It is a test whose model does not specify conditions regarding population parameters, nonparametric tests can be used if the conditions in the parametric test are not met.

Advantages of nonparametric test:

• There is no assumption that the value to be analyzed has been drawn from a population by a certain system.
• Can be used on scores that are classifications and levels or ranges.
• Can be used on small samples.
• Simple in calculation.

Weaknesses of non-parametric statistical tests:

• If the data has met all the requirements in the parametric test, then it is useless if the nonparametric test is still carried out.
• There is no nonparametric method for testing interactions in the analysis of variance (additivity) model.

Case Two Samples:

Normality test using Kolmogorov-Smirnov test. Is an analytical technique to test whether a data set follows a normal distribution with a mean and variance 2.

The proposed hypothesis is:
     Ho: Data is normally distributed
     Ha: Data is not normally distributed

Test of Randomness (Run Test). Is an analytical technique to test whether the data that is used as a sample is selected randomly from a certain population.

The proposed hypothesis is:
     Ho: The data is selected randomly
     Ha: The data is not selected randomly

Case Two Related Samples:

McNemar test. To see the significance of the changes in the design before and after, with the data on a nominal scale. The proposed hypothesis is:
     Ho: There is no significant change from before and after
     Ha: There is a significant change from before and after

Case of Two Independent Samples:

Chi-Square Test. The Chi-Square Test is suitable for analyzing data that is a discrete category with the number of categories of 2 or more, which is used to determine the significance of differences between two independent groups that have frequencies in discrete categories, and to see if the two groups are mutually independent in the sense of the word that the existence of scores in the categories in the first group is influenced by the categories in the second group.
Mann-Whitney U test. To test whether there is a difference between two independent groups whose data is on an ordinal scale. Before entering the test, it must first be determined that n1 = the number of smaller cases and n2 = the number of larger cases, then give a ranking starting from the lowest in the combined group (n1+n2).
The proposed hypothesis is:
- One-way test
     Ho: There is no difference between group 1 and group 2
     Ha: The average in group 1 is smaller than group 2
- Two-way test
     Ho: There is no difference between group 1 and group 2

     Ha: There is a difference between group 1 and group 2

Case K Related Samples:

Qochran Q test. To test whether there is a difference between k sample groups is related, with the data in that group on a nominal scale (dichotomy). Before carrying out the test, first a N x k table is made, then change the scores to 1 if it is considered "successful" and 0 if it is considered "failed".
The proposed hypothesis is:
    Ho = There is no difference between the k groups
    Ha = There is a difference between the k groups

Friedman's Two-Way Analysis of Variance. Used to test whether there is a difference between k sample groups related to data with at least ordinal scale. Prior to the test, a table of N x k is first made. Scores in each row are ranked separately, so that each row is ranked from 1 to k.
The proposed hypothesis is:
    Ho = There is no difference between the k groups
    Ha = There is a difference between the k groups

Case k Independent Sample:

Kruskall-Wallis One-Way Analysis of Variety. To test whether there is a difference between k independent groups, with data of at least ordinal scale. In the calculation, all scores in the entire sample group used are ordered in a series, starting from rank 1 to N.
The proposed hypothesis is:
    Ho = There is no difference between k independent groups
    Ha = There is a difference between k independent groups