REGRESSION ANALYSIS OF NATURAL SELECTION: STATISTICAL INFERENCE AND BIOLOGICAL INTERPRETATION

REGRESSION ANALYSIS OF NATURAL SELECTION: STATISTICAL INFERENCE AND BIOLOGICAL INTERPRETATION

November, 1987 | THOMAS MITCHELL-OLDS AND RUTH G. SHAW
The paper discusses the statistical and biological issues in applying Lande and Arnold's (1983) multiple-regression approach to measure natural selection in wild populations. It highlights the challenges posed by multicollinearity and unmeasured factors affecting fitness, which can complicate inference about selection. The authors emphasize that while the fitness-regression method is a valuable tool for suggesting hypotheses about selection forces, it should be used cautiously due to potential biases and the inability to unambiguously attribute differences in fitness to specific characters. They suggest that manipulative experiments are more effective for inferring causal relationships between phenotypic characters and fitness. The paper also reviews the assumptions of multiple regression and discusses alternative definitions of stabilizing and disruptive selection. Finally, it provides guidelines for validating the model and emphasizes the importance of experimental validation to address the limitations of observational data.The paper discusses the statistical and biological issues in applying Lande and Arnold's (1983) multiple-regression approach to measure natural selection in wild populations. It highlights the challenges posed by multicollinearity and unmeasured factors affecting fitness, which can complicate inference about selection. The authors emphasize that while the fitness-regression method is a valuable tool for suggesting hypotheses about selection forces, it should be used cautiously due to potential biases and the inability to unambiguously attribute differences in fitness to specific characters. They suggest that manipulative experiments are more effective for inferring causal relationships between phenotypic characters and fitness. The paper also reviews the assumptions of multiple regression and discusses alternative definitions of stabilizing and disruptive selection. Finally, it provides guidelines for validating the model and emphasizes the importance of experimental validation to address the limitations of observational data.
Reach us at info@study.space