Don't Know the Half of It

Despite its collective efforts, the scientific community knows a lot less about how the world works than I usually give it credit for.

Bayes As Odds

 An intuitive way to understand Bayesian updating is via odds instead of probabilities or percentages. (chapter 8 - Updating - of Reason Better by David Manley)

Map Ain't Territory Reminder

Scientific models like causal DAGs are probably much cruder instruments for understanding real-life complex phenomena than I have been hoping recently.

Science is 'Kaleidoscopic'

Science is collaborative in a variety of ways: many individuals working on the same or related problems, and those individuals employing several distinct methodologies to attack those problem clusters.

Fake-Data Simulation

Fake-data simulation is an extension of abductive reasoning, specifically exploring the implications of various competing hypotheses.

Statistical Garnish

"Statistical garnish" is prevalent in opinion articles & debates: using a small set of fancy-seeming stats to support your case, contra a genuine attempt to understand the scientific/statistical literature on an issue (like SSC often does).

Forward & Backward "Why" Questions

There are at least two distinct types of 'why' questions: specific ones about (forward) causal inference, and abductive ones about generating new hypotheses (reverse).

Air Conditioning & COVID-19

Potential interaction effects make modeling associations harder.

Simplest Critique of The Book of Why

A causal diagram approach may just shift where the questionable assumptions lie.

Collider Bias

There are more complicated correlations that don't equal causation, such as collider bias. (I learned this reading chapters 4-6 of The Book of Why.)

Statistics Is About Ignorance

Statistics is not about detecting facts, but about ignorance.

Laundering Uncertainty

A danger of statistics is using it to launder uncertainty.