Copyright (c) 2004 Brooks/Cole, a division of Thomson Learning, Inc. Chapter 14 Goodness-of-Fit Tests and Categorical Data Analysis.

Chapter 14

Goodness-of-Fit Tests and

Categorical Data Analysis

14.1

Goodness-of-fit Tests When Category

Probabilities Are Completely Specified

Terminology

A binomial experiment consists of a sequence of independent trials in which each trial can result in one of two possible outcomes. A multinomial experiment generalizes a binomial experiment by allowing each trial to result in one of k outcomes, where k is an integer greater than 2.

Multinomial Experiment

The expected number of trials resulting in category i is E(Ni) = npi. When H0:p1 = p10,…,pk = pk0 is true, these expected values become E(N1) = np10,…, E(Nk) = npk0.

Recall: Chi-squared Critical Value

Let , called a chi-squared critical value, denote the number of the measurement axis such that of the area under the chi-squared curve with v df lies to the right of 2

, .v

2,v

Notation Illustrated2,v

2,v

shaded area =

2 pdfv

Multinomial Experiment

Provided that npi 5 for every i, the random variable

2 22

1 all cells

( ) (observed expected)

e xpected

ki i

ii

N np

np

has approximately a chi-squared distribution with k – 1 df.

Test With Significance Level

Null hypothesis H0: p1 = p10,…,pk = pk0

Alternative hypoth Ha: at least one 0.i ip p2

2 0

01

( )ki i

ii

n np

np

Test statistic value:

Rejection region: 2 2, 1k

P-Values for Chi-Squared Tests

The P-value for an upper-tailed chi-squared test is the area under the curve to the right of the calculated

2.

2v

When the pi’s Are Functions of Other Parameters

2

Frequently the pi’s are hypothesized to depend on a smaller number of parameters (m < k). Then a specific hypothesis involving the yields specific pi0’s, which are then used in the test.

2

'si1,..., m

When the Underlying Distribution is Continuous

2

1

0 1 0( ) ( )i

i

a

i i ia

p P a X a f x dx

Let X denote the variable being sampled. The hypothesized pdf is f0(x). Subdivide the measurement scale of X into k intervals [a0, a1),…, [ak, ak-1). The cell properties specified by H0 are

14.2

Goodness of Fit for

Composite Hypotheses

When Parameters Are Estimated

2

The null hypothesis states that each pi is a function of a small number of parameters with the otherwise unspecified.

'si1,..., m

0 1 1: ( ),... ( )θ θi iH p p 1( ,..., )θ= m Ha: the hypothesis is not true

11 1 1( ,..., ) ... knn

k k kP N n N n p p

For general k, the joint distribution of N1,…,Nk is the multinomial distribution with

Joint Distribution

11 1 1( ,..., ) ( ) ... ( )θ θ kn n

k k kP N n N n

when H0 is true this becomes

Method of Estimation

Let n1,…,nk denote the observed values of N1,…,Nk. Then are those values of the that maximize

1̂ˆ,..., m

'si 1

1 1 1( ,..., ) ( ) ... ( )θ θ kn nk k kP N n N n

1̂ˆ,..., m are the maximum likelihood

estimators of 1,..., .m

TheoremUnder general “regularity” conditions on and the ’s if are estimated by the method of maximum likelihood as described previously and n is large,

1,..., m 1,..., m ( )θi

22

1

ˆ[ ( )]ˆ( )

θ

θ

ki i

i i

N n

n

has approximately a chi-squared distribution with k – 1 – m df when H0 is true.

Level Test

An approximate level test of H0 versus Ha is then to reject H0 if . In practice, the test can be used if

2 2, 1k m

ˆ( ) 5θin for every i.

Degrees of Freedom

2 number of freely number of independent df

determined cell counts parameters estimated

A general rule of thumb for degrees of freedom in a chi-squared test is

Test Procedure

2 2 2, 1 , 1If withhold judgement.k m k

2 2, 1 0If , reject .k H

2 2, 1 0If , do not reject .k m H

Let be the maximum likelihood estimators of based on the full sample X1,…,Xn and let denote the statistic based on these estimators. Then the critical value that specifies a level upper-tailed test satisfies

1,..., m

Goodness of Fit for Discrete Distributions

1̂ˆ,..., m

2

c

2 2, 1 , 1k m kc

Goodness of Fit for Continuous Distributions

The chi-squared test can be used to test whether the sample comes from a specified family of continuous distributions. Once the cells are chosen (independent of the observations) it is usually difficult to estimate unspecified parameters from the observed cell counts, so mle’s based on the full sample are computed.

Special Test for Normality

H0: the population distribution is normal

Ha: the pop. distribution is not normalversus

consists of rejecting H0 when .r c

1Let [( 3.75) /( .25)]iy i n andCompute r for the pairs (x(1),y1),…,(x(n),yn). The Ryan-Joiner test of

14.3

Two-Way Contingency Tables

Data With Counts or Frequencies

1. There are I populations of interest, each corresponding to a different row of the table, and each population is divided into the same J categories. A sample is taken from the ith population, and the counts are entered in the cells in the ith row of the table.

Data With Counts or Frequencies

2. There is a single population of interest, with each individual in the population categorized with respect to two different factors. There are I categories associated with the first factor and J categories associated with the second factor.

n11 n12 n1j n1J

n21

ni1 nij

nI1 nIJ

Two-Way Contingency Table

Estimated Expected Counts Under H0

(Homogeneity)

.

ˆ estimated expected count in cell ( , )

th row total th column total

ij

ji

e i j

nn

ni j

n

Null hypothesis: H0: p1j = p2j =…= pIj

Alternative hypoth.: Ha: H0 is not true2

2

1 1

ˆ( )

ˆ

I Jij ij

iji j

n e

e

Test statistic value:

Rejection region: 2 2,( 1)( 1)I J

Test for Homogeneity

Apply as long as ˆ 5 for all cells.ije

Multinomial Experiment

Provided that npi 5 for every i, the random variable

2 22

1 all cells

( ) (observed expected)

e xpected

ki i

ii

N np

np

has approximately a chi-squared distribution with k – 1 df.

Estimated Expected Counts (Independence)

ˆ estimated expected count in cell ( , )

ˆ ˆ

th row total th column total

ij

j i jii j

e i j

n n nnn p p n

n n ni j

n

Null hypothesis: H0:

Alternative hypoth.: Ha: H0 is not true2

2

1 1

ˆ( )

ˆ

I Jij ij

iji j

n e

e

Test statistic value:

Rejection region: 2 2,( 1)( 1)I J

Test for Independence

Apply as long as ˆ 5 for all cells.ije

ij i jp p p