Two diagnostic methods of mathematical models in economic studies that are based on the statistical analysis of residual errors O-S (Observation-Calculation) are considered in the article. The first method is recommended for sample sizes n within 30 < n < 500, using the x-test to verify the normality of the O-C values. The second method is recommended for the diagnosis of O-C errors with the volume n > 500. The famous Cambridge professor H. Jeffreys made the conclusion that such samples usually follow the Pearson distribution of type VII, which has a positive kurtosis. The mathematical model for volumes of O-C values n > 500 is considered adequate if the kurtosis for O-C is in the limits of 6,0-1,2 with insignificant asymmetries. The model is acceptable if the kurtosis for O-C is in the limits of 1,2-0,0. The model is considered inadequate (unacceptable) if the errors of O-C have a significant negative kurtosis or significant asymmetry.