Statistical testing

Question 1

What is the purpose of statistical testing?

Answer 1

• To determine whether a significant difference or correlation exists, and whether the null hypothesis should be accepted or rejected

Question 2

What are the 3 factors that should be considered when deciding which test to use?

Answer 2

1) Difference (/correlation)
2) Design
3) Data

Question 3

What are the 3 levels of measurement?

Answer 3

1) Nominal (category data- shows preference)
2) Ordinal (ordered with an uneven scale- ratings)
3) Interval (ordered with standardised scale)

Question 4

What are parametric tests and when are they used?

Answer 4

• Powerful and robust tests- able to detect significance
• Used when: data is interval, population has a normal distribution, homogenity of variance (similar standard deviation)

Question 4

What are non-parametric tests and when are they used?

Answer 4

• Less powerful- used when parametric conditions are not used
• Used when: data is nominal/ ordinal, population has a skewed distribution

Question 5

What test would I use if…?
- difference
- nominal
- unrelated

Answer 5

Chi-squared

Question 6

What test would I use if…?
- difference
- nominal
- related

Answer 6

Sign test

Question 7

What test would I use if…?
- difference
- ordinal
- unrelated

Answer 7

• Mann-Whitney U

Question 8

What test would I use if…?
- difference
- ordinal
- related

Answer 8

• Wilcoxon

Question 9

What test would I use if…?
- difference
- interval
- unrelated

Answer 9

• Unrelated t-test

Question 10

What test would I use if…?
- difference
- interval
- related

Answer 10

• Related t-test

Question 11

What test would I use if…?
- correlation
- nominal

Answer 11

• Chi-squared

Question 12

What test would I use if…?
- correlation
- interval

Answer 12

• Pearson’s r

Question 12

What test would I use if…?
- correlation
- ordinal

Answer 12

• Spearman’s rho

Question 13

What are type 1 errors?

Answer 13

• Null hypothesis has been rejected
• But null hypothesis is TRUE
• False positive

Question 14

What is a type 2 error?

Answer 14

• Null hypothesis has been accepted
• But null hypothesis is FALSE
• False negative

Question 15

When is a type 1 error more likely to occur?

Answer 15

• When the significance level is too lenient (too high), such as 0.1 (10%)

Question 16

When is a type 2 error more likely to occur?

Answer 16

• When the significance level is too stringent (too low), e.g: 0.01 (1%)