A Parametric Bootstrap for the Mean Measure of Divergence

Area of Science:

• Anthropology
• Bioinformatics
• Statistical Genetics

Background:

• The Mean Measure of Divergence (MMD) is a key tool in anthropology for analyzing non-metric traits.
• Small sample sizes and insufficient trait measurements pose challenges in anthropological and palaeoanthropological research.
• Bootstrapping techniques have historically aided in overcoming data limitations since 1969.

Purpose of the Study:

• To present a novel parametric bootstrap technique for analyzing non-metric traits.
• To address the limitations of small or incomplete datasets in anthropological studies.
• To offer a statistically powerful alternative to non-parametric methods when distributional properties are known.

Main Methods:

• Utilizing an Anscombe transformation to stabilize variance, yielding a normally distributed variable θ.
• Developing a parametric bootstrap procedure based on the known probabilistic distribution of the transformed data.
• Validating the method with both simulated and real-world anthropological data.

Main Results:

• The transformed variable θ, after Anscombe transformation, approximates a normal distribution.
• The proposed parametric bootstrap method demonstrates statistical power, particularly with limited data.
• The technique provides a robust approach for analyzing non-metric traits in small samples.

Conclusions:

• The developed parametric bootstrap method is effective for anthropological studies with data scarcity.
• This approach offers a more powerful alternative to non-parametric bootstrapping when distributional assumptions are met.
• The method enhances the study of non-metric traits in challenging sample conditions.