You are on page 1of 10

Non-Parametric Statistics

 It is a distribution free test. That is statistical test developed without


any reference to a specific distribution.
 It is used when observations can not be measured but are available in
counts
 To discuss the subject in sequential form, we will explains non-
parametric statistics for each of the following categories:

 One Sample Test


 Two Sample Test
 Multivariate Sample Test
Non-parametric Statistics
Test Parametric Non-parametric
Test of one sample
proportion

p p n x
zcal  Pr( x  a )  n C x p x 1  p
 p 1  p  
 
 n  Pr  x  a   0 . 05

1  Pr  zcal  ztab   0 . 05

Ade, a FPY student harvested 5 sacks of good maize and 3 sacks of low quality
maize. Test the hypothesis that Ade supervisor will agree with him on the qualities
of the maize in the sack. Will Ade be satisfied with his supervisor judgement.
Solution
Pr  x  5 
The supervisor will inspect the maize and takes his decision .

Pr  x  5   0 . 2816  0 . 21122  0 . 10138  0 . 0304  0 . 005214


 0 . 629814

Will Ade be satisfied with his supervisor probability score? Please think like
a student, recall 0.37>0.05.
Test of one sample Parametric Non-parametric
mean (Wilcoxin)

T  mean
x  zcal 
t  var
  1 )
  mean 
n(n
 n  4
n(n  1 )( 2 n  1 )
var 
24

1  Pr  zcal  ztab   0 . 05
Number of students that arrived together at the farm gate were taken by the
security officer at the gate, suppose the statistics recorded are as follow:
10, 22, 15, 25, 17, 5, 12, 13, 17, 8; Test the hypothesis that an average of 15
students usually arrived together.

Solution:
Arrange: 5, 8, 10, 12, 13, 15, 17, 17, 22, 25
Rank: 1, 2, 3, 4, 5, 6, 7.5, 7.5, 9, 10
 
RTotal  15 , RTotal  40 40
 10

* 11 

zcal   4  1 . 27412


10 * 11 * 21  

 24 

Pr  zcal 1 . 27412   0 . 8987

Pr  zcal 1 . 27412   0 . 1013

Note that 0.1013 is greater than 0.05, hence we accept the null hypothesis
that an average of 15 students usually arrived together.
Test Parametric Non-parametric (Mann-Whitney U-test)

Test of two sample


means x1  x2  R1 
n1 ( n1 1 )
 R2 
n2 ( n2 1 )
t  u1
2
; u2
2
 2
 2
 mean
1
 2
zcal 
u
n1 n2 var
n1 n 2 n1 n 2  n1  n2 1 
mean  ; var 
2 12

1  Pr  zcal  ztab   0 . 05
Supposing the management decided to put counter at strategic location at
Odogbolu and Isaga-orile FUNAAB farm settlement to monitor the
arrival of FPY students at duty post everyday. Given the following
statistics:
Odogbolu : 10, 25, 15, 10, 16, 17, 8
Isaga-Orile: 15, 22, 10.
Test at 5% significant level, that the average arrival frequency of student
is the same in each farm settlement.
Solution:
First rank the entire data together. See the ranks in bracket. Obtain total
rank
Odogbolu : 10 (3), 25(10), 15(5.5), 10(3), 16(7), 17(8), 8(1) =37.5
Isaga-Orile: 15(5.5), 22(9), 10(3)=17.5.
n1 n 2 n1 n 2  n1  n2 1  21 11 
mean   10 . 5; var    19 . 25
2 12 12
17 . 5  10 . 5 7
zcal    1 . 59545
19 . 25 4 . 3875
Pr  zcal  1 . 59545   0 . 94469
Pr  zcal  1 . 59545   1  0 . 94469  0 . 05531

Since the p-value 0.05531 > 0.05, then we accept the null
hypothesis that average number of students that arrived
together in each farm settlement is the same.
Test of several means Parametric Non-parametric
Analysis of variance Kruskal wallis

In this case, we have several groups,


First we will rank the entire data N size data together
Second we will obtain the total rank in each `ith’ group
R , i  1 , 2 ,...
i

Thirdly evaluate the formula

 12 k
Ri 
2

H cal  
 N  N  1   n   3  N  1 
 i 1 i 

Pr  H cal   cal  h   0 . 05
2

If the p-value is >0.05 accept that all the groups are the same.
Assignment
• Alabata: 5, 8, 10, 12, 13, 15, 17
• Odogbolu : 10, 25, 15, 10, 16, 17, 8
• Isaga-Orile: 15, 22, 10, 22, 25

Test the data if the average frequency of arrival of students


in the three farm settlement is the same or not.

You might also like