Nonparametric test

#In Data Analyst series

Nonparametric test is used when data does not follow normal distribution, and is suggested to use when participants are less than 30. (We all know that the larger the dataset, the more normally distribution it is).

Type of nonparametric tests:

• Mann Whitney U Test: (nonparametric version of independent group t-test)

It tests whether the scores obtained from two unrelated samples differ significantly from each other.

It tests whether two unrelated groups have similar distributions in the universe in terms of the variable of interest.

The scores in different groups are given a rank number starting from the smallest and the difference between the mean rank is examined.

The dependent variable is at least on the rank scale

Observations are independent of each other.

• Wilcoxon Signed Ranks Test

It is used to test the significance of the difference between the scores of two related measurement sets.

The dependent variable is at least on the rank scale.

Where the Related/Dependent samples t-test can be used, this analysis is used if the scores do not show a normal distribution.