Busmath42: L2 Nonparametric Methods
Business Mathematics 42 – Management Science I
LESSON 2: Nonparametric Methods
Parametric Methods one and two sampled test, ANOVA, Regression etc;
➔ uses normal distribution na may table of values where assumption is there
➔ More on interval and ratio scale of measurement; hypothesis is based on mean
Nonparametric Methods counterpart of Parametric Method
Important things to remember in Parametric and Nonparametric methods
● In the Parametric test, the selection of statistical data is based on the hypothesis you need to proved.
● If there is no hypothesis, then there is no need of the statistical test.
● Parametric and Nonparametric test are mostly used in research.
● Impossible to collect data from the entire population.
● Sample is the solution.
● From sample, we can find quantities like (mean, standard deviation, variance, etc.). These quantities are called the
PARAMETERS.
● Parameters are estimation of the population.
● Estimates quantities from sample is called “statistics”.
● “Statistics estimates the Parameters”
● Parameter is the characteristic of the population or aspect of the population.
● Parameter is the assumption about the population
Nonparametric Test
● Nonparametric test is one that makes no assumptions about the specific shape of the population from which a
sample is drawn.
● A nonparametric test should be used instead of its parametric counterpart whenever
1. Data are of the nominal or ordinal scale of measurement, or
2. Data are of the interval or ratio scale but one or more other assumptions, such as the normality of the underlying
population distribution, are not met.
● Note that both Parametric and Nonparametric testing rely on the basic principles of hypothesis testing.
PARAMETRIC NONPARAMETRIC
● Assumes the date is normally distributed (bell- ● Has no assumption about the population
shaped)
● Parametric Statistics is based on parameters of ● Also known as the Distribution Free Test
normal curve (mean in the center and standard
deviation to measure distance away from the mean
● Assumption made through the sample distribution ● It is not based on the parameters of normal curve
● If the info about the population from which the ● Can use for any shape of distribution; Not based on the
sample has been drawn is completely known through parameters of normal curve
its parameters then the test is Parametric test ● If the data are not within bell-shaped distribution= then
use nonparametric
PARAMETRIC NONPARAMETRIC
One Sample test Sign Test
T-test for two independent samples Wilcoxon Rank Sum Test
Two-sample t-test Mann-Whitney U-test
Two-Sample t-test (Paired) Wilcoxon Signed-Ranks Test
One-way ANOVA Moods median test
Two-way ANOVA Friedman Test
Pearson Correlation Spearman Rank Correlation
Advantages and Disadvantages of Nonparametric Testing
Page 1 of 13
Busmath42: L2 Nonparametric Methods
Advantages
1. Fewer assumptions about the population. Most importantly, the population need not be normally distributed or
approximately so. Nonparametric tests do not assume the population has any specific distribution.
2. The techniques can be applied when sample sizes are very small.
3. Samples with data of the nominal and ordinal scales of measurement can be tested.
Disadvantages
1. Compared to a parametric test, the information in the data is used less efficiently, and the power of the test will
be lower. For this reason, a parametric test is preferable whenever its assumptions have been met
2. Nonparametric testing places greater reliance on statistical tables, if a computer statistical package or spreadsheet
is not being used.
Nonparametric Methods
● Sign Test
● Median Test
● Wilcoxon Signed Rank Test
● Wilcoxon Rank-sum Test
● Kruskal-Wallis Analysis
SIGN TEST
● Sign test is a method whether the sample could have been drawn from a population having hypothesized value as its
median. In Sign test, it assumes that data are ordinal instead of interval or ratio. I
● n the sign test, the difference between two data values and hypothesized median (one-sample test) or the difference
between two data values (Paired-Sample Test)is replaced with a plus (+) or a minus (-) sign indicating the direction of
the difference.
● Is based on the sign difference between two related observations.
● (+) sign for positive difference; (-) sign for negative difference
● Not concerned with the magnitude of difference only the direction of difference
● Any sample that has no difference must be dropped from analysis
● The sign test relies on the binomial distribution and the fact that, if 𝑚_𝑑, the population median for 𝑑𝑖=𝑥𝑖 𝑦𝑖, is
actually zero, P(+) will be 0.5 for any pair of observations that are not toed.
○
Example 1:
Calorie Watchers has low calorie breakfasts, lunches, and dinners. If you join the club you receive two packaged
meals a day. CW claims that you can eat anything you want for the third meal and still lose at least five pounds the first
month. Members of the club are weighed before commencing the program and again at the end of the first month.
The experiences of a random sample of 11 enrollees are:
Name Weight Change
Foster Lost
Taoka Lost
Lange Gained
Rouses Lost
Stephens No Change
Cantrell Lost
Hercher Lost
Cander Lost
Hinckle Lost
Hinkley Lost
Justin Lost
We are interested in whether there has been a weight loss as a result of the Calorie Watchers Program.
a. State the null and alternative hypothesis
Page 2 of 13
Busmath42: L2 Nonparametric Methods
b. Using the 0.05 level of significance, what is the decision rule?
c. What is your conclusion about the Calorie Watchers program?
SOLUTION:
o Tinimbang before and after the program to know if may changes
o May 2 Possibilities 1st: May weight Loss, 2nd: May weight gain
o Merong probability of 50% kasi its either loss or gain weight
a) State the null and alternative hypothesis.
Pie = 0.50 kasi 2 lang possible result
b) What is the decision rule?
Observe than π = 50% or 0.50 and n = 10 n = 11-1 (No change) = 10
Pag walang pagbabago sa program then idrodrop or aalisin na sa analysis para malaman ang number or test na gagamitin
CV=9, reject the null hypothesis if there are 9 or more plus signs.
Critical Value: Start from bottom to top ang pagaadd hanggang mameet yung Significance Level. It should be either
lower or equal to NOT greater than
0.001 + 0.0098 = 0.0108
NOTE: Iadd yung 0.0439 pa ulit kasi di pa naaabot yung Significance Level na 0.05
0.0108 of 8 + 0.0439 of 9 = 0.0547 🡪 exceed na sa 0.05
So, it is now between 8 and 9, ang kailangan is yung lower value na 0.0108 so ang CV is 9
c) What is your conclusion about the Calorie Watchers program?
Reject null hypothesis since the number of plus sign is 9 which is equal to the critical value 9.
Plus Signs Always reject for plus signs in decision; refer sa number of people with weight loss. Yun mga nabawasan
ng timbang or may weight loss. Yun ang plus sign kasi ang goal is pumayat or mabawasan ang timbang
Conclusion effective yung program kasi nireject yung null hypothesis
Exercise:
The director of information systems at Samuelson Chemicals recommended that an in-plant training program be
instituted for certain managers. The objective is to improve the computer knowledge base in the Payroll, Accounting, and
Production Planning Departments. A sample of 15 managers is randomly selected from the three departments. Each
manager is rated on their computer knowledge by a panel of experts. They were rated as being either outstanding,
excellent, good, fair, or poor. After the three-month training program, the same panel of experts rated each manager again.
The two ratings (before and after) are shown along with the sign of the difference. A “+” sign indicates improvement and
a “-” sign indicates that the manager’s competence using databases had declined after the training program.
a. State the null and alternative hypothesis
b. Using the 0.10 level of significance, what is the decision rule?
c. What is your decision?
Page 3 of 13
Busmath42: L2 Nonparametric Methods
Name Before After Sign
T, J. bowers Good Outstanding +
Sue Jenkins Fair Excellent +
James Brown Excellent Good -
Tad Jackson Poor Good +
Andy Love Excellent Excellent 0
Sarah Truett Good Outstanding +
Antonia Aillo Poor Fair +
Jean Unger Excellent Outstanding +
Coy Farmer Good Poor -
Troy Archer Poor Good +
V.A. Jone Good Outstanding +
Juan Guillen Fair Excellent +
Candy Fry Good Fair -
Arthur Seiple Good Outstanding +
Sandy Gumpp Poor Good +
Ordinal Outstanding, excellent, good, fair, poor; May hierarchy
Pag di nagimprove magiging negative (-) sign. Pag nagimprove positive (+)
SOLUTION:
a. State the null and alternative hypothesis.
𝐻_0: 𝜋=0.5 𝐻_1: 𝜋 0.5
b. Using the 0.10 level of significance, what is the decision rule?
Observe that n=14, and 𝜋=0.50
n = 14 kasi di kasama si Andy Love kasi no change
CV=10, reject the null hypothesis if there are 10 or more plus signs.
CV = 0.0001 + 0.0009 + 0.0056 + 0.0222 + 0.0611 = 0.0899 🡪 10
c. What is your decision?
We reject the null hypothesis since the number of plus sign exceeds our Critical Value which is 10.
Plus sign 🡪 the basis, sign for improvement, ginogoal yung positive signs such as improving
Conclusion 🡪 Program showed changes in the managers, there’s improvement kaya + sign
Using the Normal Approximation to the Binomial
Page 4 of 13
Busmath42: L2 Nonparametric Methods
( x ± 0.50 ) −0.50 n
Test statistics z is z=
.50 √ ❑
( x−0.50 ) −0.50 n
● If the positive signs are more than n/2 use: z=
.50 √ ❑
( x+ 0.50 )−0.50 n
● If the positive signs are less than n/2 use: z=
.50 √ ❑
Pagkuha ng Z values
X = positive signs; 0.50n = population mean (pie n); 0.50n = sigma, sd, (pie - n)
More than 🡪 Minus, Negative ; Less than 🡪 Plus, Positive
EXAMPLE:
A sample of 45 overweight men participated in an exercise program At the confusion of the program, 32 had lost weight
At the 0.05 significance level, can we conclude the program is effective?
a. State the null hypothesis and the alternative hypothesis
b. State the decision rule
c. Compute the value of the test statistic
d. What is your decision regarding the null
hypothesis?
SOLUTION:
a) State the null hypothesis and the alternative hypothesis
𝐻0: 𝜋 0.50
𝐻1: 𝜋 > 0.50
b) State the decision rule.
At 0.05 significance level
Reject the null hypothesis if the computed value is greater
than 1.645
Critical Value 🡪 hanapin ang 0.05 sa standard normal table don sa mga areas;
which falls in 0.0505 and 0.0495 🡪 (1.64 + 1.65) /2 = 1.645
CV sign is nakabased sa direction ng Alternative Hypothesis (H1) which is pointing to the right so positive.
Thus, CV = +1.645
C) Compute the value of the test statistic
At n=45, n/2=22.5
( x ± 0.50 ) −0.50 n
Test statistics z is z=
.50 √ ❑
( x−0.50 ) −0.50 n
● If the positive signs are more than n/2 use: z=
.50 √ ❑
● Positive Signs = 32 > n/2 of 22.5. So use negative/minus
( 32−0.50 )−0.50(45)
z= 31.5 – 22.5 = 9 / 3.354 = 2.68 ANS: 2 to 4 decimal places
.50 √ ❑
D) What is your decision regarding the null hypothesis?
Reject the null hypothesis since the computed value is greater than the critical value 1.645
Page 5 of 13
Busmath42: L2 Nonparametric Methods
Decision Rule: Reject the null hypothesis if the computed value is greater than 1.645
MEDIAN TEST
● Hypothesized value based on Median; Testing hypothesis about a median
● TESTING A HYPOTHESIS ABOUT A MEDIAN
● Is used to test a hypothesis about a population median.
Example
The median salary for a chiropractor in the United States is $81,500 per year, according to the US Department of Labor.
A group of recent graduates believe this amount is too low. In a random sample of 205 chiropractors who recently
graduated, 170 began with a salary of more than $81,500 and five earned a salary of exactly $81,500.
a. State the null and alternative hypothesis.
b. State the decision rule. Use the 0.05 significance level.
c. Do the necessary computations and interpret the results
SOLUTION:
a) State the null and alternative hypothesis.
𝐻0: 𝑚𝑒𝑑𝑖𝑎𝑛 $81, 500
𝐻1: 𝑚𝑒𝑑𝑖𝑎𝑛 > $81, 500
Alternative Hypothesis
stated sa prob na more than so yun ang H1 and the opposite is the null hypothesis; Always based sa dulon
b) State the decision rule. Use the 0.05 significance level.
CV=1.645, if the computed value is greater than 1.645, then reject the null hypothesis.
( x−0.50 ) −0.50 n
c) Do the necessary computations and interpret the results. z=
.50 √ ❑
n=200 n = 205 – 5 = 200 The 5 people earns exactly 81,500 so walang pinagbago sa media thus idrodrop sila
x = 170 > than n/2 = 100 Use Negative/Minus
( 170−0.50 )−0.50 (200)
z= 169.5 – 100 = 69.5 / 7.0711 = 9.83
.50 √ ❑
Reject the null hypothesis,
Conclusion: The Chiropractor earn more than $81,500
WILCOXON SIGNED-RANK TEST
→ Sign is ranked from lowest to highest regardless of sign
● Normality assumption is not required
● Data must be at least ordinal scale, and the samples must be dependent
● Reject the null hypothesis if the T-value is smaller than the critical
➔ Ordinal Scale 🡪 lowest measurement in this test, bawal maging nominal
➔ Variables should have ordinality
➔ Samples must be dependent 🡪 the next sample is affected by the 1st sample
➔ Sa other test ang kinukuha is higher that CV but in Wilcoxon ang kinukuha yung smaller value than critical
value to reject null hypothesis
➔ Only test that uses less than in Critical Value
STEPS:
Page 6 of 13
Busmath42: L2 Nonparametric Methods
1. Rank absolute differences between the related observations.
a. After kunin ang differences between the observation, kunin mo absolute value (gagawing positive and
negative differences) and rank them from lowest to highest
2. Apply the sign of the differences to the ranks.
3. Sum negative ranks and positive ranks.
4. The smaller of the two sums is the computed T value.
a. Always yung smaller value and kukunin.
b. T Value 🡪 value compared to Critical Value in order to interpret based on H0 and H1
5. Refer to Appendix B.7 for the critical value, and make a decision regarding null hypothesis
a. Table for Wilcoxon has values for 2 tailed and 1 tailed so no need to divide by 2
One Tailed directional, inequality, <, > Two Tailed nondirectional; equality =,/=
It is based on hypothesis
EXAMPLE:
An industrial psychologist selected a random sample of seven young urban professional couples who own their
homes. The size of their home (square feet) is compared with that of their parents. At the 0.05 significance level, can we
conclude that the professional couples live in larger homes than their parents?
a) State the null hypothesis and alternative hypothesis
b) At 0.05 significance level
c) What is your decision regarding the null hypothesis?
SOLUTION:
Given:
Couple Name Professional Parent Difference Rank R+ R-
Gordon 1725 1175 550 7 7
Sharkey 1310 1120 190 5 5
Usehelding 1670 1420 250 6 6
Bell 1520 1640 -120 3 3
Kuhlman 1290 1360 -70 1 1
Welch 1880 1750 130 4 4
Anderson 1530 1440 90 2 2
24 T=4
Rank 🡪 The differences are considered positive regardless of the sign.
Pag ipagmaminus ibased sa alternative hypothesis (H1). Professional couples live in larger homes so Professional minus
Parents.
T Value would be 4 which is smaller between the R+ and R
a) State the null hypothesis and alternative hypothesis
𝐻0: 𝑇ℎ𝑒 𝑝𝑟𝑜𝑓𝑒𝑠𝑠𝑖𝑜𝑛𝑎𝑙 𝑐𝑜𝑢𝑝𝑙𝑒𝑠 𝑙𝑖𝑣𝑒 𝑖𝑛 𝑠𝑚𝑎𝑙𝑙𝑒𝑟 ℎ𝑜𝑚𝑒𝑠 𝑡ℎ𝑎𝑛 𝑡ℎ𝑒𝑖𝑟 𝑝𝑎𝑟𝑒𝑛𝑡𝑠
𝐻1: 𝑇ℎ𝑒 𝑝𝑟𝑜𝑓𝑒𝑠𝑠𝑖𝑜𝑛𝑎𝑙 𝑐𝑜𝑢𝑝𝑙𝑒𝑠 𝑙𝑖𝑣𝑒 𝑖𝑛 𝑙𝑎𝑟𝑔𝑒𝑟 ℎ𝑜𝑚𝑒𝑠 𝑡ℎ𝑎𝑛 𝑡ℎ𝑒𝑖𝑟 𝑝𝑎𝑟𝑒𝑛𝑡𝑠
b) At 0.05 significance level
CV=3, Reject the null hypothesis if the computed value is less than 3.
Page 7 of 13
Busmath42: L2 Nonparametric Methods
n=7 SL = 0.05 CV = 3 at one-tailed
c) What is your decision regarding the null hypothesis?
Since the T-value is 4 which is greater than the critical value 3, then do not reject the null hypothesis. Therefore, there is
no difference in the sizes of two basis
Therefore, the professional couples live in smaller homes than their parents
Page 8 of 13
Busmath42: L2 Nonparametric Methods
EXERCISE:
Fricker’s is a family restaurant chain located primarily in the southeastern part of the United States. It offers a full
dinner menu, but its specialty is chicken. Recently, Bernie Frick, the owner and founder, developed a new spicy flavor for
the batter in which the chicken is cooked. Before replacing the current flavor, he wants to conduct some tests to be sure
that patrons will like the spicy flavor better. To begin, Bernie selects a random sample of 15 customers. Each sampled
customer is given a small piece of the current chicken and asked to rate its overall taste on a scale of 1 to 20. A value near
20 indicates the participant liked the flavor, whereas a score near 0 indicates they did not like the flavor. Next, the same
15 participants
a) State the null hypothesis and alternative hypothesis
b) At 0.05 significance level
c) What is your decision regarding the null hypothesis?
SOLUTION:
Participant Spicy Current Difference Rank R+ R-
Arquette 14 12 2 1 1
Jone 8 16 -8 6 7 6 7
Fish 6 2 4 3 3
Wagner 18 4 14 13 13
Badenhop 20 12 8 6 7 6 7
Hall 16 16 0 * * *
Fowler 14 5 9 9 10 9 10
Virost 6 16 -10 11 11
Garcia 19 10 9 9 10 9 10
Sundar 18 10 8 6 7 6 7
Miller 16 13 3 2 2
Peterson 18 2 16 14 14
Bogart 4 13 -9 9 10 9 10
Hein 7 14 -7 4 4
Whitten 16 4 12 12 12
Difference Rank R+ R-
2 1 1
-8 6 7 6 7
4 3 3
14 13 13
8 6 7 6 7
0 * * *
9 9 10 9 10
-10 11 11
9 9 10 9 10
8 6 7 6 7
3 2 2
16 14 14
-9 9 10 9 10
-7 4 4
12 12 12
a) State the null hypothesis and alternative hypothesis
𝐻0: 𝑇ℎ𝑒𝑟𝑒 𝑖𝑠 𝑛𝑜 𝑑𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑐𝑒 𝑏𝑒𝑡𝑤𝑒𝑒𝑛 𝑡ℎ𝑒 𝑡𝑤𝑜 𝑓𝑙𝑎𝑣𝑜𝑟𝑠.
𝐻1: 𝑇ℎ𝑒 𝑠𝑝𝑖𝑐𝑦 𝑓𝑙𝑎𝑣𝑜𝑟 𝑟𝑎𝑡𝑖𝑛𝑔 𝑖𝑠 ℎ𝑖𝑔ℎ𝑒𝑟.
b) At 0.05 significance level
Page 9 of 13
Busmath42: L2 Nonparametric Methods
CV=25, Reject the null hypothesis if the computed value is less than 25.
c) What is your decision regarding the null hypothesis?
Since the T-value is 30 which is greater than the critical value 25, then do not reject the null hypothesis. Therefore, there
is no difference in the sizes of two flavors.
Page 10 of 13
Busmath42: L2 Nonparametric Methods
WILCOXON RANK-SUM TEST
● Is used to test whether two independent samples came from equal population
● No assumption about the shape of the population is required
● The data must be at least ordinal scale
● Each sample must contain at least eight observations
○ Bawal na below 8 observations ang each sample
● To determine the value of the test statistic W, the sample observations are ranked from low to high as if they were
from a single group.
○ Rank population A and B as if they are one pop. Rank immediately no differences needed
● The sum of ranks for each of the two samples is determined.
● W is used to compute Z. where W is the sum of the ranks for population 1.
n1 ( n1 +n2 +1 )
W−
2
Z=
√❑
➔ Use normal distribution table thus may Z-values to get the critical value
➔ Population 1 always the one na may mababang sample or observation. If equal number of populations, ang
kukunin is yung naunang sinabi sa problem
EXAMPLE:
The following observations were randomly selected from populations that were not necessarily normally
distributed. Use the 0.05 significance level, a two-tailed test, and the Wilcoxon rank-sum test to determine whether there
is a difference between the two populations.
Population A: 38, 45, 56, 57, 61, 69, 70, 79
Population B: 26, 31, 35, 42, 51, 52, 57, 62
SOLUTION:
(Two-tailed, not equal or equal)
H0: There is no difference between the 2 population
H1: There is a difference between the 2 population
POP A (8 samples) Rank POP B (8 samples) Rank
38 4 26 1
45 6 31 2
56 9 35 3
57 (10+11)/2 = 10.5 42 5
61 12 51 7
69 14 52 8
70 15 57 10.5
79 16 62 13
Total 86.5 Total 49.5
Use population A which is the first given kasi equal yung sample or population ni A and B
W = 86.5 n1 = 8 n2 = 8 z = 1.9429
8 ( 8+8+1 )
86.5−
2
Z=
√❑
CV = +-1.96, reject H0 if Z-Value > +1.96 or < -1.96
1.9429 < 1.96
Z-value < CV, Do not reject H0. So accept claim that there is no difference between 2 populations
Page 11 of 13
Busmath42: L2 Nonparametric Methods
KRUSKAL-WALLIS TEST: ANALYSIS OF VARIANCE BY RANKS
● Is used to test whether several populations are the same.
● No assumption regarding the shape of any of the population is required
● The sample must be independent and at least ordinal scale
● The sample observations are ranked from smallest to largest as though they were a single group
● The test statistics follows the chi-square distribution, provided there are at least five observations in each sample
● The value of the test statistic is computed from the following
[ ( ∑R )
]
❑ −3 ( n+ 1) ❑ ❑
(+ ∑ R )
❑ k
❑
+…+
(∑❑ R1 ) ❑
❑ 2
❑
12 ❑
H=
n( n+1) ❑
EXAMPLE:
The following sample data were obtained from three populations that did not follow a normal distribution.
Sample 1 Sample 2 Sample 3
50 48 39
54 49 41
59 49 44
59 52 47
65 56 51
57
a. State the null hypothesis.
b. Using the 0.05 level of risk, state the decision rule.
c. Compute the value of the test statistic
d. What is your decision on the null hypothesis?
SOLUTION:
a. State the null hypothesis.
𝐻𝑜:𝐴𝑙𝑙 𝑝𝑜𝑝𝑢𝑙𝑎𝑡𝑖𝑜𝑛𝑠 𝑎𝑟𝑒 𝑒𝑞𝑢𝑎𝑙.
𝐻1: 𝑁𝑜𝑡 𝑎𝑙𝑙 𝑝𝑜𝑝𝑢𝑙𝑎𝑡𝑖𝑜𝑛 𝑎𝑟𝑒 𝑡ℎ𝑒 𝑠𝑎𝑚𝑒
b. Using the 0.05 level of risk, state the decision rule.
At 0.05 level of significance and degrees of freedom equal to 2 (k-1), the critical value is 5.991
DR: Reject the null hypothesis, if the computed value is greater than 0.5991
c. Compute the value of the test statistic
[ ]
2 2 2
12 ( 64 ) ( 53 ) ( 19 )
H= + +…+ −3 ( 16+1 )=8.9809
16(16+1) 5 6 5
H=0.044117647 [ 819.2+468.17 +…+72.2 ]−51=8.9810
d. What is your decision on the null hypothesis?
Reject the null hypothesis since the computed value is greater than 0.5991.
NOTE!
TEST TABLE TO USE
Sign Test Binomial. Probability Distribution Table
Normal Approximation to binomial Areas Under Normal Curve (Z-Table)
Page 12 of 13
Busmath42: L2 Nonparametric Methods
Median Test Areas Under Normal Curve (Z-Table)
Wilcoxon Signed-Rank Test Wilcoxon T table
Wilcoxon Rank-Sum Test Areas Under Normal Curve (Z-Table)
Kruskal – walis Test Chi-square Table
Page 13 of 13
