You are on page 1of 10

Chi Square Test

Gitanjali Batmanabane

Gitanjali 1
At the end of this session you will be
able to:

Prepare a contingency table


Realise which study designs are suitable
for applying the chi square test
Understand the assumptions / limitations
of the chi square test.

tanjali 2
Know
thyself Why does
he keep
saying this
all the time?

tanjali 3
Excuse me sir, you
No, my son,
but I say “know yourself”
understand all the time but do
something YOU KNOW
about this YOURSELF?
“NOT
KNOWING”

tanjali 4
What is it?

Test of proportions
Non parametric test
Dichotomous variables are used
Tests the association between two
factors
e.g. treatment and disease
gender and mortality

tanjali 5
Associations and Causal Associations
Relationship
between variables

Not statistically Statistically


associated associated

Non-causal Causal

Indirectly Directly
causal causal
tanjali 6
Contingency (2X2) table
Outcome
Exposure
Yes No

Yes

No

 Enter number of subjects – not percentages,


ratios, averages etc.,

tanjali 7
 Each subject can be entered only once
Out of 25 women who had uterine cancer, 20
claimed to have used estrogens. Out of 30
women without uterine cancer 5 claimed to have
used estrogens.

Outcome (cancer)
Exposure Total
(estrogen) Yes No

Yes
No

tanjali 8
Total
Out of 25 women who had uterine cancer, 20
claimed to have used estrogens. Out of 30
women without uterine cancer 5 claimed to have
used estrogens.

Outcome
Exposure Total
Yes No

Yes 20 5 25

No 5 25 30

tanjali 9 Total 25 30 55
Assumptions / Limitations

Data is from a random sample.


A sufficiently large sample size is required (at
least 20)
Actual count data (not percentages)
Adequate cell sizes should be present. (>5 in
all cells- if less number present apply Yates
correction)
Observations must be independent.
Does not prove causality.
tanjali 10

You might also like