Fundamental Rule:

Choose the most powerful test possible.

STATISTICAL ANALYSIS

How to choose the appropriate test…

Ask:

How many groups?

Independent or Dependent Design?

Assumptions for test met?

PARAMETRIC TESTS

Independent t-test

Dependent (Paired or Repeated measures) t-test

ANOVA

Posthoc tests to compare group means

Assumptions

· Are samples normally distributed?

o Skewness and Kurtosis Rule

· Are samples interval or better data?

· H.O.V. (Homogeneity of Variance)

o This assumption only applies to independent design

o Four-fold Rule

NON-PARAMETRIC TESTS

Independent with two conditions: Mann-Whitney U-test

Dependent with two conditions: Wilcoxon or Sign Test

· Wilcoxon is more powerful than the Sign Test

Multigroup: Kruskal-Wallis

NOMINAL DATA

Chi-Square Test