Green R F, Suchey J M, Gokhale D V
Am J Phys Anthropol. 1979 May;50(4):629-34. doi: 10.1002/ajpa.1330500414.
At present there is no standardized method to derive frequencies from bilateral non-metric traits. In this paper the commonly used methods are evaluated in light of statistical considerations and degree of sample preservation. In particular, we explore the question of dependence between sides for the bilateral traits. Several workers have tested for bilateral trait correlation in incorrect ways, confusing tests for differences in side frequencies with tests for independence. An easy method to test for independence, using the chi-squared test, is recommended. This test is used on 16 bilateral traits in a large sample of prehistoric carnia from Central California and significant dependence is found for all 16 of these traits. We suggest the traditional method of deriving frequencies be used. Both sides of the cranium should be considered, dividing the number of times the trait occurs by the number of sides available for observation. This method can be used even in poorly preserved samples. Side to side correlation may be compensated for by modifying the constants subtracted in the mean measure of divergence and in the expression for the approximate variance of the mean measure of divergence when the samples are drawn from identical populations.
目前，尚无从双侧非度量性状推导频率的标准化方法。本文根据统计考量和样本保存程度对常用方法进行了评估。特别是，我们探讨了双侧性状两侧之间的相关性问题。一些研究人员以错误的方式测试双侧性状相关性，将两侧频率差异测试与独立性测试混为一谈。建议使用一种简单的方法，即卡方检验来测试独立性。该检验应用于加利福尼亚中部大量史前颅骨样本中的16种双侧性状，发现所有这16种性状都存在显著相关性。我们建议使用传统的频率推导方法。应考虑颅骨的两侧，将性状出现的次数除以可用于观察的侧数。即使在保存不佳的样本中也可使用此方法。当样本来自相同总体时，通过修改平均离散度测量值以及平均离散度测量值近似方差表达式中减去的常数，可以补偿两侧之间的相关性。
