The lack of statistical significance tests, p-value, does indeed allow much more, for “Irrefutable nonsense [to] rule”

Sir, Anjana Ahuja writes: “generally only studies with p-values lower than 0.05 are deemed to be of ‘statistical significance’. This magic number has calcified into the pivot on which science principally turns. Now, academics, [because of] “p-hacking”: cherry-picking experimental methods, slicing data and contorting statistical analyses to yield a desirable p-value, are arguing for “the entire concept of statistical significance to be abandoned”. “Beware making a fetish of an arbitrary number”, March 28.

But “John Ioannidis, from Stanford University, defends it a “convenient obstacle to unfounded claims”. Its absence, he warned, may unleash worse: ‘Irrefutable nonsense would rule.’”

What’s the p-value of the risk weighted bank capital requirements for banks not measuring the dangers to our bank systems correctly? I have no idea but I am sure that null hypothesis would not be rejected by a very long shot.

In fact to test, as null hypotheses, the current regulatory premises, that of what is ex ante perceived as risky being dangerous to our bank system, and that of what is ex ante perceived as safe being safe for our bank system, would surely return very low p-values and be rejected.

But Sir, no such statistical analysis was performed and so, in its absence, the “irrefutable nonsense [of the Basel Committee’s risk weighted capital requirements [does indeed] rule.”

PS. But anyone who has heard that saying attributed to Mark Twain of “A banker is a fellow who lends you his umbrella when the sun is shining, but wants it back the minute it begins to rain”, would not really do research to understand the issue.

PS. Another null hypothesis that could be tested is roulette will survive the pay out for its different bets to be tilted in favor of "safer" bets on colors and against "riskier" bets on numbers.

@PerKurowski