,
there is no significant difference in
the amount of wheat packaged by
the three machines.

You might also like