Testing Proportions - Week 9 Pt 2

Question 1

When can chi squared be used?

Answer 1

When one nominal variable with 2+ categories present
(Compare observed freq and expected freq)

Question 2

Hypothesis in one nominal variables

Answer 2

H0= No inconsistency between observed/expected frequency (same distribution)
H1= Inconsistency between observed/expected frequency (don’t follow same distribution)

Question 3

What are proportions measured through?

Answer 3

Effect Sizes

Question 4

How can sample data be analysed to reflect preference in populations?

Answer 4

Through CIs

Question 5

DF for proportion significant testing

Answer 5

Number of categories - 1

Question 6

How can CIs be used for significance?

Answer 6

If overlap by 50% = non-significant

Question 7

Hypothesis for chi squared using two nominal variables

Answer 7

H0= relative proportion of 1 variable are independent to second variable
H1= Relative proportions of 1 variable aren’t independent of second variable

Question 8

What is a p-value of 0.00 reported as?

Answer 8

p < .001

Question 9

When are chi-squared tests unreliable?

Answer 9

When any expected frequency is <5
(Especially if num of categories is small)

Question 10

How can we test proportions and statistical power?

Answer 10

Sample size and effect size = positive relationship
Large bias easier to detect than small bias
Larger N easier to detect than small N