Construction of 2x2 Contingency Tables from Normally Distributed Data. Analysis of the Contingency Tables Using a Modified Binomial Test. Accounting for Differentiating and Integrating Factors in the Calculation of Asymptotic Significance

Abstract

The work is devoted to determining the comparative efficiency of analyzers with a binary scale: 1/0, +/-, yes/no. Such a task arises, for example, when validating a new method of microbiological analysis for sterility when comparing the results of the analysis of two methods, a new one, usually more sensitive, and a traditional one. We will analyze the procedure for constructing contingency tables for normally distributed data, and then analyze the constructed tables using the modified Binomial Test. We will point out the factors that are critical for the correct calculation of the asymptotic significance (p-value).