Normality assumption
Dependent variable is normally distributed in each population.
How to test for the normality assumption?
- Shapiro-Wilk and Kolmogorov-Smirnov
- Descriptives
- Null hypothesis: Population distribution is normal
- If reject null hypothesis → check why they are’t normal: skewness and kurtosis
- Order descriptives
- If /= 0 > 2 stderr, evidence of phenomena in population
- Nonsignificance does not imply normal distribution of dependent variable
- If reject null hypothesis → check why they are’t normal: skewness and kurtosis
Related
Homogeneity of variance assumption