Limitations in the Accuracy of Estimates of Blood Group Gene Frequencies

Abstract

Summary and Conclusions1. For reporting results of calculations of blood group frequencies no more significant figures should be used than are justified by the size of the series tested.2. Technical and clerical mistakes have been responsible for gross discrepancies in reports on blood group distribution published in the literature, which qualified investigators can recognize by mere inspection of the data.3. Maximum likelihood calculations are complex and laborious and require a mastery of mathematics which few workers possess. Even in the hands of experts gross errors have resulted when carrying out the laborious computations, so that such calculations are not only expensive but impracticable and dangerous.4. The application of maximum likelihood calculations by Boyd to two series of individuals tested for the M-N-S types has yielded estimates of gene frequencies differing only in the third decimal place from the values obtained by the author with the aid of simple square root formulae. When one considers, even disregarding the possibility of technical and clerical errors, that the sizes of these two series justified the use of only two significant figures, it is evident that the maximum likelihood method contributed only a dubious increase in „ accuracy ”. Therefore, such calculations consume misdirected energy and place emphasis on superficialities while disregarding more important factors which can affect the accuracy of the data.5. Many investigators when studying the distribution of the blood groups make the tacit but incorrect assumption that the population is undergoing random mating and is therefore homogeneous. This appears to be one basis for the recent series of fallacious reports on the supposed relationship between the blood groups and various diseases.6. Confusion in the blood grouping field is due to the fact that most of the articles and books which are published have been written by individuals who have only superficial knowledge and understanding of the subject. The widespread use of the incorrect G-D-E notations reflects the lack of mastery of Rh-Hr serology shared by these workers.7. It is significant that the same workers who are enthusiastic for maximum likelihood calculations also endorse the pointless studies on blood group distribution in disease, and the use of the superficially attractive but fallacious C-D-E notations.