Sanity-check an analysis before I share it

Catches the mistakes that survive to the slide deck.

Here is an analysis I'm about to share. Try to break it.

Check for:
- Survivorship and selection bias in how the data was collected.
- Confounders that would explain the result more simply.
- Whether the sample supports the confidence being claimed.
- Denominator and time-window problems.
- Anywhere correlation is being read as causation.

For each issue, say how much it would change the conclusion if true.
If the analysis holds up, say so.

ANALYSIS:
[PASTE ANALYSIS]