choosing a statistical test

Question 1

choosing a test

Answer 1

Choosing a statistical test relies on several pieces of information.
Most of these will be given in the question:
• Test of difference or a correlation
• Test design
• Levels of measurement

Two of these should be a familiar to you.
• Test of difference or a correlation
• Should be obvious from the wording of the hypothesis / question.
• Association or relationship = correlation.
• Test design = experimental design.
• Unrelated
→ Independent groups
Related
→ Repeated measure or matched pairs.

Question 2

levels of measurement

Answer 2

Nominal
• Categorical data
• Favourite desert
• Smoker / non-smoker
• Sex
• Discrete data → only appears in 1 category.

Ordinal
• Data that can be ordered in some way.
• E.g. Rating scales.
• Subjective
• No equal intervals between values. (eg people finishing first and second in race-distance between?)

Interval
• Data based on numerical scales.
• E.g. weight, size, scores, speed.
• Objective
• Equal intervals between values.
• Most sophisticated form.

Question 3

choosing a test

Answer 3

-levels of measurement down side-nominal at top, ordinal interval (remember NOI)
-along top is test of difference and test of correlation/association, but in test of difference is unrelated and related.

then it’s Chi2, sign test, chi2, mann- whitney, wilcoxon, spearman’s rho, unrelated t test, related t test, pearsons r.

Question 4

mnemonic

Answer 4

carrots should come mashed with swede under roast potatoes