Scribd
Upload
59 views23 pages

Nonparametric Statistics

Nonparametric statistical procedures do not assume a population distribution and can be used with ordinal or continuous data. The runs test determines if the pattern of two types of observations is random by counting the number of runs. The Mann-Whitney U test and Wilcoxon signed-rank test compare two independent samples or paired samples, respectively, without assuming a normal distribution. Kruskal-Wallis test is a nonparametric equivalent of one-way ANOVA for comparing more than two independent samples.

Uploaded by

divya arya
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
59 views23 pages

Nonparametric Statistics

Nonparametric statistical procedures do not assume a population distribution and can be used with ordinal or continuous data. The runs test determines if the pattern of two types of observations is random by counting the number of runs. The Mann-Whitney U test and Wilcoxon signed-rank test compare two independent samples or paired samples, respectively, without assuming a normal distribution. Kruskal-Wallis test is a nonparametric equivalent of one-way ANOVA for comparing more than two independent samples.

Uploaded by

divya arya
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
You are on page 1/ 23

Nonparametric Statistics

Nonparametric statistical procedures are


those statistical methods that do not
concern themselves with population
distributions and/or parameters.
The Runs Test

The runs test is a statistical procedure


used to determine whether the pattern of
occurrences of two types of observations
is determined by a random process.
The Runs Test

A run is a succession of occurrences of a


certain type preceded and followed by
occurrences of the alternate type or by no
occurrences at all.
The Runs Test
(Table 15-1)

Sequence Number Code Sequence Number Code


1 0.34561 - 11 0.67201 +
2 0.42789 - 12 0.23790 -
3 0.36925 - 13 0.24509 -
4 0.89563 + 14 0.01467 -
5 0.25679 - 15 0.78345 +
6 0.92001 + 16 0.69112 +
7 0.58345 + 17 0.46023 -
8 0.23114 - 18 0.38633 -
9 0.12672 - 19 0.60914 +
10 0.88569 + 20 0.95234 +
The Runs Test
(Small Sample Example)

H0: Computer-generated numbers are random


between 0.0 and 1.0.
HA: Computer-generated numbers are not random .

--- + - ++ -- ++ --- ++ -- ++
Runs: 1 2 3 4 5 6 7 8 9 10

There are r = 10 runs


From runs table (Appendix K) with n1 = 9
and n2 = 11, the critical value of r is 6
The Runs Test
(Small Sample Example)

Test Statistic:
r = 10 runs
Critical Values from Runs Table:
Possible
Runs: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Reject H0 Do not reject H0 Reject H0

Decision:
Since r = 10, we do not reject the null hypothesis.
Large Sample Runs Test
MEAN AND STANDARD DEVIATION FOR r
2n1n2
r  1
n1  n2
(2n1n2 )(2n1n2  n1  n2 )
r 
(n1  n2 ) (n1  n2  1)
2

where:
n1 = Number of occurrences of first type
n2 = Number of occurrences of second type
Large Sample Runs Test

TEST STATISTIC FOR LARGE


SAMPLE RUNS TEST
r  r
z
r
Large Sample Runs Test
(Example 15-2)

Table 15-2
OOOUOOUOUUOOUUOOOOUUOUUOOO
UUUOOOOUUOOUUUOUUOOUUUUU
OOOUOUUOOOUOOOOUUUOUUOOOU
OOUUOUOOUUUOUUOOOOUUUOOO

n1 = 53 “O’s” n2 = 47 “U’s”
r = 45 runs
Large Sample Runs Test
(Example 15-2)

H0: Yogurt fill amounts are randomly distributed above and below 24-ounce level.
H1: Yogurt fill amounts are not randomly distributed above and below 24-ounce
level.
 = 0.05
Rejection Region Rejection Region
 /2 = 0.025  /2 = 0.025

z.025  1.96 0 z.025  1.96

r   r 45  50.82
z   1.174
r 4.95659

Since z= -1.174 > -1.96 and < 1.96, we do not reject H0,
Mann-Whitney U Test
The Mann Whitney U test can be used to compare
two samples from two populations if the
following assumptions are satisfied:
• The two samples are independent and random.
• The value measured is a continuous variable.
• The measurement scale used is at least
ordinal.
• If they differ, the distributions of the two
populations will differ only with respect to
the central location.
Mann-Whitney U Test
U-STATISTICS

n1 (n1  1)
U1  n1n2    R1
2
n2 (n2  1)
U 2  n1n2    R2
2
where:
n1 and n2 are the two sample sizes
R1 and R2 = Sum of ranks for samples 1 and 2
Mann-Whitney U Test
- Large Samples -
MEAN AND STANDARD DEVIATION FOR
THE U-STATISTIC

n1n2

2
(n1 )(n2 )(n1  n2  1)

12
where:
n1 and n2 = Sample sizes from populations 1 and 2
Mann-Whitney U Test
- Large Samples -

MANN-WHITNEY U-TEST STATISTIC


n1n2
U
z 2
(n1 )(n2 )(n1  n2  1)
12
Mann-Whitney U Test
(Example 15-4)

H 0 : ~1  ~2  0
H A : ~1  ~2  0
  0.05
Rejection Region
 = 0.05

z  1.645 ~1  ~2  0


n1n2
U
2 27,412  29,088
z   1.027
(n1 )(n2 )(n1  n2  1) (144)(404)(144  404  1)
12 12

Since z= -1.027 > -1.645, we do not reject H0,


Wilcoxon Matched-Pairs Test
The Wilcoxon matched pairs signed rank test can be
used in those cases where the following
assumptions are satisfied:
• The differences are measured on a continuous
variable.
• The measurement scale used is at least
interval.
• The distribution of the population differences
is symmetric about their median.
Wilcoxon Matched-Pairs Test
WILCOXON MEAN AND STANDARD
DEVIATION

n(n  1)

4
n(n  1)(2n  1)

24
where:
n = Number of paired values
Wilcoxon Matched-Pairs Test

WILCOXON TEST STATISTIC


n(n  1)
T
z 4
n(n  1)(2n  1)
24
Kruskal-Wallis One-Way
Analysis of Variance
Kruskal-Wallis one-way analysis of variance can be
used in one-way analysis of variance if the
variables satisfy the following:
• They have a continuous distribution.
• The data are at least ordinal.
• The samples are independent.
• The samples come from populations whose only
possible difference is that at least one may have
a different central location than the others.
Kruskal-Wallis One-Way
Analysis of Variance
H-STATISTIC
k 2
12 R
H 
N ( N  1) i 1 ni
 3( N  1), with df  k  1
i

where:
N = Sum of sample sizes in all samples
k = Number of samples
Ri = Sum of ranks in the ith sample
ni = Size of the ith sample
Kruskal-Wallis One-Way
Analysis of Variance
CORRECTION FOR TIED RANKINGS
g

 (t 3
i  ti )
1 i 1
N N 3

where:
g = Number of different groups of ties
ti = Number of tied observations in the ith tied
group of scores
N = Total number of observations
Kruskal-Wallis One-Way
Analysis of Variance
H-STATISTIC CORRECTED FOR TIED
RANKINGS
k 2
12 R

N ( N  1) i 1 ni
 3( N  1)
i

H g

 (t 3
i  ti )
1 i 1
N N 3
Key Terms
• Kruskal-Wallis One- • Run
Way Analysis of • Runs Test
Variance
• Wilcoxon Test
• Mann-Whitney U Test
• Nonparametric
Statistical Procedure

You might also like