In defense of P values.

Area of Science:

• Ecology
• Statistical Methodology

Background:

• Ecologists frequently criticize statistical hypothesis testing, particularly P values.
• Many criticisms stem from misinterpretations rather than inherent flaws in P values.

Purpose of the Study:

• To review common criticisms of P values in ecological research.
• To demonstrate the statistical linkage between P values, confidence intervals, and Akaike's Information Criterion (AIC) differences (delta AIC).
• To argue for the continued utility of P values alongside alternative metrics.

Main Methods:

• Literature review of criticisms regarding P values.
• Conceptual and mathematical linkage of P values with confidence intervals and delta AIC.
• Comparative analysis of the arbitrary nature of threshold choices in P values and delta AIC.

Main Results:

• Most criticisms of P values arise from misunderstandings or incorrect application.
• P values are statistically equivalent to confidence intervals and delta AIC.
• Threshold selection for delta AIC is as arbitrary as Type I error rates in hypothesis testing.

Conclusions:

• P values, confidence intervals, and delta AIC are all valuable statistical tools.
• The choice of metric should be application-dependent, not based on dogma.
• Modern statistical practice can effectively integrate these different measures.