Play in pairs on one phone. You are a research group that badly needs a headline result from the BRFSS 2024 teaching sample. For the game, assume the card below is the truth about the world: no causal effect. (In real life some of these pairs are associated. That is the point: the tricks below produce a result whether or not an effect exists.) Your job is to make the data say there is one.
Choose 2 to 4. Each trick pushes your odds ratio further from 1, in the direction you chose. More tricks means a bigger number and a bigger chance that a reviewer catches you. The odds ratio below is illustrative; the real figures from our dataset are on the cards.
Four short parts. Make it sound like a real abstract. Use your tricks without naming them. Be ready to read it aloud in 60 seconds if your pair is picked.
0 words
Six abstracts, written for this game on the variables in our teaching file. Some use tricks, some do not. Tick every trick you think an abstract uses, then lock in. A correct spot is plus one, a false accusation is minus one, and a clean abstract you let pass is plus two. Where a trick is arguable, ticking it is neither rewarded nor punished and the explanation says why. Numbers in the abstracts are invented unless the explanation says otherwise.
Detective points: 0 out of 16.
| Part A · Villain | 0 / 10 |
| Part B · Detective | 0 / 16 |
| Total | 0 / 26 |
Tell the room your total when asked. Highest villain score reads their abstract aloud.
Every one of these is a reason a cross-sectional study cannot show what came first or what causes what. Your own abstract must be the one none of them can touch.
A cross-sectional study shows who has what. It cannot show what came first. Write "is associated with", never "causes" or "leads to".