The Expert Witness

You're the econometrics expert on the stand. One by one, regression tables are entered into evidence, each carrying a causal claim someone badly wants certified — a wage dispute, a policing budget, a drug hearing, a market model. For every exhibit you testify twice: the ruling — uphold the causal claim, find it not proven, or overturn it — and the grounds, the single reason why. The trap is always the same: a small p-value that begs you to call correlation 'cause.' Certify enough junk and the case collapses on appeal; but refuse to certify sound, randomized evidence and you're just an obstruction. It starts with a clean trial. It ends on the landmark case, with a confident opposing expert, a sympathetic story, and a p < 0.001 that everyone in the room wants you to bless.

⚖️

CredibilityThe court's trust in your testimony. Certify junk — or refuse to certify sound evidence — and it craters. Zero → discredited, struck from the record.

🔬

RigorThe analytical merit of your reasoning — naming the actual threat, not just the verdict. The mastery gauge: no shortcut past identifying why an exhibit fails.

🧠

ComposureHolding steady under cross-examination. A confidently wrong verdict — above all, endorsing a significant trap — rattles you. Zero → you crack on the stand.

☣️

ContaminationUnchallenged bias you've certified onto the record. Every trap you uphold pours more in. Past 100% the case collapses on appeal — a mistrial.

Ten regression tables enter evidence. The significant one is usually the trap.