The purpose of the present study is to present a method for evaluating brief zygosity detemination by use of data consisted of physique which is obtained easily at schools. As a step, we investigate previous methods of zykosity tests and statistical analysis for similarity of twins. We then propose normalized statistics for physical fitness in adolescent twins to investigate the differences of distributions between within-pair of twins and the control group consisted of singletons. We make statistics which is absolute pair difference against the standard deviation in each age and sex to standardize data so that we can pool the data sets from different age and sex. For any pair of data (y_i, y_j) Statistics : Z_<ij> =ΔY_<ij>/σ where ΔY_<ij> is the absolute difference between y_i, and y_j. As the basic model, we consider a bivariate normal distribution with unknown correlation coefficient. We then consider for types of the assumptions on the equivalence of correlations of monozygotic twins (MZ) and dizygotic twins (DZ). Here we consider the following four models Model 1: ρ_1>ρ_2> 0 Model 2: ρ_1>ρ_2= 0 Model 3: ρ_1=ρ_2> 0 Model 4: ρ_1=ρ_2= 0 Where ρ_1 and ρ_2 are correlation coefficients between MZ and DZ respectively. The best model was selected by using the Akaike's Information Criterion (AIC) . AIC is statistics produced by Akaike,H. in 1971 in order to estimate best model using observed model. The model is estimated by the method of the maximum likelihood. AIC = -2 log (maximum likelihood) + 2 (number of independently adjusted parameter within the model) According to the minimum AIC procedure, we will obtain the best model by picking up the one which attains the minimum value of AIC. By selecting the best model by AIC, it was found that 1) In the case of Body Height and Sitting Height, the correlation of MZ and DZ are significantly different. 2) The correlations of MZ are very high. ρ_1 of Body Heiht = 0.92 ρ_1 of Sitting Height = 0.87 3) On the other hand, those of DZ are relatively low and for Body Height ρ_2 is considered as zero. 4) In the case of Body Weight and Chest Cirtumference, ρ_1 and ρ_2 are considered to be identical. Since the number of DZ is very small, the results obtained in the previons section are not so reliable. The future study with increased number of DZ will be necessary. In this paper, Body Height, Body Weight, Chest Circumference and Sitting Height are modeled individually. However, it is also possible to consider these four date simultaneously.
View full abstract