Eliminating "systematic survivorship bias" in the attitude of specialists to the significance of investment attractive features of examined objects

Abstract

Methods for detecting and eliminating false measurements are well known in the theory of metrology. However, the relevant methodology is not adapted to the needs of the qualitative assessment of the impact of the human factor on expert decision-making. The "systematic survivorship bias" refers to the following: when involving experts in conducting examinations, they usually focus on that part of them, where statistically probable agreed opinions are observed based on the results of these examinations. Other experts are considered "marginal", their opinions are discarded and not taken into account, which defines the "systematic survivorship bias", also called the "paradox of information availability". Although the specified "marginality" may be a consequence of the unique experience of conducting examinations or, for example, the use of modern technologies by a specific specialist, little known to the general public. It should be noted that manipulation of statistical data with an orientation only on "successful" cases could be really dangerous, for example, in studies of the human factor in complex ergastic active and organizational management systems, in particular aviation. The rationale and implementation of the algorithm for detecting and eliminating the "systematic survivorship bias" have been given in this paper. m=90 specialists who are usually involved in various examinations by UkrINTEI took part in the research. The actual elimination of the "systematic survivorship bias" occurs after the implementation of a certain number of iterations of the algorithm given in the current work. As a result of iterations of the above-mentioned algorithm, it was established that four subgroups can be distinguished from the initial sample with the number of m=90, with the following numbers: mC=30 people, mH=12 people, mM=11 people, mT=6 people. For the specified subgroups, the consistency of group opinions satisfies the entire spectrum of hypothesis testing criteria established in this paper