Before you perform any statistical analysis the characteristics
of the data you have collected must be established. This will provide
you with the information to choose between the two major groups
of statistical analyses - parametric or non-parametric tests.
Parametric tests
In general, parametric tests
- compare differences between means
- are used with interval or ratio level data
- require large sample sizes
- require data to be normally distributed
- are more sensitive (have more statistical power) than non-parametric
methods
Examples of parametric tests include are the t-test, ANOVA and
regression.
Non-parametric tests
Non-parametric tests tend to be simpler than parametric
tests, and are generally used
- to compare differences between medians
- if data are ranked or nominal
- where sample sizes are small
- when data are not normally distributed
Examples of non-parametric tests are the Mann-Whitney U-test, Kruskal-Wallis
test, Chi-squared test and Spearman's rank correlation.
back to "Types of statistical analysis"
|