The common classification of statistics is to divide it into parametric and nonparametric statistics. In the simplest form it should be said that parametric statistics are used to measure the hypotheses that are small in size. Quantitative variables are due to the fact that they are quantitative and indivisible because of the fact that they are moderate and standard deviations, and because of this characteristic, it is common for parametric tests to have assumptions that include the normal distribution of the society. Because in the absence of normal distribution, the mean and standard deviation do not represent the actual representation of the data.

Nonparametric statistics are used to test the qualitative variables and rank. These tests, also referred to as "no-default tests", do not require any special assumptions.

Regarding the conversion of variables, it should be recalled that quantitative variables can be converted into qualitative variables and evaluated them with nonparametric tests, but the opposite is not possible.

It is worth noting that the level of accuracy of parametric statistics is more than nonparametric statistical tests and it is usually suggested that nonparametric tests should not be used if parametric tests are possible. It should be noted that most behavioral science variables are judged by nonparametric tests. Placed.

As you know, the random variable may be assigned to one of the four measurement scales, such as nominal, order, distance, and relative. A statistical method is said to be nonparametric when there is at least one of the following conditions:

1- Suitable data that has a nominal scale.

2- Appropriate data that has a sequential scale.

3- Suitable data that have a relative distance scale, but the population distribution function of the random variables from which the data is obtained is not clear.

Advantages of using non-parametric methods:

1- Calculation of nonparametric methods is usually easy.

2. Nonparametric methods can be used for data that parametric methods can not be applied to. This situation is where the scale of data measurement is nominal or sequential.

3. In nonparametric methods, it is not necessary to assume that the random variable of the population has a probability distribution. These methods are based on the sampling distribution, but in the form of sampling, it is not necessary to assume a specific form for population probability distribution.

4. If a non-parametric method can be applied to a weak measurement scale, then it can be used for more robust scales.

More Maysam Toghraee's questions See All
Similar questions and discussions