Systematic selection error

Systematic selection error is a statistical concept that shows that the conclusions made in relation to a group may be inaccurate due to improper selection in this group.

Content

Results Selection Errors

May include preliminary or subsequent selection with the prevalence or exclusion of certain species. This may, of course, be a form of scientific fraud , data manipulation, but it is much more often a bona fide delusion, for example, due to the use of an inappropriate tool.

For example, in the era of using film to photograph the sky, an independent observer would definitely come to the conclusion that the number of blue galaxies is clearly greater than the number of red ones. Not because blue galaxies are more common, but only because most films are more sensitive to the blue part of the spectrum. The same independent observer would make the exact opposite conclusion now, in the era of digital photography, because the matrices of digital cameras are more sensitive to the red part of the spectrum.

Types of Systematic Errors

There are a large number of possible systematic errors, the main types:

Space

Select the first and last point in the series. For example, in order to maximize the declared trend, you can start the series with a year with unusually low rates and end the year with the highest rates.
"Timely" ending, that is, when the results fit into the desired theory.
Separating a part of the data on the basis of knowledge about the entire sample and then applying the mathematical apparatus to this part as a blind (random) sample. See Zoned Sampling , en: cluster sampling , Marksmanship Error .
The study of the process on an interval (in time or space) is obviously less than the length required for a complete picture of the phenomenon.

Data

Crossing out some “bad” data in accordance with the rules, even if these rules were contrary to the previously announced rules for this sample.

Members

Preliminary selection of participants, or, for example, placement of an announcement on the recruitment of volunteers for participation in the trials among a certain group of people. For example, to prove that smoking does not harm the results of fitness, you can place an advertisement in a local fitness center for recruiting volunteers, but for smokers, type in a master class, and non-smokers among beginners or in a section who want to lose weight. For example: "An online population survey showed that 100% of the population use the Internet."
Throwing out the sample of participants who did not reach the end of the test . In the weight loss program, we consider detailed weight loss schedules as proof of the correctness of the methodology, but these graphs do not include participants who did not reach the end and who considered that this technique does not work for them.
Systematic error of self-selection. That is, a group of people for study is formed partly of their own free will, since not all respondents will want to participate in the test.

Resolving a Systematic Error

In the general case, it is impossible to single out a systematic sampling error only on the basis of statistical methods, although, as shown in the work of Nobel laureate James Heckman , there are working strategies in some special cases.

The famous phrase is “the stories about the mind and kindness of dolphins are based on the stories of tired swimmers who they were pushing to the shore, but we are unable to hear the story of those they were pushing to the other side”.