The Mcnemar Change Test
The Mcnemar Change Test
by Nimal Baroya
The McNemar Change Test
FUNCTION
For The significance of change is particularly
to before and after designs in which each
subject is used as its own control
The measurement are made on either
nominal or ordinal scale
It might be used to test the effectiveness of
particular treatment, ex : injection drug,
training, separation for family etc.
The McNemar Change Test
Rationale and Method
A fourfold table of
after
frequencies is used
to represent the first - +
and second sets of + A B
responses from the before
same individual
- C D
A denotes number of individuals Rationale and
whose responses were (+) on the
first measure and (-) on the Method
second measure
D is the number individuals who
changed from (-) to (+) after
B is the frequency of individual - +
who responded the same (+)
C is the number of individuals + A B
before
who responded the same (-) both
before and after
- C D
A+D is the total number of people
whose responses changed
Rationale and Method
The null hypothesis is that the number of
change in each direction is equally like
Ho is true, the expected frequency in each of
the two cells is (A+D)/2
The significances of changes follow the
sampling distribution X2 with df = 1
The McNemar Test for the Significances of Changes
2
X = ( IA DI 1)2 with df = 1
A+D
Rejection Region,
Reject Ho if, X2(hitung) > X2(tabel) (; df =1)
Small expected frequencies
If the expected frequency for the McNemar
test, (A+D)/2 , is very small less than 5-, the
binomial test should be used rather than the
McNemar test
To use binomial test, let N = A+D and x be
the smaller of the two observed frequencies,
either A or D, and use table D (binomial test)
to the test significance of x
Reject H0 if p value <
Summary Procedure (Steps in the computation of the
McNemar change test)
- 105 15
Exercise 1