Choosing a statistical test

Question 1

What is a statistical test?

Answer 1

A test used to determine whether a significant difference or correlation exists

Question 2

Recall the test table

Answer 2

Nominal data:Chi-squared - Sign test - Chi-squared
Ordinal data:Mann-whitney - Wilcoxon - Spearmans rho
Interval data:Unrelated t-test - Related t-test - Pearson r

Question 3

What are the headings of the test table?

Answer 3

Test of difference

Test of association or relationship

Question 4

What are the sub-headings?

Answer 4

Unrelated design

Related design

Question 5

When deciding which statistical test, what is the first factor to know?

Answer 5

Is the researcher looking for a correlation or a difference?

Question 6

When deciding which statistical test, what is the second factor to know?

Answer 6

What experimental design are they using?
Independent groups
Repeated measures
Matched pairs design

Question 7

Which designs come under which subheadings?

Answer 7

Unrelated design:
Independent groups 
Related design:
Repeated measures 
Matched pairs

Question 8

When deciding which statistical test, what is the third factor to know?

Answer 8

What level of measurement is being used?
Nominal - the answer is categorised
Ordinal - eg: rating something 1-10 (intervals are not equal between each unit)
Interval - eg: how quick is your reaction time? (intervals are equal between each unit)

Question 9

What is the rule of ‘r’?

Answer 9

When determining the relationship between the calculated value and the critical value, if the the test has an ‘r’ within the name, it means the calculated value needs to be equal to or higher than the critical value for it to be significant

Question 10

What is a type 1 error?

Answer 10

Occurs when the researcher claims they have discovered a significant difference/association when have not ( a false positive)
Wrongly rejecting a null hypothesis

Question 11

What is a type 2 error?

Answer 11

Occurs when the researcher claims the difference/ association found is not significant when it actually is ( a false negative)
Wrongly accepting a null hypothesis

Question 12

How can you decrease the likelihood of these errors occurring?

Answer 12

Type 1 error increases when the level of significance is too lenient
Type 2 error increases if the level of significance is too strict
Therefore setting the level of significance at 0.05 is ideal as it decreases the likelihood of these errors occurring