Earth & climate · the experiment
Measure the toll's effect
A landmark toll, a spectacular drop on paper, a margin of error that swallows everything. Four questions, from beginner to expert, on what Manhattan's counters really say.
1 · Read the main result
Beginner
What does the study conclude about traffic inside the toll zone?
2 · Convert the peak-hour effect
The model works on the logarithm of hourly traffic. The author turns a coefficient into a percentage drop with 1 − e^coefficient: −0.539 gives 41.7%. At peak hours, the coefficient is −0.484, and e^−0.484 is about 0.62.
Intermediate
What drop in traffic does this coefficient represent?
3 · Interpret the fake-date test
The author reruns the calculation as if the toll had started a year earlier, on 5 January 2024. The model still finds a drop: −1.03 on the log scale, not significant.
Expert
What does this result suggest?
4 · Weigh the rise on the bridges
On the bridges in the sample, traffic rises by 25% after the toll begins, with a p-value of 0.036.
Expert
Why does this figure not prove, on its own, that drivers are going around the zone?
Toll measured
—