Mcnemar Test: Dr. Saadah Mohamed Akhir Unit Rundingcara Statistik FSK, Ukm
Mcnemar Test: Dr. Saadah Mohamed Akhir Unit Rundingcara Statistik FSK, Ukm
Mcnemar Test: Dr. Saadah Mohamed Akhir Unit Rundingcara Statistik FSK, Ukm
Test
Dr. Saadah Mohamed Akhir
Unit Rundingcara Statistik
FSK, UKM
Learning Outcomes
At
the end of the lecture, students will be
able to
Understand when to use McNemar test.
Perform McNemar test using SPSS.
Calculate the McNemar test using formula.
Interpret and report the output from SPSS.
What is McNemar test?
Non parametric test
Categorical data
a dichotomous (binary) dependent
variable between two related groups.
paired nominal data
toanalyze pretest-posttest (before and after)
study designs
Wilcoxon Sign Ranked Test and
McNemar Test – Any difference?
They do not share the same type of data.
Wilcoxon – continuous data
McNemar – categorical data
Assumptions
Has one categorical dependent variable
with two categories (dichotomous variable)
and one categorical independent variable
with two related groups.
Dichotomous variable (Binary data) – eg.
Yes and No, Normal and Abnormal.
Participants are a random sample from the
population.
Example
Categorical
dependent:
With Training Total
Knowledge status
Knowledge Knowledge
Improved not yes – improved
Improved
Without Knowledge 10 5 15
no – not improved
Two related group
Training Improved
Knowledge
not
Improved
15 20 35
– training status
Total 25 25 50 yes – with
no - without
Contingency Table
Organizationof data – table type form
Variables composed in columns and rows
Known as cross tabulation table.
With Training Total
Knowledge Knowledge
Improved not
Improved
Without Knowledge 10 5 15
Training Improved
Knowledge 15 20 35
not
Improved
Total 25 25 50
Case Study:
Bilirubin level from experimental drug
- A study was conducted to evaluate the
toxicity side effect of an experimental
therapy.
- Patients (n=86) were treated with the
experimental drug for 3 months.
- Clinical lab measured bilirubin levels of
each patient at baseline and 3 months
after drug is given.
Data - Variables
Dependent variable – bilirubin level
Normallevel of bilirubin
Abnormally high level of bilirubin
Normal 60 14
Abnormal 6 6
McNemar with SPSS
Raw Data
McNemar - SPSS
Enterraw data
Analyze – Non Parametric – Related
Samples
objective – customize analysis
Field – move pre test and post test to test
field box
Click Run
Raw data
Analyze – Non Parametric –
Related Samples
objective – customize analysis
Field – move pre test and post test to
test field box
SPSS - output
McNemar with SPSS
Raw data
SPSS (McNemar)
Enter raw data in SPSS
Analyze Descriptive statistics
Crosstabs
Move over the two variables to the row
and column boxes (I used rows for the
pre-test and columns for the post-test)
Click on Statistics and check McNemar,
then click Continue.
Click OK.
Raw data
Analyze Descriptive statistics
Crosstabs
SPSS
Output
McNemar with SPSS
2 x 2 contingency table
SPSS (McNemar)
Output from raw data – contingency table
Weight cases by frequency
Analyze Descriptive statistics Crosstabs
Move over the two variables to the row and
column boxes (I used rows for the pre-test and
columns for the post-test – it can also be the other
way round)
Click on Statistics and check McNemar, then click
Continue.
Click on Cells and then check Row and Column
under Percentages, then click Continue.
Click OK.
What should be in SPSS?
Since there are 4 cells in 2 x 2 contingency table –
there are only 4 data entry.
How many variables? There are 3 variables
Before
After
Frequency
After
Before
Normal Abnormal
Normal 60 14
Abnormal 6 6
Variable View – Data View
Create 3 columns and label
Hypothesis
null hypothesis : H 0 : p1 p2
alt. hypotehsis : H A : p1 p2
Null hypothesis: There is no change in bilirubin
level after taking experimental drug
Responders A B
Non-responder C D
38
McNemar’s test
null hypothesis : H 0 : p1 p2
alt. hypotehsis : H A : p1 p2
( B C )2
test statistic :
2
BC
decision rule : reject H 0 if 2 12 ( )
Responders A B
Non-responder C D
Formula – Calculate
(McNemar)
After
Normal Abnormal High(Y) null hypothesis : H 0 : p1 p2
(N) alt. hypotehsis : H A : p1 p2
Before
(B C )2
test statistic :
2
BC
Normal 60 (A) 14 (B) decision rule : reject H 0 if 2 12 ( )
Abnormally
High
6 (C) 6 (D)
- df = 1, the critical value =
3.84 (from the chi-square
- df = (r - 1) (c – 1) = 1
table)
- r = 2, c = 2
- calculated value of 3.20
does not exceeds this value
14 −6 2
- 2 = - there is no significant
14+6 difference in bilirubin level
after therapy
= 3.20
Chi Square table
Report (APA style)
Eighty six participants were recruited to take
part in the toxicity side effect of an
experimental drug for 3 months. A bilirubin
levels were measured to identify the toxicity
side effect of the experimental drug. Of these,
only 6 had normal bilirubin level after taking
the experimental drug for 3 months. A
McNemar test showed that there is no
significant change of bilirubin level after
taking the experimental drug, 2 (1, N = 86) =
3.20, p = 0.115.
Exercise
THANK YOU