Nonparametric statistics deals with median rather than mean, because mean can be easily influenced by the outliers or skewness of data, and data are not normally distributed.
The main difference between parametric and nonparametric tests is that in parametric tests, data are normally distributed, whereas in nonparametric tests, data are not normally distributed.
Assumptions of Parametric Statistics.
- Random sampling
- Samples are independent
- Population is normally distributed
- Each population have a same variance
Assumptions of Nonparametric Statistics
- Samples are randomly selected
- Observations are independent
Some important Nonparametric Statistics.
- Sign test – Compare one population & Compare two dependent populations.
- Mann-Whitney U test – Compare two independent populations.
- Spearman correlation – linear association between two variables.
Positive points of non parametric tests.
- It can easily apply and compute
- Make few assumptions
- Most of the time results may expect as parametric tests
- Can be used where normally distribution can not made
- Easy to understand.
