IBU Analysis of

International Burch University

Categorical Data
 Introduction

◼ Nonparametric Statistics
◼ Fewer restrictive assumptions about data
levels and underlying probability distributions
◼ Population distributions may be skewed
◼ The level of data measurement may only be
ordinal or nominal
 Goodness-of-Fit Tests:
Specified Probabilities

◼ Does sample data conform to a hypothesized

distribution?
◼ Examples:

◼ Do sample results conform to specified expected

probabilities?
◼ Are technical support calls equal across all days of
the week? (i.e., do calls follow a uniform
distribution?)
◼ Do measurements from a production process
follow a normal distribution?
 Chi-Square Goodness-of-Fit Test
(continued)
◼ Are technical support calls equal across all days of the
week? (i.e., do calls follow a uniform distribution?)
◼ Sample data for 10 days per day of week:

Sum of calls for this day:

Monday 290
Tuesday 250
Wednesday 238
Thursday 257
Friday 265
Saturday 230
Sunday 192

 = 1722
 Logic of Goodness-of-Fit Test

◼ If calls are uniformly distributed, the 1722 calls

would be expected to be equally divided across
the 7 days:
1722
= 246 expected calls per day if uniform
7
◼ Chi-Square Goodness-of-Fit Test: test to see if
the sample results are consistent with the
expected results
 Observed vs. Expected
Frequencies
Observed Expected
Oi Ei
Monday 290 246
Tuesday 250 246
Wednesday 238 246
Thursday 257 246
Friday 265 246
Saturday 230 246
Sunday 192 246
TOTAL 1722 1722
 Chi-Square Test Statistic
H0: The distribution of calls is uniform
over days of the week
H1: The distribution of calls is not uniform

◼ The test statistic is

K
(O − E ) 2
2 =  i i
(where d.f. = K − 1)
i=1 Ei
where:
K = number of categories
Oi = observed frequency for category i
Ei = expected frequency for category i
 The Rejection Region
H0: The distribution of calls is uniform
over days of the week
H1: The distribution of calls is not uniform

K
(O − E ) 2
2 =  i i

i=1 Ei

◼ Reject H0 if  
2 2
α


(with k – 1 degrees
of freedom) 0 2
Do not Reject H0
reject H0 2 
 Observed vs. Expected
Frequencies
Observed Expected
Oi Ei (Oi – Ei) (Oi – Ei)2 (Oi – Ei)2/Ei
Monday 290 246 44 1936 7.870
Tuesday 250 246 4 16 0.065
Wednesday 238 246 -8 64 0.260
Thursday 257 246 11 121 0.492
Friday 265 246 19 361 1.467
Saturday 230 246 -16 256 1.041
Sunday 192 246 -54 2916 11.854

TOTAL 1722 1722 2 = 23.049

 Chi-Square Test Statistic
H0: The distribution of calls is uniform
over days of the week
H1: The distribution of calls is not uniform

(290 − 246) 2
(250 − 246) 2
(192 − 246) 2
2 = + + ... + = 23.049
246 246 246
K – 1 = 6 (7 days of the week) so
use 6 degrees of freedom:
2.05 = 12.5916
 = .05
Conclusion:
2 = 23.05 > 2 = 12.5916 so
reject H0 and conclude that the 0 2
Do not Reject H0
distribution is not uniform reject H0

2.05 = 12.5916
 Goodness-of-Fit Tests:
Population Parameters Unknown

Idea:
◼ Test whether data follow a specified distribution

(such as binomial, Poisson, or normal) . . .

◼ . . . without assuming the parameters of the

distribution are known

◼ Use sample data to estimate the unknown

population parameters
 Goodness-of-Fit Tests:
Population Parameters Unknown
(continued)
◼ Suppose that a null hypothesis specifies category
probabilities that depend on the estimation (from the
data) of m unknown population parameters
◼ The appropriate goodness-of-fit test is the same as in
the previously section . . .
K
(Oi − Ei )2
 =
2

i=1 Ei
◼ . . . except that the number of degrees of freedom for
the chi-square random variable is
Degrees of Freedom = (K − m − 1)
◼ Where K is the number of categories
 Test of Normality

◼ The assumption that data follow a normal

distribution is common in statistics

◼ Evidence of normality was assessed in prior

chapters
(for example, with normal probability plots in
Chapter 5)
◼ Here, a chi-square test is developed
 Test of Normality
(continued)

◼ Two population parameters can be estimated using

sample data:
n

 i
(x − x) 3

Skewness = i−1
ns3
n

 i
(x − x) 4

Kurtosis = i−1
ns 4
◼ For a normal distribution,
Skewness = 0
Kurtosis = 3
 Jarque-Bera
Test for Normality
◼ Consider the null hypothesis that the population
distribution is normal
◼ The Jarque-Bera Test for Normality is based on the closeness the
sample skewness to 0 and the sample kurtosis to 3
◼ The test statistic is

 (Skewness) 2 (Kurtosis − 3)2 

JB = n + 
 6 24 
◼ as the number of sample observations becomes very large, this
statistic has a chi-square distribution with 2 degrees of freedom
◼ The null hypothesis is rejected for large values of the test statistic
 Jarque-Bera
Test for Normality
(continued)
◼ The chi-square approximation is close only for very
large sample sizes
◼ The test statistic is compared to significance points from
text Table 14.9
Sample 10% 5% Sample 10% 5%
size n point point size n point point
20 2.13 3.26 200 3.48 4.43
30 2.49 3.71 250 3.54 4.61
40 2.70 3.99 300 3.68 4.60
50 2.90 4.26 400 3.76 4.74
75 3.09 4.27 500 3.91 4.82
100 3.14 4.29 800 4.32 5.46
125 3.31 4.34 ∞ 4.61 5.99
150 3.43 4.39
 Example: Jarque-Bera
Test for Normality
◼ The average daily temperature has been recorded for
200 randomly selected days, with sample skewness
0.232 and kurtosis 3.319
◼ Test the null hypothesis that the true distribution is
normal
 (Skewness) 2 (Kurtosis − 3)2   (0.232)2 (3.319 − 3)2 
JB = n +  = 200  +  = 2.642
 6 24   6 24 

◼ From Table 14.9 the 10% critical value for n = 200 is

3.48, so there is not sufficient evidence to reject that the
population is normal
 Contingency Tables

Contingency Tables
◼ Used to classify sample observations according
to a pair of attributes
◼ Also called a cross-classification or cross-
tabulation table
◼ Assume r categories for attribute A and c
categories for attribute B
◼ Then there are (r x c) possible cross-classifications
 r x c Contingency Table

Attribute B

Attribute A 1 2 ... c Totals

1 O11 O12 … O1c R1

2 O21 O22 … O2c R2
. . . … . .
. . . … . .
. . . … . .
r Or1 Or2 … Orc Rr
Totals C1 C2 … Cc n
 Test for Association
◼ Consider n observations tabulated in an r x c
contingency table
◼ Denote by Oij the number of observations in
the cell that is in the ith row and the jth column
◼ The null hypothesis is
H0 : No association exists
between the two attributes in the population
◼ The appropriate test is a chi-square test with
(r-1)(c-1) degrees of freedom
 Test for Association
(continued)
◼ Let Ri and Cj be the row and column totals
◼ The expected number of observations in cell row i and
column j, given that H0 is true, is

R iC j
Eij =
n
◼ A test of association at a significance level  is based
on the chi-square distribution and the following decision
rule
r c (Oij − Eij )2
Reject H0 if χ 2 =   χ (r2 −1)(c −1),α
i=1 j=1 Eij
 Contingency Table Example

Left-Handed vs. Gender

▪ Dominant Hand: Left vs. Right
▪ Gender: Male vs. Female

H0: There is no association between

hand preference and gender
H1: Hand preference is not independent of gender
 Contingency Table Example
(continued)

Sample results organized in a contingency table:

Hand Preference
sample size = n = 300:
Gender Left Right
120 Females, 12
were left handed Female 12 108 120
180 Males, 24 were
left handed Male 24 156 180

36 264 300
 Logic of the Test
H0: There is no association between
hand preference and gender
H1: Hand preference is not independent of gender

◼ If H0 is true, then the proportion of left-handed females

should be the same as the proportion of left-handed
males
◼ The two proportions above should be the same as the
proportion of left-handed people overall
 Finding Expected Frequencies
120 Females, 12 Overall:
were left handed
180 Males, 24 were P(Left Handed)
left handed = 36/300 = .12
If no association, then
P(Left Handed | Female) = P(Left Handed | Male) = .12

So we would expect 12% of the 120 females and 12% of the 180
males to be left handed…

i.e., we would expect (120)(.12) = 14.4 females to be left handed

(180)(.12) = 21.6 males to be left handed
 Expected Cell Frequencies
(continued)

◼ Expected cell frequencies:

th th
R iC j
(i Row total)(j Column total)
Eij = =
n Total sample size

Example:
(120)(36)
E11 = = 14.4
300
 Observed vs. Expected
Frequencies

Observed frequencies vs. expected frequencies:

Hand Preference
Gender Left Right
Observed = 12 Observed = 108
Female 120
Expected = 14.4 Expected = 105.6
Observed = 24 Observed = 156
Male 180
Expected = 21.6 Expected = 158.4

36 264 300
 The Chi-Square Test Statistic

The Chi-square test statistic is:

r c (Oij − Eij )2
 = 
2
with d.f . = (r − 1)(c − 1)
i=1 j=1 Eij

◼ where:
Oij = observed frequency in cell (i, j)
Eij = expected frequency in cell (i, j)
r = number of rows
c = number of columns
 Observed vs. Expected
Frequencies
Hand Preference
Gender Left Right
Observed = 12 Observed = 108
Female 120
Expected = 14.4 Expected = 105.6
Observed = 24 Observed = 156
Male 180
Expected = 21.6 Expected = 158.4

36 264 300

(12 − 14.4)2 (108 − 105.6)2 (24 − 21.6)2 (156 − 158.4)2

 =
2
+ + + = 0.7576
14.4 105.6 21.6 158.4
 Contingency Analysis

 2 = 0.7576 with d.f. = (r - 1)(c - 1) = (1)(1) = 1

Decision Rule:
If 2 > 3.841, reject H0,
otherwise, do not reject H0
Here, 2 = 0.7576
 = 0.05 < 3.841, so we
do not reject H0
2.05 = 3.841  and conclude that
gender and hand
Do not reject H0 Reject H0 preference are not
associated