Permutation test ( aka randomization aka exact test). Terms of use?

Question

Python

Hello! My question is mostly caused by the lack of Russian-language sources. indications (the results of elections by municipalities in two regions) and the statistical criterion of this hypothesis was the equality of the average percentage of voting for a candidate.The essence of the method was as follows: Two arrays of sizes n and m merge into one, then the order of the elements changed randomly and simply the first n elements were assigned the first array , and the rest - the second.The statistic criterion was considered and such mixing was carried out a large number of times.

Thank you for your response!

Answer 1

There is a "blurred" probability at the moment that one candidate is 3 times cooler than the other: 25%/75%.
It is necessary to find out a more accurate percentage of their leadership in most cases.
The permutation test provides a search for a balanced mean for a given number of inputs (a more accurate separation boundary, for example: 35% / 65%) with a given accuracy of all percentages for all tests performed.
In fact, it reduces the "scatter" of scores (of a single test) to a single most likely-possible value (using multiple tests) using random inputs.
It can be used where the numerical actual indicator of measuring one measured output value is not constant, but you cannot influence the factors that change it in any way (or get rid of their influence completely).