Effects of Intra- and Inter-Individual Variation on Distributions of Single Measurements

[摘要] When healthy individuals are surveyed to estimate the "normal range" of some measured variable, generally only a single determination of the variable is obtained for each person. The distribution of such values reflects intra-individual variations, (including analytic deviations) as well as the differences among individuals with respect to such parameters as mean or variance. These underlying sources of variation have been expressed in a conditional probability model from which general equations have been derived showing the effects of these variations on the shape parameters (skewness and kurtosis) of a single-sample distribution. These results may help to explain the shape of a given distribution. More generally, they imply that methods of calculating normal ranges would benefit from a study of various mathematical transformations that could convert distributions of almost any shape to approximately gaussian form. Data from recent blood-chemistry studies are used to compare observed shape statistics with those calculated from the model.

[学科分类] 过敏症与临床免疫学