© 2011 Pearson Education, Inc

Statistics for Business and

Economics

Chapter 14
Nonparametric Statistics

© 2011 Pearson Education, Inc

 Content
14.1 Introduction: Distribution-Free Tests
14.2 Single Population Inferences
14.3 Comparing Two Populations:
Independent Samples
14.4 Comparing Two Populations: Paired
Difference Experiment
14.5 Comparing Three or More Populations:
Completely Randomized Design
© 2011 Pearson Education, Inc
 Content
14.6 Comparing Three or More Populations:
Randomized Block Design
14.7 Rank Correlation

© 2011 Pearson Education, Inc

 Learning Objectives

• Develop the need for inferential techniques

that require fewer, or less stringent,
assumptions than the methods of earlier
chapters
• Introduce nonparametric tests that are based
on ranks (i.e., on an ordering of the sample
measurements according to their relative
magnitudes)
© 2011 Pearson Education, Inc
 14.1

Introduction:
Distribution-Free Tests

© 2011 Pearson Education, Inc

 Parametric Test Procedures
• Involve population parameters
— Example: population mean

• Require interval scale or ratio scale

— Whole numbers or fractions
— Example: height in inches (72, 60.5, 54.7)

• Have stringent assumptions

— Example: normal distribution

• Examples: z-test, t-test, F-test, 2-test

© 2011 Pearson Education, Inc
 Nonparametric Test
Procedures
• Do not involve population parameters
— Example: probability distributions, independence
• Data measured on any scale
— Ratio or interval
— Ordinal
 Example: good-better-best
— Nominal
 Example: male-female
• Example: Wilcoxon rank sum test
© 2011 Pearson Education, Inc
Nonnormal Distributions -
t-Statistic is Invalid

© 2011 Pearson Education, Inc

 Distribution-Free Tests
Distribution-free tests are statistical tests that do
not rely on any underlying assumptions about the
probability distribution of the sampled
population.
The branch of inferential statistics devoted to
distribution-free tests is called nonparametrics.
Nonparametric statistics (or tests) based on the
ranks of measurements are called rank statistics
(or rank tests).
© 2011 Pearson Education, Inc
 Advantages of
Nonparametric Tests
• Used with all scales
• Easier to compute
— Developed originally before wide
computer use
• Make fewer assumptions
• Need not involve population
parameters
• Results may be as exact as
parametric procedures © 1984-1994 T/Maker Co.

© 2011 Pearson Education, Inc

 Disadvantages of
Nonparametric Tests
• May waste information
— If data permit using parametric © 1984-1994 T/Maker Co.

procedures
— Example: converting data from
ratio to ordinal scale
• Difficult to compute by hand for
large samples
• Tables not widely available

© 2011 Pearson Education, Inc

 Frequently Used
Nonparametric Tests
• Sign Test
• Wilcoxon Rank Sum Test
• Wilcoxon Signed Rank Test
• Kruskal Wallis H-Test
• Spearman’s Rank Correlation
Coefficient

© 2011 Pearson Education, Inc

 14.2

Single Population Inferences

© 2011 Pearson Education, Inc

 Sign Test

• Tests one population median,  (eta)

• Corresponds to t-test for one mean
• Assumes population is continuous
• Small sample test statistic: Number of sample
values above (or below) median
— Alternative hypothesis determines
• Can use normal approximation if n 30

© 2011 Pearson Education, Inc

 Sign Test Uses p-Value
to Make Decision
P(X)
P(X) Binomial: n = 8 p = 0.5
30%
30% .273
.273
.219
.219 .219
.219
20%
20%
.109
.109 .109
.109
10%
10%
.031
.031 .031
.031 .004
.004
.004 .004
0%
0%
00 11 22 33 44 55 66 77 88 X
X

P-value is the probability of getting an observation at least as

extreme as we got. If 7 of 8 observations ‘favor’ Ha, then
p-value = p(x  7) = .031 + .004 = .035.
If  = .05, then reject ©H2011
0 since p-value  .
Pearson Education, Inc
 Sign Test for a Population
Median 
One-Tailed Test
H0:  = 0
Ha:  > 0 [or Ha:  < 0 ]

Test statistic:
S = Number of sample measurements greater
than0 [or S = number of measurements less than
0]
© 2011 Pearson Education, Inc
 Sign Test for a Population
Median 
Observed significance level:
p-value = P(x ≥ S)
where x has a binomial distribution with
parameters n and p = .5
(Use Table II, Appendix B)

Rejection region: Reject H0 if p-value ≤ .05

© 2011 Pearson Education, Inc

 Sign Test for a Population
Median 
Two-Tailed Test
H0:  = 0
Ha:  ≠ 0

Test statistic:
S = Larger of S1 and S2, where S1 is the number of
sample measurements less than0 and S2 is the
number of measurements greater than 0
© 2011 Pearson Education, Inc
 Sign Test for a Population
Median 
Observed significance level:
p-value = 2P(x ≥ S)
where x has a binomial distribution with
parameters n and p = .5
(Use Table II, Appendix B)

Rejection region: Reject H0 if p-value ≤ .05

© 2011 Pearson Education, Inc

Conditions Required for Valid
Application of the Sign Test
The sample is selected randomly from a
continuous probability distribution.
[Note: No assumptions need to be made about the
shape of the probability distribution.]

© 2011 Pearson Education, Inc

 Large Sample Sign Test for a
Population Median 
One-Tailed Test
H0:  = 0
Ha:  > 0 [or Ha:  < 0 ]
( S −.5) −.5n
Test statistic: z =
.5 n
S = Number of sample measurements greater
than0 [or S = number of measurements less than
0] The “– .5” is the “correction
© 2011 Pearson Education, Inc
for continuity.”
 Large Sample Sign Test for a
Population Median 
The null hypothesized mean value is np = .5n,
and the standard deviation is

npq = n (.5) (.5) = .5 n

Rejection region: z > z

where tabulated z-values can be found in Table
IV of Appendix B.

© 2011 Pearson Education, Inc

 Large Sample Sign Test for a
Population Median 
Two-Tailed Test
H0:  = 0
Ha:  ≠ 0
( S −.5) −.5n
Test statistic: z =
.5 n

S = Larger of S1 and S2, where S1 is the number of

sample measurements less than0 and S2 is the
number of measurements greater
© 2011 Pearson Education, Inc than 0
 Large Sample Sign Test for a
Population Median 
The null hypothesized mean value is np = .5n,
and the standard deviation is

npq = n (.5) (.5) = .5 n

Rejection region: z > z

where tabulated z-values can be found in Table
IV of Appendix B.

© 2011 Pearson Education, Inc

 Sign Test Example
You’re an analyst for Chef-Boy-
R-Dee. You’ve asked 7 people
to rate a new ravioli on a 5-
point Likert scale (1 = terrible to
5 = excellent). The ratings are:
2 5 3 4 1 4 5
At the .05 level of significance,
is there evidence that the
median rating is less than 3?

© 2011 Pearson Education, Inc

 Sign Test Solution
• H0:  = 3 p-value:
• Ha:  < 3 P(x  2) = 1 – P(x  1)
= .937
•  = .05
(Binomial Table, n = 7,
• Test Statistic: p = 0.50)
S=2
(Ratings 1 & 2 are Decision:
Do not reject at  = .05
less than  = 3:
2, 5, 3, 4, 1, 4, 5) Conclusion:
There is no evidence
median is less than 3
© 2011 Pearson Education, Inc
 14.3

Comparing Two Populations:

Independent Samples

© 2011 Pearson Education, Inc

 Wilcoxon Rank Sum Test

• Tests two independent population probability

distributions
• Corresponds to t-test for two independent means
• Assumptions
— Independent, random samples
— Populations are continuous
• Can use normal approximation if ni  10

© 2011 Pearson Education, Inc

 Wilcoxon Rank Sum Test:
Independent Samples
Let D1 and D2 represent the probability
distributions for populations 1 and 2, respectively.
One-Tailed Test
H0: D1 and D2 are identical
Ha: D1 is shifted to the right of D2
[or D1 is shifted to the left of D2]
Test statistic:
T1, if n1 < n2; T2, if n2 < n1
(Either rank sum can be used if n1 = n2.)
© 2011 Pearson Education, Inc
 Wilcoxon Rank Sum Test:
Independent Samples
Rejection region:
T1: T1 ≥ TU [or T1 ≤ TL]
T2: T2 ≤ TL [or T2 ≥ TU]
where TL and TU are obtained from Table XIV of
Appendix B

Ties: Assign tied measurements the average of the ranks

they would receive if they were unequal but occurred in
successive order. For example, if the third-ranked and
fourth-ranked measurements are tied, assign each a rank
© 2011 Pearson Education, Inc
of (3 + 4)/2 = 3.5
 Wilcoxon Rank Sum Test:
Independent Samples
Let D1 and D2 represent the probability
distributions for populations 1 and 2, respectively.
Two-Tailed Test
H0: D1 and D2 are identical
Ha: D1 is shifted to the left or to the right of D2
Test statistic:
T1, if n1 < n2; T2, if n2 < n1
(Either rank sum can be used if n1 = n2.) We will
denote this rank sum© 2011 as T.Education, Inc
Pearson
 Wilcoxon Rank Sum Test:
Independent Samples
Rejection region:
T ≤ TL or T ≥ TU
where TL and TU are obtained from Table XIV of
Appendix B

Ties: Assign tied measurements the average of the ranks

they would receive if they were unequal but occurred in
successive order. For example, if the third-ranked and
fourth-ranked measurements are tied, assign each a rank
of (3 + 4)/2 = 3.5
© 2011 Pearson Education, Inc
Conditions Required for Valid
Wilcoxon Rank Sum Test
1. The two samples are random and independent.
2. The two probability distributions from which
the samples are drawn are continuous.

© 2011 Pearson Education, Inc

 Wilcoxon Rank Sum Test
Procedure
1. Assign ranks, Ri, to the n1 + n2 sample
observations
• If unequal sample sizes, let n1 refer to smaller-
sized sample
• Smallest value = 1
• Average ties

2. Sum the ranks, Ti, for each sample

3. Test statistic is Ti (smallest sample)
© 2011 Pearson Education, Inc
 Wilcoxon Rank Sum Test
Example
You’re a production planner. You want to see if
the operating rates for two factories is the same.
For factory 1, the rates (% of capacity) are 85,
82, 94, and 97. For factory 2, the rates are 71,
82, 77, 92, and 88 . Do the factory rates have
the same probability distributions at the
.10 level of significance?

© 2011 Pearson Education, Inc

 Wilcoxon Rank Sum Test
Solution
• H0: Identical Distrib.
• Ha: Shifted Left or Right
•  = .10
• n =4
1 n 5=
2

• Critical Value(s):

© 2011 Pearson Education, Inc

 Wilcoxon Rank Sum
Table (Portion)
 = .05 one-tailed;  = .10 two-tailed
n1
3 4 5 ..
TL TU TL TU TL TU ..
3 6 15 7 17 7 20 ..
n2 4 7 17 12 24 13 27 ..
5 7 20 13 27 19 36 ..
: : : : : : : :
© 2011 Pearson Education, Inc
 Wilcoxon Rank Sum Test
Solution
• H0: Identical Distrib.
• Ha: Shifted Left or Right
•  = .10
• n1 = 4 n2 = 5
• Critical Value(s):
Reject Do Not Reject
H0 Reject H0 H0

13 27  Ranks
© 2011 Pearson Education, Inc
Wilcoxon Rank Sum Test
Computation Table

Factory 1 Factory 2
Rate Rank Rate Rank
85 5 71 1
82 3 3.5 82 4 3.5
94 8 77 2
97 9 92 7
... 88 6
Rank Sum 25.5 19.5

© 2011 Pearson Education, Inc

 Wilcoxon Rank Sum Test
Solution
• H0: Identical Distrib. Test Statistic:
• Ha: Shifted Left or Right T1 = 5 + 3.5 + 8+ 9 = 25.5
(Smallest sample)
•  = .10
• n1 = 4 n2 = 5
• Critical Value(s): Decision:
Do not reject at  = .10
Reject Do Not Reject
H0 Reject H0 H0 Conclusion:
There is no evidence
13 27  Ranks
distrib.
© 2011 Pearson Education, Inc are not identical
Wilcoxon Rank Sum Test for Large
Samples (n1 ≥ 10 and n2 ≥ 10)

Let D1 and D2 represent the probability distributions for

populations 1 and 2, respectively.

One-Tailed Test
H0: D1 and D2 are identical
Ha: D1 is shifted to the right of D2
[or D1 is shifted to the left of D2]

© 2011 Pearson Education, Inc

Wilcoxon Rank Sum Test for Large
Samples (n1 ≥ 10 and n2 ≥ 10)
n1 ( n1 + n2 + 1)
T1 −
2
Test statistic: z =
n1n2 ( n1 + n2 + 1)
12

Rejection region:
z > z (or z < –z)
© 2011 Pearson Education, Inc
Wilcoxon Rank Sum Test for Large
Samples (n1 ≥ 10 and n2 ≥ 10)

Let D1 and D2 represent the probability

distributions for populations 1 and 2, respectively.

Two-Tailed Test
H0: D1 and D2 are identical
Ha: D1 is shifted to the right or to the left of D2

© 2011 Pearson Education, Inc

Wilcoxon Rank Sum Test for Large
Samples (n1 ≥ 10 and n2 ≥ 10)
n1 ( n1 + n2 + 1)
T1 −
2
Test statistic: z =
n1n2 ( n1 + n2 + 1)
12

Rejection region:
| z | > z/2
© 2011 Pearson Education, Inc
 14.4

Comparing Two Populations:

Paired Differences Experiment

© 2011 Pearson Education, Inc

 Wilcoxon
Signed Rank Test
• Tests probability distributions of two related
populations
• Corresponds to t-test for dependent (paired) means
• Assumptions
— Random samples
— Both populations are continuous
• Can use normal approximation if n  25

© 2011 Pearson Education, Inc

Wilcoxon Signed Rank Test for a
Paired Difference Experiment

Let D1 and D2 represent the probability

distributions for populations 1 and 2, respectively.

One-Tailed Test
H0: D1 and D2 are identical
Ha: D1 is shifted to the right of D2
[or D1 is shifted to the left of D2]
© 2011 Pearson Education, Inc
Wilcoxon Signed Rank Test for a
Paired Difference Experiment

Calculate the difference within each of the n

matched pairs of observations. Then rank the
absolute value of the n differences from the
smallest (rank 1) to the highest (rank n) and
calculate the rank sum T– of the negative
differences and the rank sum T+ of the positive
differences. [Note: Differences equal to 0 are
eliminated, and the number n of differences is
reduced accordingly.]
© 2011 Pearson Education, Inc
Wilcoxon Signed Rank Test for a
Paired Difference Experiment
Test statistic:
T–, the rank sum of the negative differences [or T+,
the rank sum of the positive differences] Rejection
region:
T– ≤ T0 [or T+ ≤ T0]
where T0 is given in Table XV of Appendix B
Ties: Assign tied measurements the average of the ranks
they would receive if they were unequal but occurred in
successive order. For example, if the third-ranked and
fourth-ranked measurements are tied, assign each a rank
of (3 + 4)/2 = 3.5 © 2011 Pearson Education, Inc
Wilcoxon Signed Rank Test for a
Paired Difference Experiment

Let D1 and D2 represent the probability

distributions for populations 1 and 2, respectively.

Two-Tailed Test
H0: D1 and D2 are identical
Ha: D1 is shifted to the left or to the right of D2

© 2011 Pearson Education, Inc

Wilcoxon Signed Rank Test for a
Paired Difference Experiment

Calculate the difference within each of the n

matched pairs of observations. Then rank the
absolute value of the n differences from the
smallest (rank 1) to the highest (rank n) and
calculate the rank sum T– of the negative
differences and the rank sum T+ of the positive
differences. [Note: Differences equal to 0 are
eliminated, and the number n of differences is
reduced accordingly.]
© 2011 Pearson Education, Inc
Wilcoxon Signed Rank Test for a
Paired Difference Experiment
Test statistic:
T–, the rank sum of the negative differences [or
T+, the rank sum of the positive differences]
Rejection region:
T ≤ T0
where T0 is given in Table XV of Appendix B
Ties: Assign tied measurements the average of the ranks
they would receive if they were unequal but occurred in
successive order. For example, if the third-ranked and
fourth-ranked measurements are tied,
© 2011 Pearson Education, Inc assign each a rank
of (3 + 4)/2 = 3.5
 Conditions Required for a
Valid Signed Rank Test
1. The sample of differences is randomly
selected from the population of differences.
2. The probability distribution from which the
sample of paired differences is drawn is
continuous.

© 2011 Pearson Education, Inc

 Signed Rank Test
Procedure
1. Obtain difference scores, Di = X1i – X2i
2. Take absolute value of differences, |Di|
3. Delete differences with 0 value
4. Assign ranks, Ri, where smallest = 1
5. Assign ranks same signs as Di
6. Sum ‘+’ ranks (T+) and ‘–’ ranks (T–)
1. Test statistic is T– or T+ (one-tail test)
2. Test statistic is T = smaller of T– or T+ (two-tail
test)
© 2011 Pearson Education, Inc
 Signed Rank Test
Computation Table
X1i X2i Di = X1i - X2i |Di| Ri Sign Sign Ri
X11 X21 D1 = X11 - X21 |D1| R1 ± ± R1
X12 X22 D2 = X12 - X22 |D2| R2 ± ± R2
X13 X23 D3 = X13 - X23 |D3| R3 ± ± R3
: : : : : : :
X1n X2n Dn = X1n - X2n |Dn| Rn ± ± Rn
Total T+ & T –
© 2011 Pearson Education, Inc
 Signed Rank Test
Example
You work in the finance department. Is the new
financial package faster (α = .05)? You collect the
following data entry times:
User Current New
Donna 9.98 9.88
Santosha 9.88 9.86
Sam 9.90 9.83
Tamika 9.99 9.80
Brian 9.94 9.87
Jorge 9.84 9.84

© 2011 Pearson Education, Inc © 1984-1994 T/Maker Co.

 Signed Rank Test
Solution
• H0: Identical Distrib.
• Ha: Current Shifted
Right
•  = .05
• n' =
• Critical Value(s):

© 2011 Pearson Education, Inc

 Signed Rank Test
Computation Table
X1i X2i Di |Di | Ri Sign Sign Ri
9.98 9.88 +0.10 0.10 4 + +4
9.88 9.86 +0.02 0.02 1 + +1
9.90 9.83 +0.07 0.07 2 2.5 + +2.5
9.99 9.80 +0.19 0.19 5 + +5
9.94 9.87 +0.07 0.07 3 2.5 + +2.5
9.84 9.84 0.00 0.00 ... ... Discard
Total T+= 15, T–= 0
© 2011 Pearson Education, Inc
 Signed Rank Test
Solution
• H0: Identical Distrib.
• Ha: Current Shifted
Right
•  = .05
• n' = 5 (not 6; 1 elim.)
• Critical Value(s):

© 2011 Pearson Education, Inc

 Wilcoxon Signed Rank
Table (Portion)
One-Tailed Two-Tailed n=5 n=6 n=7 ..
 = .05  = .10 1 2 4 ..
 = .025  = .05 1 2 ..
 = .01  = .02 0 ..
 = .005  = .01 ..
n = 11 n = 12 n = 13
: : : :
© 2011 Pearson Education, Inc
 Signed Rank Test
Solution
• H0: Identical Distrib.
Test Statistic:
• Ha: Current Shifted T– = 0
Right
•  = .05
• n' = 5 (not 6; 1 elim.)
• Critical Value(s): Decision:
Reject at  = .05
Reject Do Not
H0 Reject Conclusion:
H0 There is evidence new
1 T0 package
© 2011 Pearson Education, Inc is faster
 Wilcoxon Signed Rank Test for
Large Samples (n ≥ 25)

Let D1 and D2 represent the probability

distributions for populations 1 and 2, respectively.

One-Tailed Test
H0: D1 and D2 are identical
Ha: D1 is shifted to the right of D2
[or D1 is shifted to the left of D2]
© 2011 Pearson Education, Inc
Wilcoxon Signed Rank Test for a
Paired Difference Experiment

T+ −⎡ ⎣n ( n + 1) 4 ⎤
⎦
Test statistic: z =
n ( n + 1) ( 2 n + 1) 24

Rejection region:
z > z [or z < –z ]

© 2011 Pearson Education, Inc

Wilcoxon Signed Rank Test for a
Paired Difference Experiment

Assumptions:
The sample size n is greater than or equal to 25.
Differences equal to 0 are eliminated, and the
number n of differences is reduced accordingly.
Tied absolute differences receive ranks equal to
the average of the ranks they would have
received had they not been tied.

© 2011 Pearson Education, Inc

 Wilcoxon Signed Rank Test for
Large Samples (n ≥ 25)

Let D1 and D2 represent the probability

distributions for populations 1 and 2, respectively.

Two-Tailed Test
H0: D1 and D2 are identical
Ha: D1 is shifted to the left or to the right of D2

© 2011 Pearson Education, Inc

Wilcoxon Signed Rank Test for a
Paired Difference Experiment

T+ −⎡ ⎣n ( n + 1) 4 ⎤
⎦
Test statistic: z =
n ( n + 1) ( 2 n + 1) 24

Rejection region:
| z | > z

© 2011 Pearson Education, Inc

Wilcoxon Signed Rank Test for a
Paired Difference Experiment

Assumptions:
The sample size n is greater than or equal to 25.
Differences equal to 0 are eliminated, and the
number n of differences is reduced accordingly.
Tied absolute differences receive ranks equal to
the average of the ranks they would have
received had they not been tied.

© 2011 Pearson Education, Inc

 14.5

Comparing Three or More

Populations:
Completely Randomized Design

© 2011 Pearson Education, Inc

 Kruskal-Wallis H-Test
• Tests the equality of more than two (p)
population probability distributions
• Corresponds to ANOVA for more than two
means
• Used to analyze completely randomized
experimental designs
• Uses 2 distribution with p – 1 df
— if sample size nj ≥ 5
© 2011 Pearson Education, Inc
 Kruskal-Wallis H-Test for
Comparing k Probability
Distributions

H0: The k probability distributions are identical

Ha: At least two of the k probability distributions
differ in location.

12 2
Test statistic: H=
n ( n + 1)
∑ n j( R j − R )

© 2011 Pearson Education, Inc

 Kruskal-Wallis H-Test for
Comparing k Probability
Distributions
where
nj = Number of measurements in sample j
Rj = Rank sum for sample j, where the rank of
each measurement is computed according to
its relative magnitude in the totality of data
for the k samples
Rj = Rj/nj = Mean rank sum for jth sample
R = Mean of all ranks = (n + 1)/2
n = Total sample ©size = nEducation,
2011 Pearson
1 + nInc2 + . . . + nk
 Kruskal-Wallis H-Test for
Comparing k Probability
Distributions
Rejection region:
2
H > χα with (k – 1) degrees of freedom

Ties: Assign tied measurements the average of the ranks

they would receive if they were unequal but occurred in
successive order. For example, if the third-ranked and
fourth-ranked measurements are tied, assign each a rank
of (3 + 4)/2 = 3.5. The number should be small relative
to the total number of observations.
© 2011 Pearson Education, Inc
 Conditions Required for the
Validity of the Kruskal-Wallis
H-Test
1. The k samples are random and independent.
2. There are five or more measurements in each
sample.
3. The k probability distributions from which the
samples are drawn are continuous

© 2011 Pearson Education, Inc

 Kruskal-Wallis H-Test
Procedure
1. Assign ranks, Ri , to the n combined
observations
• Smallest value = 1; largest value = n
• Average ties
2. Sum ranks for each group
3. Compute test statistic Squared total of
each group
⎛ 12 R 2j ⎞
H =⎜ ∑ ⎟ − 3 (n + 1)
⎜ n (n + 1) n ⎟
⎝ © 2011 Pearson Education,
j ⎠
Inc
 Kruskal-Wallis H-Test
Example
As production manager, you Mach1 Mach2 Mach3
want to see if three filling 25.40 23.40 20.00
machines have different 26.31 21.80 22.20
filling times. You assign 15
24.10 23.50 19.75
similarly trained and
experienced workers, 5 per 23.74 22.75 20.60
machine, to the machines. At 25.10 21.60 20.40
the .05 level of significance,
is there a difference in the
distribution of filling times?

© 2011 Pearson Education, Inc

 Kruskal-Wallis H-Test
Solution
• H0: Identical Distrib.
• Ha: At Least 2 Differ
•  = .05
• df = p – 1 = 3 – 1 = 2
• Critical Value(s):

 = .05

0 5.991 ©2 2011 Pearson Education, Inc

 Kruskal-Wallis H-Test
Solution
Raw Data Ranks
Mach1 Mach2 Mach3 Mach1 Mach2 Mach3
25.40 23.40 20.00
26.31 21.80 22.20
24.10 23.50 19.75
23.74 22.75 20.60
25.10 21.60 20.40

© 2011 Pearson Education, Inc

 Kruskal-Wallis H-Test
Solution
Raw Data Ranks
Mach1 Mach2 Mach3 Mach1 Mach2 Mach3
25.40 23.40 20.00
26.31 21.80 22.20
24.10 23.50 19.75 1
23.74 22.75 20.60
25.10 21.60 20.40

© 2011 Pearson Education, Inc

 Kruskal-Wallis H-Test
Solution
Raw Data Ranks
Mach1 Mach2 Mach3 Mach1 Mach2 Mach3
25.40 23.40 20.00 2
26.31 21.80 22.20
24.10 23.50 19.75 1
23.74 22.75 20.60
25.10 21.60 20.40

© 2011 Pearson Education, Inc

 Kruskal-Wallis H-Test
Solution
Raw Data Ranks
Mach1 Mach2 Mach3 Mach1 Mach2 Mach3
25.40 23.40 20.00 2
26.31 21.80 22.20
24.10 23.50 19.75 1
23.74 22.75 20.60
25.10 21.60 20.40 3

© 2011 Pearson Education, Inc

 Kruskal-Wallis H-Test
Solution
Raw Data Ranks
Mach1 Mach2 Mach3 Mach1 Mach2 Mach3
25.40 23.40 20.00 14 9 2
26.31 21.80 22.20 15 6 7
24.10 23.50 19.75 12 10 1
23.74 22.75 20.60 11 8 4
25.10 21.60 20.40 13 5 3

© 2011 Pearson Education, Inc

 Kruskal-Wallis H-Test
Solution
Raw Data Ranks
Mach1 Mach2 Mach3 Mach1 Mach2 Mach3
25.40 23.40 20.00 14 9 2
26.31 21.80 22.20 15 6 7
24.10 23.50 19.75 12 10 1
23.74 22.75 20.60 11 8 4
25.10 21.60 20.40 13 5 3
Total 65 38 17

© 2011 Pearson Education, Inc

 Kruskal-Wallis H-Test
Solution
⎛ 12 Rj ⎞
2

H =⎜ ∑ ⎟ − 3 (n + 1)
⎜ n (n + 1) n j ⎟
⎝ ⎠
⎛ 12 ⎛ (65 )2 (38 )2 (17 )2 ⎞⎞
=⎜ ⎜ + + ⎟⎟ − 3 (16 )
⎜ (15 )(16 ) ⎜ 5 5 5 ⎟⎟
⎝ ⎝ ⎠⎠
⎛ 12 ⎞
=⎜ ⎟(191.6 ) − 48 12
⎝ 240 ⎠ 2

= 11.58
H=
n ( n + 1)
∑ n j( R j − R )
© 2011 Pearson Education, Inc
 Kruskal-Wallis H-Test
Solution
• H0: Identical Distrib.
Test Statistic:
• Ha: At Least 2 Differ H = 11.58
•  = .05
• df = p – 1 = 3 – 1 = 2
• Critical Value(s):
Decision:
Reject at  = .05
 = .05
Conclusion:
There is evidence population
0 5.991 ©2 2011 Pearsondistrib.
Education, Inc are different
 14.6

Comparing Three or More

Populations:
Randomized Block Design

© 2011 Pearson Education, Inc

 Friedman Fr-statistic

• provides another method for testing to detect a

shift in location of a set of k populations that
have the same spread (or, scale)
• is based on the rank sums of the treatments,
measures the extent to which the k samples
differ with respect to their relative ranks within
the blocks

© 2011 Pearson Education, Inc

 Friedman Fr-Test for a
Randomized Block Design
H0: The probability distributions for the k
treatments are identical
Ha: At least two of the probability distributions
differ in location
12b 2

Test statistic: Fr = k k + 1 ∑( R j −R )
( )

© 2011 Pearson Education, Inc

 Friedman Fr-Test for a
Randomized Block Design
where
b = Number of blocks
k = Number of treatments
Rj = Rank sum of the jth treatment, where the rank
of each measurement is computed relative to
its position within its own block
Test statistic:
2
χα
Fr > with (k – 1) degrees of freedom
© 2011 Pearson Education, Inc
 Friedman Fr-Test for a
Randomized Block Design
Ties: Assign tied measurements the average of
the ranks they would receive if they were
unequal but occurred in successive order. For
example, if the third-ranked and fourth-ranked
measurements are tied, assign each a rank of (3 +
4)/2 = 3.5. The number should be small relative
to the total number of observations.

© 2011 Pearson Education, Inc

 Friedman Fr -Test Example
Consider the data in the table. A pharmaceutical
firm wants to compare the reaction times of
subjects under the influence of three different
drugs that it produces. Apply the Friedman Fr-
test to the data. What conclusion can you draw?
Test using  = .05.

© 2011 Pearson Education, Inc

 Friedman Fr -Test Example
H0: The population distributions of reaction
times are identical for the three drugs
Ha: At least two of the three drugs have reaction
time distributions that differ in location
From the table: k = 3 treatments, b = 6 blocks,
R1 = 7, R2 = 12, R3 = 17; so R1 = 7 6 = 1.167,
R2 = 12 6 = 2.0, R3 = 17 6 = 2.833; R = ( 3 + 1) 2 = 2.0
12 ( 6 )
H= ⎡(1.167 −2.0 ) + ( 2.0 −2.0 ) + ( 2.833 −2.0 ) ⎤
2 2 2

( 3) ( 4 ) ⎣ ⎦
= 6 (1.388) = 8.33 © 2011 Pearson Education, Inc
 Friedman Fr -Test Example
For  = .05,  2.05 = 5.99147, therefore

Rejection region: H > 5.99147

© 2011 Pearson Education, Inc

 Friedman Fr -Test Example
Conclusion: Because H = 8.33 exceeds the
critical value of 5.99, we reject the null
hypothesis and conclude that at least two of the
three drugs have distributions of reaction times
that differ in location. That is, at least one of the
drugs tends to yield reaction times that are
faster than the others.

© 2011 Pearson Education, Inc

 14.7

Rank Correlation

© 2011 Pearson Education, Inc

 Spearman’s Rank
Correlation Coefficient
• Measures correlation between ranks
• Corresponds to Pearson product moment correlation
coefficient
• Values range from –1 to +1
• Formula (shortcut)

6∑ d i 2
rs = 1−
n ( n −1)
2

di = ui – vi (difference in ranks of ith observation

for samples 1 and 2)
n = number of pairs
© 2011 ofEducation,
Pearson observations
Inc
Spearman’s Rank Correlation
Procedure

1. Assign ranks, Ri , to the observations of each

variable separately
2. Calculate differences, di , between each pair of ranks
3. Square differences, di2, between ranks
4. Sum squared differences for each variable
5. Use shortcut approximation formula

© 2011 Pearson Education, Inc

 Spearman’s Rank
Correlation Example
You’re a research assistant for the FBI. You’re
investigating the relationship between a person’s attempts
at deception and percent Subj. Deception
changes in their pupil Pupil
size. You ask subjects
1 87 10
a series of questions,
some of which they 2 63 6
must answer 3 95 11
dishonestly. At the .05 4 50 7
5 correlation
level of significance, what is the 43 coefficient?
0

© 2011 Pearson Education, Inc

Spearman’s Rank Correlation
Table
2
Subj. Decep. R1i Pupil R2i di di
1 87 4 10 4 0 0
2 63 3 6 2 1 1
3 95 5 11 5 0 0
4 50 2 7 3 -1 1
5 43 1 0 1 0 0
Σdi2= 2
© 2011 Pearson Education, Inc
 Spearman’s Rank
Correlation Solution
6∑ di2
rs = 1 −
n (n 2 − 1)
6 (2 )
= 1−
5 (5 − 1)
2

= 1 − 0.10
= 0.90
© 2011 Pearson Education, Inc
 Spearman’s Nonparametric
Test for Rank Correlation
One-Tailed Test
H 0:  = 0
Ha:  > 0 (or Ha:  < 0)
6∑ d i 2
Test statistic: rs = 1−
n ( n −1)
2

Rejection region: rs > rs, or rs <–rs, when s < 0

where rs, is the value from Table XVI corresponding
to the upper-tail area  and n pairs of observations
© 2011 Pearson Education, Inc
 Spearman’s Nonparametric
Test for Rank Correlation

Ties: Assign tied measurements the average of the

ranks they would receive if they were unequal but
occurred in successive order. For example, if the
third-ranked and fourth-ranked measurements are
tied, assign each a rank of (3 + 4)/2 = 3.5. The
number should be small relative to the total number
of observations.

© 2011 Pearson Education, Inc

 Spearman’s Nonparametric
Test for Rank Correlation
Two-Tailed Test
H0:  = 0
Ha:  ≠ 0

Test statistic: 6∑ d i 2
rs = 1−
n ( n −1)
2

Rejection region: |rs | > rs,

where rs, is the value from Table XVI corresponding to
the upper-tail area  and n pairs of observations
© 2011 Pearson Education, Inc
 Spearman’s Nonparametric
Test for Rank Correlation

Ties: Assign tied measurements the average of the

ranks they would receive if they were unequal but
occurred in successive order. For example, if the
third-ranked and fourth-ranked measurements are
tied, assign each a rank of (3 + 4)/2 = 3.5. The
number should be small relative to the total number
of observations.

© 2011 Pearson Education, Inc

 Conditions Required for a
Valid Spearman’s Test

1. The sample of experimental units on which the

two variables are measured is randomly
selected.
2. The probability distributions of the two
variables are continuous.

© 2011 Pearson Education, Inc

 Key Ideas

Distribution-Free Tests
Do not rely on assumptions about the
probability distribution of the sampled
population

© 2011 Pearson Education, Inc

 Key Ideas
Nonparametrics
Distribution-free tests that are based on rank
statistics
One-sample nonparametric test for the
population median – sign test
Nonparametric test for two independent samples
– Wilcoxon rank sum test
Nonparametric test for matched pairs –
Wilcoxon signed rank test
© 2011 Pearson Education, Inc
 Key Ideas
Nonparametrics
Nonparametric test for a completely
randomized design – Kruskal-Wallis test
Nonparametric test for a randomized block
design – Friedman test
Nonparametric test for rank correlation –
Spearman’s test

© 2011 Pearson Education, Inc