Factors affecting choice of Stat test Flashcards
8 stat tests
- Sign test
- Chi-Squared test
- Wilcoxon
- Mann-Whitney
- Related t-test
- Unrelated t-test
- Spearman’s rho
- Pearson’s r
choose the appropriate statistical test, there are three questions that need to be answered
- Do we need a test of difference or a test of relationship?
- If test of difference – what type of experimental design has been used?
If test of relationship – are we looking for an association or correlation? - What level of data has been collected?
appropriate statistical test can be identified from the table below
Test of Difference:
Independent groups:………..Repeated Measures:
Nom: Chi-squared Test…….Sign Test
Ord: Mann-Whitney U……….Wilcoxon
Int: Unrelated t-test…………..Related t-test
Test of Relationship
Association:……………..Correlation:
Nom: Chi-squared ……………. -
Ord: - ………………….Spearmans rho
Int: - ………………….. Pearsons’s are
non-parametric tests
*Spearman’s rho
*Wilcoxon
*Mann-Whitney
*Chi-Squared test
*Sign test
parametric tests
*Pearson’s r
*Related t-test
*Unrelated t-test
Why are parametric tests are preferred to non-parametric tests
Parametric tests are preferred to non-parametric tests as they are more powerful. This means that they are more likely to detect significance. However, there are 3 conditions that must be met in order for them to be used:
- Data must be interval level.
- Population data must be normally distributed (remember that many characteristics are normally distributed such as height, weight, IQ etc).
- The variances (based on standard deviation) for the two sets of data should be similar. This is termed ‘homogeneity of variance’.