Open AccessCommon errors in filling in spreadsheets and how to avoid them
Author(s) -
С. Н. Лысенков
Publication year - 2021
Publication title -
rossijskij žurnal geriatričeskoj mediciny
Language(s) - English
Resource type - Journals
eISSN - 2686-8709
pISSN - 2686-8636
DOI - 10.37586/2686-8636-1-2021-105-109
Subject(s) - zero (linguistics) , statistics , systematic error , random error , non sampling error , computer science , sample size determination , statistical analysis , mixing (physics) , sample (material) , algorithm , data mining , mathematics , philosophy , physics , quantum mechanics , chemistry , chromatography , linguistics
Errors during research can occur not only at the stage of planning, data collection and statistical analysis, but also at the stage of lling in spreadsheets. In most cases they can be easily corrected, otherwise results may be distorted. These errors can be categorized into invalid format and input errors. In the rst case, the mixing of the absence of data and the zero values is especially dangerous, since it can lead to systematic errors: overestimation (if the zero values are lled in as the absence of data) or underestimation (if the absence of data is lled in as zero values) of the mean or the prevalence. Input errors are most often random, and their effect decreases with increasing sample size, but for the better analysis they should also be corrected as much as possible. The article provides an algorithm that allows you to nd input errors and correct them before statistical analysis.