Proof Is the conflict-of-interest pattern real?
Our core claim is strong, so it deserves a strong test: do members of Congress trade more in the sectors their committees oversee than you'd expect by pure chance? Instead of asserting it, we prove it with a permutation test — a standard, assumption-light statistical method. No story, just the data shuffled thousands of times.
the member actually oversees
(average of 5,000 shuffles)
Across 19,439 trades by members who oversee at least one sector, 8.85% landed in a sector that member's own committees oversee. When we randomly shuffle which member made which trade — breaking any real link between a member and their jurisdiction — that figure falls to about 10.71% (± 0.2%). The observed figure sits within the range chance produces (p = 1.0000), so on this data we cannot claim a systematic jurisdiction effect. We report that honestly rather than overstating the case.
How the test works
A permutation test asks a simple question: if there were no real link between a member and their committee's sectors, how often would we see an overlap this large just by luck?
- Observed statistic. Count the share of trades whose sector the trading member oversees — 8.85%.
- Shuffle. Randomly reassign which member made which trade, 5,000 times. This destroys any genuine member↔jurisdiction link while keeping the overall mix of trades and committees identical.
- Compare. Recompute the share each time to build the "by chance" distribution (centred at 10.71%). The p-value is the fraction of shuffles that reach the observed figure.
Method note: only trades with a resolved sector, made by members who oversee at least one sector, are included — the rest can't inform the test. BUY and SELL are both counted. p is add-one smoothed, so it is never reported as exactly zero from a finite number of shuffles. The test is in app/analytics/null_model.py.
This measures whether a pattern exists — members trading within their oversight — not whether any individual trade is improper. A trade in your committee's sector can be perfectly legal and routine. The point of accountability is that the pattern is visible and verifiable, trade by trade, against the original filings.