Double Your Variance, Dirtify Your Bayes, Devour Your Pufferfish, and Draw your Kidstrogram

This article expands upon my presentation to the panel on “The Radical Prescription for Change” at 2017 ASA (American Statistical Association) symposium A World Beyond $p<0.05$. It emphasizes that, greatly enhance reliability of—and hence public trust in—statistical and data scientific findings, we need take a holistic approach. We lead by example, incentivize study quality, inoculate future generations with profound appreciations world of uncertainty world. The four “radical” proposals in title—with all their inherent defects trade-offs—are designed provoke reactions actions. First, research methodologies are trustworthy only if they deliver what promise, even this means that have be overly protective, necessary trade-off practicing quality-guaranteed statistics. guiding principle may compel us doubling variance some situations, strategy also coincides call raise bar from $p<0.05$ $p<0.005$ [3]. Second, teaching principled practicality or corner-cutting is promising community’s as well general public’s ability spot—and deter—flawed arguments findings. remarkable quick-and-dirty Bayes formula rare events, which simply divides prevalence sum false positive rate (or total error rate), featured popular radio show Car Talk, illustrates effectiveness strategy. Third, it should routine mental exercise put ourselves shoes those who would affected our finding, order combat tendency rushing conclusions overstating confidence pufferfish/selfish test can serve an effective reminder, help institute mantra “Thou shalt not sell thou refuseth buy” most basic professional decency. Considering personal stakes statistical endeavors points concept behavioral statistics, spirit economics. Fourth, current mathematical education paradigm puts “deterministic first, stochastic second” likely responsible difficulties reasoning under uncertainty, situation improved introducing histogram, rather kidstogram, early counting.