Zalety i ograniczenia wnioskowania w modelu permutacyjnym Fishera-Pitmana
The advantages and limitations of inference in the Fisher-Pitman permutation model

Summary/Abstract: Classical statistical inference is based on the Neyman-Pearson population model. An alternative to it is the permutation model, which is used much more rarely, but which has many advantages. In the permutation model, hypothesis testing is performed using permutation tests, which do not require assumptions regarding the form of random variable distributions, and which are especially recommended for very small samples. It is also possible to perform statistical inference for fixed, non-random groups. The aim of this article is to indicate key advantages and possibilities of applying statistical inference based on the Fisher-Pitman permutation model to scientific research, particularly in the field of economics. The article presents arguments in favour of the wider use of the permutation model in analyses where the classical population approach is insufficient or impossible to use. The article also shows the limitations of permutation tests. Their practical worth is demonstrated on the example of the assessment of changes in the live birth rate per 1,000 population in powiats by voivodship in 2023 and 2024. The analysis showed that permutation inference allows the detection of significant differences even when comparing non-random groups.

• Details
• Contents