Before p < 0.05 to Beyond p < 0.05: Using History to Contextualize p-Values and Significance Testing

Figures & data

Table 1 Criteria for a lasting framework for inference beyond p < 0.05

Criterion	How p < 0.05 Meeting criterion	Does p < 0.005 meet criterion?
Provides quantitative basis to address uncertainty	p-value varies with precision of estimator	Same as p < 0.05
Rewards increased precision	Significance achieved more easily with higher sample size	Same as p < 0.05
Matches intuitive understanding	Statistic akin to p-value developed several times	Stricter threshold easily understood, but requires departure from current intuition
Advocated by statisticians and nonstatisticians	Promoted by Fisher, his students, and scientists in a variety of fields	Supported by some statisticians and practitioners, but value still disputed
Computationally feasible for nonstatisticians	Fisher and Yates made tables user-friendly and accessible to scientists and practitioners	Any threshold feasible with modern software
Responsive to changing conditions	p < 0.05 met needs of a time when few tests were conducted while varying thresholds allowed responses to multiple testing	Addresses current preponderance of tests, but viewed by advocates as a stopgap measure