Chi Squared

Question 1

What type of statistical test is chi squared?

Answer 1

An inferential test - allows us to say something about a population from looking at a sample

Question 2

What data does it deal with?

Answer 2

Nominal (categorised) data

Question 3

What does chi squared test for a difference between

Answer 3

Expected frequencies and observed frequencies in one or more categories - tests if the observed is statistically significant

Question 4

Is chi squared a Non parametric or parametric test?

Answer 4

Non parametric - data doesn’t have to be perfect

Question 5

What is the parametric equivalent of the chi squared test.

Answer 5

T-test

Question 6

What are the chi squared assumptions

Answer 6

• must have nominal data
• total number of observations >20
• cannot have 20% or more categories with calculated expected frequencies <5
• no category can have an expected value of <1
• observations must have frequencies not percentages (%)

Question 7

What is the chi squared test procedure?

Answer 7

• Set up hypothesis of H0 and H1
• Determine acceptable confidence level (for 95% confidence, significance must be <0.05
• calculate the chi squared test statistic
• determine the number of degrees of freedom
• determine the significance level

Question 8

The equation of the test: what is xi and ei?

Answer 8

xi = observed frequencies 
ei = expected frequencies

Question 9

Is the observed and expected frequencies are the same then what is the value of chi squared?

Answer 9

Zero (0)

Question 10

As we increase n, the statistical power of the test…

Answer 10

Increases.

Question 11

When the sample size is raised to 100…

Answer 11

The test power increases considerably

Large samples allow very small changes to be detected

Question 12

Any sample bigger than ________ is too big (looses power)

Answer 12

> 300

You gain statistical significance just due to the large sample size