Chi test

Greg Martin’s statistics cheat sheet Data:

T test
Observed difference Two categorical variables e.g. Observed difference
For more statistics teaching
Sex (male / female)
go to www.learnmore365.com
Height category (short, medium height, and tall)
Data: M
A numeric variable (like weight) Observa,on from the data:
Weight

Sex
A categorical variable (like sex) As we look at the observa8ons in the different height categories, we see
different propor8ons of men and women. We would also see the
Observa,on from the data: F associa8on if we looked at the sex groups separately: the propor8on of
The average weight for males is more than that for females people who were short, medium and tall would be different in the two
Weight groups. S M T
Research ques,on:
Is this difference due to chance or is it a sta8s8cally significant Null hypothesis Research ques,on: Null hypothesis
difference? Is this associa8on due to chance or is it sta8s8cally significant?

Start with a null hypothesis: M Start with a null hypothesis:

Sex
Perform a Student’s T test Perform a Chi test
If the p value is less than 0.05 then you can reject the null hypothesis F If the p value is less than 0.05 then you can reject the null hypothesis
and conclude that the difference between means (averages) in the two and conclude that the associa8on between Sex and Height Category is
categories is sta8s8cally significant Weight sta8s8cally significant S M T

Data:
A numeric variable (like weight)
ANOVA Observed difference
Data: Correlation test Observed relationship
A categorical variable with 3 or more categories (like na8onality: Two numeric variables like Weight and Age
American; Bri8sh; Russian)
A
Observa,on from the data:
Observa,on from the data:

Weight
There is an associa8on (correla8on) between the two variables: as the
The average weight is not the same for each na8onality B
value of one changes, the value of the other seems to change with it
(either up or down).
Research ques,on: R
Is this difference due to chance or is it a sta8s8cally significant Age
Weight Research ques,on:
difference? Is the associa8on / correla8on between the two variables a chance
finding or is it sta8s8cally significant? Null hypothesis
Start with a null hypothesis: Null hypothesis
The average weight in each group is the same Start with a null hypothesis:
There is no associa8on between the two variables.
Perform an ANOVA A
If the p value is less than 0.05 then you can reject the null hypothesis Perform a Correla,on test

Weight
and conclude that the difference between means (averages) in the B
If the p value is less than 0.05 then you can reject the null hypothesis
three categories is sta8s8cally significant. Note that this test doesn’t tell and conclude that the correla8on between the two variables is
you which of the three is different. It just tells you that they are not all R
sta8s8cally significant. This test will also give you the correla8on
the same. coefficient (that tells you about the strength of the associa8on) Age
Weight
Learn More 365
Share this cheat sheet with others

Course on:
Research methods
Statistics
Epidemiology
Public health
R Programming
... and more

Get access to the Resource Library by

Statistics and Research Methods

Or scan this QR code to go to