Parametric

Parametric analyses assume a distribution of data concerning the population under study. These require the same assumptions as anova-assumptions:

• independence
• normality
• homoscedasticity

Non-parametric

Non-parametric analyses don’t make these underlying distributions assumptions. They convert the data into ranks and analysing these ranks in various ways.

Using parametric analyes we gain power, but we need to comply to assumption. Non-parametric don’t require these assumptions, but are less powerful.

References

• basic-experiment-design-concepts
• measuring-results
• variable-types
• test-of-proportions
• confounding-variables
• exerting-control
• measurement-errors
• anova-assumptions