Related Experiment Videos
Statistics of sexual size dimorphism
1Department of Anthropology, Washington University, 1 Brookings Drive, St Louis, MO 63130, USA. rjsmith@artsci.wustl.edu
Journal of Human Evolution
|April 20, 1999
Summary
Sexual size dimorphism (SSD) quantification methods are debated. This study finds that while many ratios are statistically inferior, two specific ratios (linear and logarithmic) are sufficient and appropriate for comparative SSD studies.
Area of Science:
- Evolutionary biology
- Quantitative genetics
- Comparative anatomy
Background:
- Methods for quantifying sexual size dimorphism (SSD) are controversial in comparative studies.
- Commonly used ratio-based methods for SSD may violate statistical assumptions and lead to spurious correlations.
Purpose of the Study:
- To evaluate the statistical properties of various methods used to quantify SSD.
- To compare ratio-based methods with residual-based methods in SSD analysis.
- To provide recommendations for appropriate statistical approaches in SSD research.
Main Methods:
- Analysis of 25 datasets from existing SSD literature.
- Replication of common statistical procedures using five different ratios and ratio-avoiding methods.
- Statistical evaluation of ratio and residual properties.
Main Results:
- Many commonly used SSD ratios are unnecessary; only linear and logarithmic scale ratios are recommended.
- Ratio-based methods do not inherently cause spurious correlations or non-normality issues in common SSD analyses.
- Residuals are not suitable replacements for ratios in many SSD pattern evaluations and are often misused.
Conclusions:
- Specific linear and logarithmic ratios are adequate for comparative SSD studies.
- Residual-based methods are frequently misapplied in SSD research, often requiring multiple regression instead.
- The findings are applicable to comparative methods with or without phylogenetic corrections.