department of department of

statistics Learning Outcomes statistics Parametric vs. Nonparametric Statistical Analysis

Nonparametric Methods At the end of this lesson, you would be able to: Parametric tests assume underlying statistical distributions in the data.
Therefore, several conditions of validity must be met so that the result of a
Glyzel Grace M. Francisco
STAT1200 – Management Science 1. Decipher when to perform nonparametric tests. parametric test is reliable. For example, Student’s t-test for two
2nd Semester, 2022-2023
independent samples is reliable only if each sample follows a normal
2. Perform Spearman Rank Correlation
distribution and if sample variances are homogeneous.
3. Perform Chi-square Test for Independence
Nonparametric tests do not rely on any distribution. They can thus be
applied even if parametric conditions of validity are not met.

CENTRAL LUZON STATE UNIVERSITY NONPARAMETRIC METHODS | 2 NONPARAMETRIC METHODS | 3

department of department of department of

statistics Advantages of Nonparametric Methods statistics Disadvantages of Nonparametric Methods statistics Assumptions of Nonparametric Statistics
1. They are less sensitive than their parametric counterparts when the assumptions of
1. They can be used to test population parameters when the variable is not the parametric methods are met. Therefore, larger differences are needed before 1. The sample or samples are randomly selected.
normally distributed. the null hypothesis can be rejected. 2. If two or more samples are used, they must be independent of each other
2. They can be used when the data are nominal or ordinal. 2. They tend to use less information than the parametric tests. For example, the sign unless otherwise stated.
test requires the researcher to determine only whether the data values are above or
3. They can be used to test hypotheses that do not involve population
below the median, not how much above or below the median each
parameters. Remarks:
value is.
4. In some cases, the computations are easier than those for the parametric 3. They are less efficient than their parametric counterparts when the assumptions of • If the parametric assumptions can be met, the parametric methods are
counterparts. the parametric methods are met. That is, larger sample sizes are needed to preferred.
5. They are easy to understand. overcome the loss of information. For example, the nonparametric sign test is • When parametric assumptions cannot be met, the nonparametric
6. There are fewer assumptions that have to be met, and the assumptions about 60% as efficient as its parametric counterpart, the z test. Thus, a sample size methods are a valuable tool for analyzing the data.
are easier to verify. of 100 is needed for use of the sign test, compared with a sample size of 60 for use
of the z test to obtain the same results.
NONPARAMETRIC METHODS | 4 NONPARAMETRIC METHODS | 5 NONPARAMETRIC METHODS | 6

department of department of department of

statistics Selection of statistical tools statistics Selection of statistical tools statistics

Parametric Test Nonparametric Test Assessing normality using different statistical graphs/plots
Conditions/ Purposes Parametric Test Nonparametric Test
Normal Distribution Non-normal Distribution Conditions/ Purposes • A normal quantile plot (or normal probability plot) is a graph of points (x,y) where
Normal Distribution Non-normal Distribution each x value is from the original set of sample data, and each y value is the
One sample z-test (if 𝜎 is known)
Compare a mean with
standard value
and Wilcoxon test Compare >2 means of corresponding z score that is a quantile value expected from the standard normal
One sample t-test (if 𝜎 is unknown) Multi-factor ANOVA Friedman test
matched data sets distribution.
Two independent samples z-test
Compare two means of
(if 𝜎1 𝑎𝑛𝑑 𝜎2 are known) Find the relationship Procedure for determining whether it is reasonable to assume that sample data are
and Mann-Whitney test Pearson’s correlation Spearman’s correlation
unpaired data sets between two variables from a normally distributed population:
Two independent samples t-test
(If 𝜎1 𝑎𝑛𝑑 𝜎2 is unknown) Predict the values of 1. Histogram: Construct a histogram. Reject normality if the histogram departs
Simple linear or
Compare two means of one variable from Spearman’s correlation dramatically from a bell shape.
Paired-sample t-test Wilcoxon test nonlinear regression
paired data sets another
2. Outliers: Identify outliers. Reject normality if there is more than one outlier
Compare >2 means of Find the relationship Multiple regression Kendall’s coefficient of present. (Just one outlier could be an error or the result of chance variation, but
One-way ANOVA Kruskal-Wallis test
unmatched data sets among several variables (linear/nonlinear) concordance be careful, because even a single outlier can have a dramatic effect on results.)
NONPARAMETRIC METHODS | 7 NONPARAMETRIC METHODS | 8 NONPARAMETRIC METHODS | 9
 department of department of department of
statistics statistics Example (Normal) statistics Example (Uniform)
3. Normal quantile plot: If the histogram is basically symmetric and there is at most one The first case shows a histogram of IQ scores that is close to being bell-shaped, so the The second case shows a histogram of data having a uniform distribution. The
outlier, use technology to generate a normal quantile plot. Use the following criteria histogram suggests that the IQ scores are from a normal distribution. The corresponding normal quantile plot suggests that the points are not normally
to determine whether or not the distribution is normal. (These criteria can be used corresponding normal quantile plot shows points that are reasonably close to a distributed because the points show a systematic pattern that is not a straight-
loosely for small samples, but they should be used more strictly for large samples.) straight-line pattern, and the points do not show any other systematic pattern that is line pattern. These sample values are not from a population having a normal
Normal Distribution: not a straight line. It is safe to assume that these IQ scores are from a normally
distribution.
The population distribution is normal if the pattern of the points is reasonably close to a straight line distributed population.
and the points do not show some systematic pattern that is not a straight-line pattern.

Not a Normal Distribution:

The population distribution is not normal if either or both of these two conditions applies:
• The points do not lie reasonably close to a straight line.
• The points show some systematic pattern that is not a straight-line pattern. Later in this section
we will describe the actual process of constructing a normal quantile plot, but for now we focus
on interpreting such a plot.
NONPARAMETRIC METHODS | 10 NONPARAMETRIC METHODS | 11 NONPARAMETRIC METHODS | 12

department of department of department of

statistics Spearman Rank Correlation Coefficient statistics Spearman Rank Correlation Coefficient statistics Spearman Rank Correlation Coefficient
Assumptions for Spearman’s Rank Correlation Coefficient
Steps in Performing Spearman’s Rank Correlation Coefficient
• The Spearman rank correlation coefficient is a nonparametric statistic that 1. The sample is a random sample.
Step 1: State the hypotheses.
uses ranks to determine if there is a relationship between two variables. 2. The data consist of two measurements or observations taken on the
Step 2: Find the critical value.
• The computations for the rank correlation coefficient are simpler than same individual.
those for the Pearson coefficient and involve ranking each set of data. Step 3: Find the test value.
Formula for Computing the Spearman Rank Correlation Coefficient a. Rank the values in each data set.
• The difference in ranks is found, and 𝑟𝑠 is computed by using these 6 σ 𝑑2 b. Subtract the rankings for each pair of data values
differences. 𝑟𝑠 = 1 −
𝑛(𝑛2 − 1) c. Square the differences.
• If both sets of data have the same ranks, 𝑟𝑠 will be +1. d. Find the sum of the squares.
• If the sets of data are ranked in exactly the opposite way, 𝑟𝑠 will be -1. Where: 𝑑 = difference in ranks e. Substitute in the formula of 𝑟𝑠
𝑛 = number of data pairs
• If there is no relationship between the rankings, 𝑟𝑠 will be near 0. Step 4: Make the decision.
Decision Rule: Reject Ho if |𝑟𝑠 | ≥ 𝑐𝑟𝑖𝑡𝑖𝑐𝑎𝑙 𝑣𝑎𝑙𝑢𝑒 Step 5: Summarize the results.
NONPARAMETRIC METHODS | 13 NONPARAMETRIC METHODS | 14 NONPARAMETRIC METHODS | 15

department of department of department of

statistics Example 1 statistics Example 1 statistics Example 1
Critical Values for the Rank Correlation Coefficient 4. Computation
Nursing a.) Rank each data set as shown in the table. Let 𝑋1 be the hospitals and
Hospitals Claim: there is a correlation between the two variables
Homes 𝑋2 be the nursing homes.
Find the Spearman rank correlation coefficient
107 230 1. Ho: 𝜌 = 0 (There is no correlation between the number of Hospitals Nursing Homes
for the following data, which represent the Rank of 𝑿𝟏 Rank of 𝑿𝟐 𝒅 = 𝑿𝟏 − 𝑿𝟐 𝒅𝟐
61 134 hospitals and nursing homes) (𝑿𝟏 ) (𝑿𝟐 )
number of hospitals and nursing homes in each
Ha: 𝜌 ≠ 0 (There is correlation between the number of 107 2 230 2 0 0
of seven randomly selected states. At the 0.05 202 704
level of significance, is there enough evidence hospitals and nursing homes) -> claim 61 1 134 1 0 0
133 376
to conclude that there is a correlation between 202 7 704 7 0 0
145 431 2. Test: Spearman’s Rank Correlation Coefficient
the two? 133 5 376 4 1 1
117 538
𝑛 = 7 , 𝛼 = 0.05 , critical value = 0.786 145 6 431 5 1 1
108 3 117 4 538 6 -2 4
3. Decision Rule: Reject Ho if |𝑟𝑠 | ≥ 0.786. σ 𝑑2 = 6
108 3 373 3 0 0
NONPARAMETRIC METHODS | 16 NONPARAMETRIC METHODS | 17 NONPARAMETRIC METHODS | 18
 department of department of department of
statistics Example 1 statistics statistics Example 2
English Math
b.) Substitute in the formula for
The following data shows the final term 56 66
6 σ 𝑑2 6 6 exam scores in English and Math of 10 75 70
𝑟𝑠 = 1 − =1− = 𝟎. 𝟖𝟗𝟐𝟗
𝑛 𝑛2 − 1 7 72 − 1 students. At the 0.01 level of 45 40
5. Decision significance, is there enough evidence to 71 60
Since 0.8929 > critical value (0.786), we reject 𝐻𝑜 . conclude that there is a correlation 62 65
between the two variables? 64 56
6. Conclusion ✓ 58 59
At 5% level of significance, the sample data support the claim that there is
✓ 80 77
correlation between the number of hospitals and nursing homes. 76 67
Claim: ρ ≠ 0 61 63
(see next slide for the wording of final conclusion)
Decision: Reject 𝑯𝒐
NONPARAMETRIC METHODS | 19 INTRODUCTION TO CORRELATION AND LINEAR REGRESSION | 20 NONPARAMETRIC METHODS | 21

department of department of department of

statistics Example 2 statistics Example 2 statistics Example 2
Critical Values for the Rank Correlation Coefficient
Claim: there is a correlation between the two 4. Computation
a.) Rank each data set as shown in the table. Let 𝑋1 be English scores and b.) Substitute in the formula for
variables
𝑋2 be the Math scores. 6 σ 𝑑2 6 54
𝑟𝑠 = 1 − =1− = 𝟎. 𝟔𝟕𝟐𝟕
1. Ho: 𝜌 = 0 (There is no correlation between the English (𝑿𝟏 ) Rank of 𝑿𝟏 Math(𝑿𝟐 ) Rank of 𝑿𝟐 𝒅 = 𝑿𝟏 − 𝑿𝟐 𝒅𝟐 𝑛 𝑛2 − 1 10 102 − 1
number of hospitals and nursing homes) 56 9 66 4 5 25 5. Decision
75 3 70 2 1 1 Since 0.6727 ≥ critical value (0.794), we failed to reject 𝐻𝑜 .
Ha: 𝜌 ≠ 0 (There is correlation between the number 45 10 40 10 0 0
of hospitals and nursing homes) -> claim 71 4 60 7 3 9
62 6 65 5 1 1 6. Conclusion
2. Test: Spearman’s Rank Correlation Coefficient 64 5 56 9 4 16
At 1% level of significance, there is no sufficient sample evidence to support
58 8 59 8 0 0
𝑛 = 10 , 𝛼 = 0.01 , critical value = 0.794 80 1 77 1 0 0 the claim that there is correlation between the English and Math scores of
76 2 67 3 1 1 σ 𝑑2 = 54 students. (see next slide for the wording of final conclusion)
3. Decision Rule: Reject Ho if |𝑟𝑠 | ≥ 0.794. 61 7 63 6 1 1
NONPARAMETRIC METHODS | 22 NONPARAMETRIC METHODS | 23 NONPARAMETRIC METHODS | 24

department of department of department of

statistics statistics Chi-Squared Test of Independence statistics Chi-Squared Test of Independence

Test Statistic Value:

• tests the null hypothesis that the row variable and column variable in a
𝑘
contingency table are not related 𝑂𝑖 − 𝐸𝑖 2
2𝑐 = ෍
𝐸𝑖
𝑖=1
Ho: The row variable and column variable are not related
Where:
Ha: The row variable and column variable are related
(𝑟𝑜𝑤 𝑡𝑜𝑡𝑎𝑙)(𝑐𝑜𝑙𝑢𝑚𝑛 𝑡𝑜𝑡𝑎𝑙)
𝐸𝑖 =
✓ Assumptions:
𝑔𝑟𝑎𝑛𝑑 𝑡𝑜𝑡𝑎𝑙
• The sample data are randomly selected.
• For every cell in the contingency table, the expected frequency is at least 5.
Claim: ρ ≠ 0
Decision: Failed to reject 𝑯𝒐
✓ Decision Rule: Reject Ho if 2𝑐 > 2𝛼,(𝑟−1)(𝑐−1) where r is the number of rows
and c is the number of columns in a contingency table
INTRODUCTION TO CORRELATION AND LINEAR REGRESSION | 25 NONPARAMETRIC METHODS | 26 NONPARAMETRIC METHODS | 27
 department of department of department of
statistics Chi-Squared Test of Independence statistics Example 1 statistics Example 1
Claim: The sex and hand preference are related
Based on the table below, is there evidence to suggest that sex is related to
Steps in Performing Chi-Squared Test of Independence 1. Ho: The sex and hand preference are not related
whether a person is left-handed or right-handed? Test at 0.05 level of
Ha: The sex and hand preference are related -> claim
Step 1: State the hypotheses. significance.
Step 2: Find the critical value. 2. Test: Chi-Squared Test
Step 3: Find the test value. Hand Preference 𝑑𝑓 = 𝑟 − 1 𝑐 − 1 = 2 − 1 2 − 1 = 1 (see slide 27)
Sex Total 𝛼 = 0.05 , critical value = 3.841
a. First, find the expected values for each cell of the contingency table. Left Right
b. Find the test value using the formula of 2𝑐 .
Female 12 108 120
Step 4: Make the decision.
Male 24 156 180
Step 5: Summarize the results.
Total 36 264 300
3. Decision Rule: Reject Ho if 2𝑐 > 3.841.
NONPARAMETRIC METHODS | 28 NONPARAMETRIC METHODS | 29 NONPARAMETRIC METHODS | 30

department of department of
statistics Example 1 statistics Example 1
4. Computation
First, find the expected values for each cell of the contingency table. 5. Decision
(𝑟𝑜𝑤 𝑡𝑜𝑡𝑎𝑙)(𝑐𝑜𝑙𝑢𝑚𝑛 𝑡𝑜𝑡𝑎𝑙)
Hand Preference (j) 𝐸𝑖𝑗 =
𝑔𝑟𝑎𝑛𝑑 𝑡𝑜𝑡𝑎𝑙
Since 0.7576 > critical value (3.841), we failed to reject 𝐻𝑜 .
Sex (i) Total
Left (1) Right (2)
(120)(36) (120)(264)
𝐸11 = = 14.4 𝐸12 = = 105.6 6. Conclusion
Female (1) 12 108 120 300 300
Male (2) 24 156 180 At 5% level of significance, we can conclude that the sex and the hand
(180)(36) (180)(264)
Total 36 264 300 𝐸21 = = 21.6 𝐸22 = = 158.4 preference are not related.
300 300

𝑘
𝑂𝑖 − 𝐸𝑖 2 12 − 14.4 2 108 − 105.6 2 24 − 21.6 2 156 − 158.4 2
2𝑐 = ෍ = + + + = 𝟎. 𝟕𝟓𝟕𝟔 Ho: The sex and hand preference are not related
𝐸𝑖 14.4 105.6 21.6 158.4
𝑖=1 Ha: The sex and hand preference are related

NONPARAMETRIC METHODS | 31 NONPARAMETRIC METHODS | 32