It is very important for health instructors to get some imformations on when each pupil will have the first menstruation. The significant relationships between sexual maturation and anthropometric measures have been realized in many studies and also in our studies. Therefore, the prediction of first menstruation with anthropometric and growth speed measures was attempted in this paper. Two hundred junior high school girls who all experienced their first menstruation already were used as the sample. The predictor variables are as follows; stature, body weight, chest girth, sitting height, and growth speed measure, which were computed in (growth amount per a year) /1O in the four anthropometric measures. The sample was classified into four groups ; the first whose menarche ages were later than 10 years of age, the second later than 11, the third later than 12, and fourth later than 13. Let R_<12> be the row vector of correlations of menarche age with the predictor variables and R_<22> be the correlation matrix of predictor variables, so the coefficients of prediction formulaes were given by R_<12> R_<22>^<-1>X, where X stands for the column vector of predictor variables. Then, the prediction formulaes were investigated with such simulation technique as dropping the variables which showed minimum in the products of β-coefficient and correlation of menarche age with prediction variable and also with evaluating the multiple correlation. After all, for the second group, the prediction formula with all variable ( 8 anthropometric and 8 growth speed measures) showed the multiple correlation of 0.942, the one with 15 variables, excluding the stature of 6th grade, 0.902, and their 95% error of estimate were ± 0:541 year, and ± 0.715 year, respectively. The prediction formulaes for other sample groups were constructed, too, but they did not show good precision enough to be applied. Therefore, it can be concluded that prediction of menarche age with anthropometric measures of 5th and 6th grade is most effective. And it was proved that there were little error between the actual menartche age and the predicted one in other sample.
View full abstract