Drawing, I dare to say, is in practice something that is commonly believed, and it can be done in various, usually precisely defined ways. In general, it is nothing like the “random” option or the random option that a person is supposed to meet “coincidentally”. Therefore, the disclosure of the algorithm for drawing judges – which is demanded by the judicial minister by civil organizations before the Supreme Administrative Court (see Eva Ivanova, “Geobro is to disclose reports from judges” – judges randomly rewarded Should be considered necessary for the evaluation and control of the mechanism of cases.
Unlike many popular associations, drawing is not something disorganized, indescribable or even some mysterious activity. Conversely, randomization practices a rigorous implementation of a fully deterministic human algorithm. Such an algorithm is no more than a detailed sequence, which is performed in the order of a set of units (such as cases of judges or judges) from which the drawing is made. The sequence of these operations is subject to prior selection of a specific sampling technique (in statistics it is called sampling scheme).
The person commissioning the development of the sampling algorithm points to one of the sampling techniques, primarily directed at the purpose it is to serve, and – rarely mentioned – the initial knowledge of the community in which To sample from. These are two major, interrelated issues that determine the choice of this or that drawing technique.
The most basic drawing scheme, the so-called simple individual drawing, which for many of us is the only known method of drawing, and moreover which best expresses the essence of random selection, is rarely used in practice. First, because in internally diverse communities this may result in obtaining a less representative sample. To protect yourself against this, it is good to first analyze everything you know about the entire community and use this information in the drawing process. Each use of information of this type of preferences about a given community means a departure from the simple individual drawing scheme or at least its modification. For example, the person ordering the draw may first divide a group of units of interest into teams (layers) and draw separately from each layer. In this way, it would ensure that a small number of layers of the community would also have their representatives in the drawn sample. But it cannot be a mystery, first, according to which the features or characteristics were divided into layers of community, and secondly, how individuals at the individual level are attracted. Both these elements should not only be made public, but also well established. It is important how and why layering was done in one way and not the other. It is also important how units are drawn from layers, ie according to which scheme and how many units from each layer. Without this information, it is impossible to estimate the statistical properties of randomly selected samples, especially whether in the light of available knowledge about community units, the use of the algorithm serves the main purpose of its application.
The same applies to any other drawing scheme, which should be well explained and detailed decisions involved.
Only complete transparency in this regard will free us from the need to assume that the drawing is done according to strict laws of statistics and probability.
Those who create and use random algorithms see us as rational and important.
Pro dr hab. Miroslav Szreder
Head of the Department of Statistics, University of Gdansk
Write to: list @ wyborcza.pl
Analyst. Amateur problem solver. Wannabe internet expert. Coffee geek. Tv guru. Award-winning communicator. Food nerd.