Empirical comparison of distance equations using discrete traits

Area of Science:

• Biological anthropology
• Paleostatistics
• Population genetics

Background:

• The Grewal-Smith statistic is a traditional method for assessing biological distance in skeletal populations.
• Its application has faced scrutiny regarding reliability and variance stabilization.
• Recent advancements propose corrections and transformations to enhance its utility, particularly for non-metric data.

Purpose of the Study:

• To compare the efficacy of the Grewal-Smith statistic against several corrected and transformed equations.
• To evaluate the performance of different statistical models using real-world population data.
• To determine if newer statistical approaches offer significant advantages over the Grewal-Smith method.

Main Methods:

• Rank order correlation statistics were employed to compare distance measures.
• Thirteen equations, including Grewal-Smith, Freeman-Tukey, Anscombe, and Bartlett transformations, were tested.
• Actual biological distance data from existing literature on skeletal populations were utilized.

Main Results:

• Little variation in results was observed between the tested equations when applied to the selected populational data sets.
• The Grewal-Smith statistic demonstrated comparable performance to more sophisticated models in this analysis.
• The study found that while newer methods may offer specific benefits, they did not universally outperform Grewal-Smith on this data.

Conclusions:

• The Grewal-Smith statistic is not inferior to newer, more complex models for measuring biological distance in skeletal populations.
• Advanced transformations and corrections may provide specific advantages for small sample sizes or low trait frequencies.
• The choice of statistical method should consider the specific characteristics of the population data being analyzed.