INVESTIGATION OF MEASUREMENT PRECISION OF LATENT VARIABLES DEPENDING ON THE NOISE TEST RESULTS

Abstract

Measurement precision is an important aspect of experimental studies because it affects the effectiveness of decisions. The precision of measurements is influenced by many factors, including the quality of the original data or how the data was obtained. When measuring a latent variable, one of these factors is the precision of the indicators. For example, there may be situations in which, for various reasons, the subject gave the wrong answer instead of the correct answer and vice versa. This fully applies to conducting surveys. In general, we are talking about the noise of the original data. The aim of the work is to analyze the precision of measurement of the latent variable depending on the degree of noise of the original data. The results of the simulation experiment are used as initial data, which allows to simulate the necessary experimental situations. The analysis is conducted within the framework of the theory of latent variables, which allows to obtain estimates of the latent variable on a linear scale. The measurement precision of the latent variable is determined on the basis of absolute error. The noise level of the original data varies from 0 to 0.10%. It is shown that even minor noise of the original data significantly reduces the precision of the measurement of the latent variable.