Hauspie R C, Susanne C, Defrise-Gussenhoven E
Ann Hum Biol. 1985 Sep-Oct;12(5):429-40. doi: 10.1080/03014468500007991.
Factor analysis with VARIMAX rotation was used to analyse 15 face measurements in Belgian same-sexed twins, aged 18-25 years: 39 dizygotic and 57 monozygotic male pairs and 42 dizygotic and 67 monozygotic female paris. According to Christian's model, we used the ratio of the within-mean squares of dizygotic and monozygotic pairs to test for the presence of a genetic component in the variance of the facial dimensions and of all the rotated factors were statistically significant (P less than 0.05), suggesting a genetic component in the variance. The probabilities of the F values were generally lower in males than in females. The factor analysis yielded five main factors of which three were well separated: face height, ear size and lips. The two others were breadth factors, but were less clearly defined, probably due to a bad selection of variables. A comparison of the F values of the factors with those of their contributing variables seemed to indicate that well-defined factors may better describe genetically determined structures than the original variables can.
采用方差最大化旋转的因子分析方法，对18至25岁的比利时同性双胞胎的15项面部测量数据进行分析：39对异卵双生子和57对同卵双生子男性，以及42对异卵双生子和67对同卵双生子女性。根据克里斯蒂安模型，我们使用异卵双生子和同卵双生子对内均方的比率，来检验面部尺寸方差中是否存在遗传成分，所有旋转因子均具有统计学意义（P小于0.05），表明方差中存在遗传成分。F值的概率通常男性低于女性。因子分析产生了五个主要因子，其中三个因子区分明显：面部高度、耳朵大小和嘴唇。另外两个是宽度因子，但定义不太清晰，可能是由于变量选择不当。将因子的F值与其贡献变量的F值进行比较，似乎表明定义明确的因子比原始变量更能描述遗传决定的结构。
