NONPARAMETRIC
STATISTIC TESTS
Parametric vs. Nonparametric statistic tests
Parametric statistics:
Tests: - Z-tests …
- T-tests …
- ANOVA tests …
Assumptions: - Data are randomly selected
- Data are normal distribution
- Interval or Ratio level data
- Large sample size (≥ 20)
- ……
Nonparametric statistics:
Tests: - Mann-Whitney U test
- Wilcoxon Signed Rank T test
- Kruskal-Wallis K test
- Friedman test
Assumptions: - Distribution-free
- Nominal or Ordinal data
- ……
20,
20,
Small sample procedure Large sample procedure
Using Mann-Whitney table ([Link]
Mann-Whitney U test: Examples 1
= [Link](…)
Assumption: distribution-free
samples are randomly
Mann-Whitney U test: Examples 1
Small samples size Conclusion ???
Small sample procedure Large sample procedure
Using Wilcoxon Signed Rank table ([Link]
signed-ranks-table/)
Wilcoxon Matched-Pairs Signed Rank Test: Examples 1
Test the hypothesis if there is difference in amount of
household spending on healthcare between 2 places:
Pennsylvania and California. The significant level is
0.05.
Wilcoxon Matched-Pairs Signed Rank Test: Examples 1
Assumption: distribution-free
samples are randomly
2 dependent samples
Wilcoxon Matched-Pairs Signed Rank Test: Examples 1
Small
sample size
Conclusion ???
1-right-tailed test
Conclusion???
The Randomized Block Design
experimental units is divided into b homogeneous groups of size c
homogeneous groups are called blocks
the treatments are then randomly assigned to the experimental units in
each block - one treatment to a unit in each block.
For randomized block designs, there are:
one factor is the primary interest
several other nuisance factors - may affect
the measured result, but not the primary
interest
Example
Suppose we are interested in how weight gain (Y) in rats
is affected by 2 factors:
- Source of protein (Beef, Cereal, and Pork)
- Level of Protein (High or Low)
There are a total of t = 32 = 6 treatment combinations
1. Beef - High Protein
2. Cereal - High Protein
3. Pork - High Protein
4. Beef - Low Protein
5. Cereal - Low Protein
6. Pork - Low Protein
Randomized Block Design
6 treatment combinations
* need at least 10 rats for each treatment
* same conditions to eliminate environmental factors
Think: 60 rats, in one place, at one time, feeding same time,
cleaning same time, same temperature, same light, same
life schedule, same exercise program…
Impossible for one person to do it alone
But, if you divide them, they will have different conditions
Then, you have randomized block design
Randomized Block Design
1 2 3
1 2 3 4 Stu.
5 B 6
Stu. A
4 5 6
High protein of pork will positively
affect the weight gain1 in rats
2 3
1 2 3 Stu. D
Stu. C 4 5 6
4 5 6
1 2 3
4 Stu.
5E 6
1. Beef - High Protein 2. Cereal - High Protein
3. Pork - High Protein 4. Beef - Low Protein
5. Cereal - Low Protein 6. Pork - Low Protein
Randomized Block Design
Block Block
1 107 96 112 83 87 90 6 128 89 104 85 84 89
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
2 102 72 100 82 70 94 7 56 70 72 64 62 63
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
3 102 76 102 85 95 86 8 97 91 92 80 72 82
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
4 93 70 93 63 71 63 9 80 63 87 82 81 63
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
5 111 79 101 72 75 81 10 103 102 112 83 93 81
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
1. Beef - High Protein 2. Cereal - High Protein
3. Pork - High Protein 4. Beef - Low Protein
5. Cereal - Low Protein 6. Pork - Low Protein
1-right-tailed test
Conclusion???
Quiz
Question 1
PBS PBS non-PBS non-PBS
24,500 62,500 41,000 43,500
0.01
39,400
36,800
61,400
53,000
32,500
33,000
51,900
27,800
Assumption: distribution-free
samples are randomly 44,300 27,000 21,000 21,500
57,960 59,600 40,500 39,500
32,000 48,200 32,400 38,500
61,000 25,040 16,000 15,850
34,000 37,190 21,500 16,700
43,500 18,900 39,500 29,680
55,000 33,500
39,000
Question 2
Test the hypothesis if there is difference in amount of
pollution level between 2 years: 1979 and 2011. The
significant level is 0.05.
18.5
Question 3
College students were assigned to three study methods in an
experiment to determine the effect of study technique on
learning. The three methods are: read only, read and
underline,and read and take notes. The test scores are
recorded. Test to compare the three group.
Test scores
Read only 15 14 16 13 11 14
Read and underline 15 14 25 10 12 14
Read and take notes 18 18 18 16 18 20
Question 4
A researcher examined Pet shop in a city. They
collected numbers of 3 types of animals on display by the
shop: fishes, reptiles, and mammals. The data was presented
with numbers of fishes, reptiles and mammals, respectively.
The researcher hypothesized that stores would tend to
display more fishes than other types of animals, reptiles and
mammals. Test the hypothesis at 0.05 of significant level.
