Review of methods for transformation skinfolds data to normality for boys and girls from 7 to 19 years old.
The most popular method to estimate fatness is anthropometry, especially measurement of skinfold thicknesses. However, such kind of data seriously departs from normal distribution. The present study reviews some methods not age or sex dependent for normalizing triceps and subscapular skinfolds data distribution. Material consists 1408 boys and 1390 girls aged 7-19 from Warsaw measured in 1987-88. The following methods were tested: log(x), log(x-c), -1/x, -1/square route of x. Approximation to the normality were judged by means of the Shapiro-Wilks W test. The results indicate that log(x-c) gave the best transformation of triceps skinfold for boys and girls. The best method for subscapular site appeared -1/x for boys and -1/square route of x for girls. The most popular log(x) is not sufficient for correct normalization. Nevertheless, none of methods have brought required distribution in each age class. It is suggested, depending on the character of material being transformed, to use all reviewed methods in order to achieve proper distribution in each age class.