Causal claims — diagnosis cards
Eight claims, each broken in a different way. Cut along the dashed lines and deal one per team. Page 2 is the diagnosis sheet.
Breakfast
"Students who eat breakfast score higher. Schools should serve breakfast."
A school district, proposing a budget line
Diagnose using this morning's vocabulary. Draw it.
The consultants
"Firms that hire consultants grow faster than firms that don't. So hire consultants."
A consulting firm's own marketing deck
Diagnose using this morning's vocabulary. Draw it.
The hiring puzzle
"Among the people we hired, the ones with the best test scores had the worst interviews. Our test must be measuring the wrong thing."
A head of recruiting, about to scrap the test
Diagnose. This one is not what it looks like.
The loyalty app
"Customers on our app spend 40% more. Getting the other half onto the app would raise revenue 40%."
A retailer's board paper
Diagnose using this morning's vocabulary. Draw it.
The promotion gap
"Once we control for salary band, the gender gap in promotion disappears. So there is no bias in promotion."
An HR review, closing the matter
Diagnose. Controlling for the wrong thing is also a mistake.
The founders
"Nine of the ten most valuable startups were founded by people who dropped out. Dropping out is an advantage."
A widely shared post
Diagnose. Who is missing from this comparison?
The turnaround team
"We send the turnaround team to our worst-performing stores. Those stores improve. The programme works."
An operations VP, requesting more budget
Diagnose. What would have happened anyway?
The crime wave
"Reported cybercrime doubled after 2019. Cybercrime is out of control."
A vendor selling security software
Diagnose. What changed besides the world?
Diagnosis sheet — team ______ · card ____
How this runs
Every team works its card at the same time. Then one team is drawn at random to defend at the front while everyone else attacks — and we draw again, and again. You will not know whether you are up, so be ready. Points for an attack that lands, and you have to name what is wrong, not just dislike it.
The graph behind the claim — draw it
Name it with the vocabulary: confounder · collider · mediator · selection · reverse causality · survivorship · regression to the mean · the measure changed
What is actually wrong here
What would you have to condition on — and what must you leave alone?
What comparison would settle it, if you could run any study you liked?
From the floor: the objection that landed